umgeher/leveled
0
Fork 0
Code Issues Pull requests Releases Activity Actions
leveled/test/end_to_end/basic_SUITE.erl

1294 lines
53 KiB
Erlang
Raw Normal View History

Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
-module(basic_SUITE).
Switch to logger (#442) * Switch to logger Use logger rather than io:format when logging. The ct tests have besn switched to log to file, testutil/init_per_suite/1 may offer useful guidance on configuring logger with leveled. As all logs are produced by the leveled_log module, the MFA metadata is uninteresting for log outputs, but can be used for explicit filter controls for leveled logs. * iolist_to_binary not unicode_binary() logger filters will be error and be removed if the format line is a binary(). Must be either a charlist() or a unicode_binary() - so iolist_to_binary() can't be used * Add metadata for filter * Update test/end_to_end/tictac_SUITE.erl Co-authored-by: Thomas Arts <thomas.arts@quviq.com> --------- Co-authored-by: Thomas Arts <thomas.arts@quviq.com>
2024-09-06 11:18:24 +01:00
-include("leveled.hrl").
-export([all/0, init_per_suite/1, end_per_suite/1]).
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
-export([simple_put_fetch_head_delete/1,
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
many_put_fetch_head/1,
journal_compaction/1,
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
fetchput_snapshot/1,
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
load_and_count/1,
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
load_and_count_withdelete/1,
Add test for is_empty Bucket listing didn't care if keys were active - now does.
2017-05-23 11:59:44 +01:00
space_clear_ondelete/1,
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
is_empty_test/1,
Add test of switching from a big journal to a little journal And sending objects over the journal size
2018-09-25 16:56:24 +01:00
many_put_fetch_switchcompression/1,
Add ct test Which passes
2018-09-27 11:32:42 +01:00
bigjournal_littlejournal/1,
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
bigsst_littlesst/1,
Test to expose startup issue See #194
2018-10-03 18:29:20 +01:00
safereaderror_startup/1,
Add performance/profiling test (#424) * Add performance/profiling test Add test to perf_SUITE to do performance tests and also profile different activities in leveled. This can then be used to highlight functions with unexpectedly high execution times, and prove the impact of changes. Switch between riak_ctperf and riak_fullperf to change from standard test (with profile option) to full-scale performance test * Change shape of default perfTest * Change fullPerf Change the fullPerf test to run more tests, but with fewer keys. Given that RS of 512 is being pushed in Riak, 2M objects is till a 300M+ object cluster. 10M >> 1B. so these are still reasonable sizes to test. A profilePerf test also added to generate all the profiles base don 2M objects. * Extend test Queries where previously all returning a large number of index entries - changes made to make number of entries per query more realistic. Also an update process added to show difference between loading and rotating keys. * Relabel as AAE fold * Test v5 Test mini-queries - where generally a small number of entries are returned * Default to ctperf
2023-12-19 11:56:03 +00:00
remove_journal_test/1,
bigpcl_bucketlist/1
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
]).
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
Broken snapshot test The test confirming that deleting sft files wer eheld open whilst snapshots were registered was actually broken. This test has now been fixed, as well as the logic in registring snapshots which had used ledger_sqn mistakenly rather than manifest_sqn.
2016-10-21 11:38:30 +01:00
all() -> [
Uncomment tests!
2019-01-27 23:31:44 +00:00
simple_put_fetch_head_delete,
many_put_fetch_head,
journal_compaction,
fetchput_snapshot,
load_and_count,
load_and_count_withdelete,
space_clear_ondelete,
is_empty_test,
many_put_fetch_switchcompression,
bigjournal_littlejournal,
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
bigsst_littlesst,
Handle missing journal files without data loss Make sure data added after the journal has been lost will not be lost (even though lost data may stay lost)
2018-10-03 18:38:56 +01:00
safereaderror_startup,
Add performance/profiling test (#424) * Add performance/profiling test Add test to perf_SUITE to do performance tests and also profile different activities in leveled. This can then be used to highlight functions with unexpectedly high execution times, and prove the impact of changes. Switch between riak_ctperf and riak_fullperf to change from standard test (with profile option) to full-scale performance test * Change shape of default perfTest * Change fullPerf Change the fullPerf test to run more tests, but with fewer keys. Given that RS of 512 is being pushed in Riak, 2M objects is till a 300M+ object cluster. 10M >> 1B. so these are still reasonable sizes to test. A profilePerf test also added to generate all the profiles base don 2M objects. * Extend test Queries where previously all returning a large number of index entries - changes made to make number of entries per query more realistic. Also an update process added to show difference between loading and rotating keys. * Relabel as AAE fold * Test v5 Test mini-queries - where generally a small number of entries are returned * Default to ctperf
2023-12-19 11:56:03 +00:00
remove_journal_test,
bigpcl_bucketlist
Improve test coverage Some additional tests following previous refactoring for abstraction, primarily to make manifest print safer an dprove co-existence of Riak and non-Riak objects.
2016-10-14 22:58:01 +01:00
].
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
Further work on system tests Another issue exposed with laziness in the using an incomplete ledger when checking for presence during compaction.
2016-10-05 18:28:31 +01:00
Switch to logger (#442) * Switch to logger Use logger rather than io:format when logging. The ct tests have besn switched to log to file, testutil/init_per_suite/1 may offer useful guidance on configuring logger with leveled. As all logs are produced by the leveled_log module, the MFA metadata is uninteresting for log outputs, but can be used for explicit filter controls for leveled logs. * iolist_to_binary not unicode_binary() logger filters will be error and be removed if the format line is a binary(). Must be either a charlist() or a unicode_binary() - so iolist_to_binary() can't be used * Add metadata for filter * Update test/end_to_end/tictac_SUITE.erl Co-authored-by: Thomas Arts <thomas.arts@quviq.com> --------- Co-authored-by: Thomas Arts <thomas.arts@quviq.com>
2024-09-06 11:18:24 +01:00
init_per_suite(Config) ->
testutil:init_per_suite([{suite, "basic"}|Config]),
Config.
end_per_suite(Config) ->
testutil:end_per_suite(Config).
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
simple_put_fetch_head_delete(_Config) ->
Support for log settings at startup Both log level and forced_logs. Allows for log_level to be changed at startup ad runtime. Also allow for a list of forced logs, so if log_level is set > info, individual info logs can be forced to be seen (such as to see stats logs).
2018-12-05 00:17:39 +00:00
io:format("simple test with info and no forced logs~n"),
simple_test_withlog(info, []),
io:format("simple test with error and no forced logs~n"),
simple_test_withlog(error, []),
io:format("simple test with error and stats logs~n"),
Develop 3.1 d30update (#386) * Mas i370 patch d (#383) * Refactor penciller memory In high-volume tests on large key-count clusters, so significant variation in the P0031 time has been seen: TimeBucket PatchA a.0ms_to_1ms 18554 b.1ms_to_2ms 51778 c.2ms_to_3ms 696 d.3ms_to_5ms 220 e.5ms_to_8ms 59 f.8ms_to_13ms 40 g.13ms_to_21ms 364 h.21ms_to_34ms 277 i.34ms_to_55ms 34 j.55ms_to_89ms 17 k.89ms_to_144ms 21 l.144ms_to_233ms 31 m.233ms_to_377ms 45 n.377ms_to_610ms 52 o.610ms_to_987ms 59 p.987ms_to_1597ms 55 q.1597ms_to_2684ms 54 r.2684ms_to_4281ms 29 s.4281ms_to_6965ms 7 t.6295ms_to_11246ms 1 It is unclear why this varies so much. The time to add to the cache appears to be minimal (but perhaps there is an issue with timing points in the code), whereas the time to add to the index is much more significant and variable. There is also variable time when the memory is rolled (although the actual activity here appears to be minimal. The refactoring here is two-fold: - tidy and simplify by keeping LoopState managed within handle_call, and add more helpful dialyzer specs; - change the update to the index to be a simple extension of a list, rather than any conversion. This alternative version of the pmem index in unit test is orders of magnitude faster to add - and is the same order of magnitude to check. Anticipation is that it may be more efficient in terms of memory changes. * Compress SST index Reduces the size of the leveled_sst index with two changes: 1 - Where there is a common prefix of tuple elements (e.g. Bucket) across the whole leveled_sst file - only the non-common part is indexed, and a function is used to compare. 2 - There is less "indexing" of the index i.e. only 1 in 16 keys are passed into the gb_trees part instead of 1 in 4 * Immediate hibernate Reasons for delay in hibernate were not clear. Straight after creation the process will not be in receipt of messages (must wait for the manifest to be updated), so better to hibernate now. This also means the log PC023 provides more accurate information. * Refactor BIC This patch avoids the following: - repeated replacement of the same element in the BIC (via get_kvrange), by checking presence via GET before sing SET - Stops re-reading of all elements to discover high modified date Also there appears to have been a bug where a missing HMD for the file is required to add to the cache. However, now the cache may be erased without erasing the HMD. This means that the cache can never be rebuilt * Use correct size in test results erts_debug:flat_size/1 returns size in words (i.e. 8 bytes on 64-bit CPU) not bytes * Don't change summary record As it is persisted as part of the file write, any change to the summary record cannot be rolled back * Clerk to prompt L0 write Simplifies the logic if the clerk request work for the penciller prompts L0 writes as well as Manifest changes. The advantage now is that if the penciller memory is full, and PUT load stops, the clerk should still be able to prompt persistence. the penciller can therefore make use of dead time this way * Add push on journal compact If there has been a backlog, followed by a quiet period - there may be a large ledger cache left unpushed. Journal compaction events are about once per hour, so the performance overhead of a false push should be minimal, with the advantage of clearing any backlog before load starts again. This is only relevant to riak users with very off/full batch type workloads. * Extend tests To more consistently trigger all overload scenarios * Fix range keys smaller than prefix Can't make end key an empty binary in this case, as it may be bigger than any keys within the range, but will appear to be smaller. Unit tests and ct tests added to expose the potential issue * Tidy-up - Remove penciller logs which are no longer called - Get pclerk to only wait MIN_TIMEOUT after doing work, in case there is a backlog - Remove update_levelzero_cache function as it is unique to handle_call of push_mem, and simple enough to be inline - Alight testutil slow offer with standard slow offer used * Tidy-up Remove pre-otp20 references. Reinstate the check that the starting pid is still active, this was added to tidy up shutdown. Resolve failure to run on otp20 due to `-if` sttaement * Tidy up Using null rather then {null, Key} is potentially clearer as it is not a concern what they Key is in this case, and removes a comparison step from the leveled_codec:endkey_passed/2 function. There were issues with coverage in eunit tests as the leveled_pclerk shut down. This prompted a general tidy of leveled_pclerk (remove passing of LoopState into internal functions, and add dialyzer specs. * Remove R16 relic * Further testing another issue The StartKey must always be less than or equal to the prefix when the first N characters are stripped, but this is not true of the EndKey (for the query) which does not have to be between the FirstKey and the LastKey. If the EndKey query does not match it must be greater than the Prefix (as otherwise it would not have been greater than the FirstKey - so set to null. * Fix unit test Unit test had a typo - and result interpretation had a misunderstanding. * Code and spec tidy Also look to the cover the situation when the FirstKey is the same as the Prefix with tests. This is, in theory, not an issue as it is the EndKey for each sublist which is indexed in leveled_tree. However, guard against it mapping to null here, just in case there are dangers lurking (note that tests will still pass without `M > N` guard in place. * Hibernate on BIC complete There are three situations when the BIC becomes complete: - In a file created as part of a merge the BIS is learned in the merge - After startup, files below L1 learn the block cache through reads that happen to read the block, eventually the while cache will be read, unless... - Either before/after the cache is complete, it can get whiped by a timeout after a get_sqn request (e.g. as prompted by a journal compaction) ... it will then be re-filled of the back of get/get-range requests. In all these situations we want to hibernate after the BIC is fill - to reflect the fact that the LoopState should now be relatively stable, so it is a good point to GC and rationalise location of data. Previously on the the first base was covered. Now all three are covered through the bic_complete message. * Test all index keys have same term This works functionally, but is not optimised (the term is replicated in the index) * Summaries with same index term If the summary index all have the same index term - only the object keys need to be indexes * Simplify case statements We either match the pattern of <<Prefix:N, Suffix>> or the answer should be null * OK for M == N If M = N for the first key, it will have a suffix of <<>>. This will match (as expected) a query Start Key of the sam size, and be smaller than any query Start Key that has the same prefix. If the query Start Key does not match the prefix - it will be null - as it must be smaller than the Prefix (as other wise the query Start Key would be bigger than the Last Key). The constraint of M > N was introduced before the *_prefix_filter functions were checking the prefix, to avoid issues. Now the prefix is being checked, then M == N is ok. * Simplify Correct the test to use a binary field in the range. To avoid further issue, only apply filter when everything is a binary() type. * Add test for head_only mode When leveled is used as a tictacaae key store (in parallel mode), the keys will be head_only entries. Double check they are handled as expected like object keys * Revert previous change - must support typed buckets Add assertion to confirm worthwhile optimisation * Add support for configurable cache multiple (#375) * Mas i370 patch e (#385) Improvement to monitoring for efficiency and improved readability of logs and stats. As part of this, where possible, tried to avoid updating loop state on READ messages in leveled processes (as was the case when tracking stats within each process). No performance benefits found with change, but improved stats has helped discover other potential gains.
2022-12-18 20:18:03 +00:00
simple_test_withlog(
error,
[b0015, b0016, b0017, b0018, p0032, sst12, cdb19, sst13, i0019]).
Support for log settings at startup Both log level and forced_logs. Allows for log_level to be changed at startup ad runtime. Also allow for a list of forced logs, so if log_level is set > info, individual info logs can be forced to be seen (such as to see stats logs).
2018-12-05 00:17:39 +00:00
Allow for run-time changes to log_level and forced_logs Will not lead to immediate run time changes in SST or CDB logs. These log settings will only change once the new files are re-written. To completely change the log level - a restart of the store is necessary with new startup options.
2018-12-11 21:59:57 +00:00
Support for log settings at startup Both log level and forced_logs. Allows for log_level to be changed at startup ad runtime. Also allow for a list of forced logs, so if log_level is set > info, individual info logs can be forced to be seen (such as to see stats logs).
2018-12-05 00:17:39 +00:00
simple_test_withlog(LogLevel, ForcedLogs) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
RootPath = testutil:reset_filestructure(),
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
StartOpts1 = [{root_path, RootPath},
Support for log settings at startup Both log level and forced_logs. Allows for log_level to be changed at startup ad runtime. Also allow for a list of forced logs, so if log_level is set > info, individual info logs can be forced to be seen (such as to see stats logs).
2018-12-05 00:17:39 +00:00
{sync_strategy, testutil:sync_strategy()},
{log_level, LogLevel},
{forced_logs, ForcedLogs}],
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{TestObject, TestSpec} = testutil:generate_testobject(),
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
ok = testutil:book_riakput(Bookie1, TestObject, TestSpec),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
testutil:check_formissingobject(Bookie1, "Bucket1", "Key2"),
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
ok = leveled_bookie:book_close(Bookie1),
Startup work - baffled Changes the stratup otpions to a prolist to make it easier to get environment variables as default. Tried application:start - and completely baffled as to how to get this to work.
2016-11-02 12:58:27 +00:00
StartOpts2 = [{root_path, RootPath},
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
{max_journalsize, 3000000},
Support for log settings at startup Both log level and forced_logs. Allows for log_level to be changed at startup ad runtime. Also allow for a list of forced logs, so if log_level is set > info, individual info logs can be forced to be seen (such as to see stats logs).
2018-12-05 00:17:39 +00:00
{sync_strategy, testutil:sync_strategy()},
{log_level, LogLevel},
{forced_logs, ForcedLogs}],
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
{ok, Bookie2} = leveled_bookie:book_start(StartOpts2),
Allow for run-time changes to log_level and forced_logs Will not lead to immediate run time changes in SST or CDB logs. These log settings will only change once the new files are re-written. To completely change the log level - a restart of the store is necessary with new startup options.
2018-12-11 21:59:57 +00:00
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie2, TestObject),
ObjList1 = testutil:generate_objects(5000, 2),
Shift pause out of leveled Leveled will now signal the need for a pause due to back-pressure, but not actually pause itself. The hope is that in a riak implementation this pause can be managed by the put_fsm, and so not lock the store.
2016-11-07 10:27:38 +00:00
testutil:riakload(Bookie2, ObjList1),
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
ChkList1 = lists:sublist(lists:sort(ObjList1), 100),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, ChkList1),
testutil:check_forobject(Bookie2, TestObject),
testutil:check_formissingobject(Bookie2, "Bucket1", "Key2"),
Improve test coverage Some additional tests following previous refactoring for abstraction, primarily to make manifest print safer an dprove co-existence of Riak and non-Riak objects.
2016-10-14 22:58:01 +01:00
ok = leveled_bookie:book_put(Bookie2, "Bucket1", "Key2", "Value2",
[{add, "Index1", "Term1"}]),
{ok, "Value2"} = leveled_bookie:book_get(Bookie2, "Bucket1", "Key2"),
Use backwards compatible term_to_binary (#408) * Use backwards compatible term_to_binary So that where we have hashed term_to_binary output in OTP25 or earlier, that has will be matched in OTP 26. * Test reliability If all keys are put in order, the max_slots may not be used, as the driver at L0 is penciller cache size, and merge to new files (managed by the parameter) only occurs when there are overlapping files the level below
2023-10-05 10:33:20 +01:00
{ok, {62888926, S, undefined}} =
leveled_bookie:book_head(Bookie2, "Bucket1", "Key2"),
true = (S == 58) or (S == 60),
% After OTP 26 the object is 58 bytes not 60
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_formissingobject(Bookie2, "Bucket1", "Key2"),
Improve test coverage Some additional tests following previous refactoring for abstraction, primarily to make manifest print safer an dprove co-existence of Riak and non-Riak objects.
2016-10-14 22:58:01 +01:00
ok = leveled_bookie:book_put(Bookie2, "Bucket1", "Key2", <<"Value2">>,
[{remove, "Index1", "Term1"},
{add, "Index1", <<"Term2">>}]),
{ok, <<"Value2">>} = leveled_bookie:book_get(Bookie2, "Bucket1", "Key2"),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
ok = leveled_bookie:book_close(Bookie2),
Improve test coverage Some additional tests following previous refactoring for abstraction, primarily to make manifest print safer an dprove co-existence of Riak and non-Riak objects.
2016-10-14 22:58:01 +01:00
{ok, Bookie3} = leveled_bookie:book_start(StartOpts2),
{ok, <<"Value2">>} = leveled_bookie:book_get(Bookie3, "Bucket1", "Key2"),
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
ok = leveled_bookie:book_delete(Bookie3, "Bucket1", "Key2",
[{remove, "Index1", "Term1"}]),
not_found = leveled_bookie:book_get(Bookie3, "Bucket1", "Key2"),
Quality review Minor test fix-up and quality changes
2016-10-30 22:06:44 +00:00
not_found = leveled_bookie:book_head(Bookie3, "Bucket1", "Key2"),
Improve test coverage Some additional tests following previous refactoring for abstraction, primarily to make manifest print safer an dprove co-existence of Riak and non-Riak objects.
2016-10-14 22:58:01 +01:00
ok = leveled_bookie:book_close(Bookie3),
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
{ok, Bookie4} = leveled_bookie:book_start(StartOpts2),
not_found = leveled_bookie:book_get(Bookie4, "Bucket1", "Key2"),
Add destruction Allow a store to be cleared out and destroyed
2016-11-21 12:34:40 +00:00
ok = leveled_bookie:book_destroy(Bookie4).
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
many_put_fetch_head(_Config) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
RootPath = testutil:reset_filestructure(),
Develop 3.1 d30update (#386) * Mas i370 patch d (#383) * Refactor penciller memory In high-volume tests on large key-count clusters, so significant variation in the P0031 time has been seen: TimeBucket PatchA a.0ms_to_1ms 18554 b.1ms_to_2ms 51778 c.2ms_to_3ms 696 d.3ms_to_5ms 220 e.5ms_to_8ms 59 f.8ms_to_13ms 40 g.13ms_to_21ms 364 h.21ms_to_34ms 277 i.34ms_to_55ms 34 j.55ms_to_89ms 17 k.89ms_to_144ms 21 l.144ms_to_233ms 31 m.233ms_to_377ms 45 n.377ms_to_610ms 52 o.610ms_to_987ms 59 p.987ms_to_1597ms 55 q.1597ms_to_2684ms 54 r.2684ms_to_4281ms 29 s.4281ms_to_6965ms 7 t.6295ms_to_11246ms 1 It is unclear why this varies so much. The time to add to the cache appears to be minimal (but perhaps there is an issue with timing points in the code), whereas the time to add to the index is much more significant and variable. There is also variable time when the memory is rolled (although the actual activity here appears to be minimal. The refactoring here is two-fold: - tidy and simplify by keeping LoopState managed within handle_call, and add more helpful dialyzer specs; - change the update to the index to be a simple extension of a list, rather than any conversion. This alternative version of the pmem index in unit test is orders of magnitude faster to add - and is the same order of magnitude to check. Anticipation is that it may be more efficient in terms of memory changes. * Compress SST index Reduces the size of the leveled_sst index with two changes: 1 - Where there is a common prefix of tuple elements (e.g. Bucket) across the whole leveled_sst file - only the non-common part is indexed, and a function is used to compare. 2 - There is less "indexing" of the index i.e. only 1 in 16 keys are passed into the gb_trees part instead of 1 in 4 * Immediate hibernate Reasons for delay in hibernate were not clear. Straight after creation the process will not be in receipt of messages (must wait for the manifest to be updated), so better to hibernate now. This also means the log PC023 provides more accurate information. * Refactor BIC This patch avoids the following: - repeated replacement of the same element in the BIC (via get_kvrange), by checking presence via GET before sing SET - Stops re-reading of all elements to discover high modified date Also there appears to have been a bug where a missing HMD for the file is required to add to the cache. However, now the cache may be erased without erasing the HMD. This means that the cache can never be rebuilt * Use correct size in test results erts_debug:flat_size/1 returns size in words (i.e. 8 bytes on 64-bit CPU) not bytes * Don't change summary record As it is persisted as part of the file write, any change to the summary record cannot be rolled back * Clerk to prompt L0 write Simplifies the logic if the clerk request work for the penciller prompts L0 writes as well as Manifest changes. The advantage now is that if the penciller memory is full, and PUT load stops, the clerk should still be able to prompt persistence. the penciller can therefore make use of dead time this way * Add push on journal compact If there has been a backlog, followed by a quiet period - there may be a large ledger cache left unpushed. Journal compaction events are about once per hour, so the performance overhead of a false push should be minimal, with the advantage of clearing any backlog before load starts again. This is only relevant to riak users with very off/full batch type workloads. * Extend tests To more consistently trigger all overload scenarios * Fix range keys smaller than prefix Can't make end key an empty binary in this case, as it may be bigger than any keys within the range, but will appear to be smaller. Unit tests and ct tests added to expose the potential issue * Tidy-up - Remove penciller logs which are no longer called - Get pclerk to only wait MIN_TIMEOUT after doing work, in case there is a backlog - Remove update_levelzero_cache function as it is unique to handle_call of push_mem, and simple enough to be inline - Alight testutil slow offer with standard slow offer used * Tidy-up Remove pre-otp20 references. Reinstate the check that the starting pid is still active, this was added to tidy up shutdown. Resolve failure to run on otp20 due to `-if` sttaement * Tidy up Using null rather then {null, Key} is potentially clearer as it is not a concern what they Key is in this case, and removes a comparison step from the leveled_codec:endkey_passed/2 function. There were issues with coverage in eunit tests as the leveled_pclerk shut down. This prompted a general tidy of leveled_pclerk (remove passing of LoopState into internal functions, and add dialyzer specs. * Remove R16 relic * Further testing another issue The StartKey must always be less than or equal to the prefix when the first N characters are stripped, but this is not true of the EndKey (for the query) which does not have to be between the FirstKey and the LastKey. If the EndKey query does not match it must be greater than the Prefix (as otherwise it would not have been greater than the FirstKey - so set to null. * Fix unit test Unit test had a typo - and result interpretation had a misunderstanding. * Code and spec tidy Also look to the cover the situation when the FirstKey is the same as the Prefix with tests. This is, in theory, not an issue as it is the EndKey for each sublist which is indexed in leveled_tree. However, guard against it mapping to null here, just in case there are dangers lurking (note that tests will still pass without `M > N` guard in place. * Hibernate on BIC complete There are three situations when the BIC becomes complete: - In a file created as part of a merge the BIS is learned in the merge - After startup, files below L1 learn the block cache through reads that happen to read the block, eventually the while cache will be read, unless... - Either before/after the cache is complete, it can get whiped by a timeout after a get_sqn request (e.g. as prompted by a journal compaction) ... it will then be re-filled of the back of get/get-range requests. In all these situations we want to hibernate after the BIC is fill - to reflect the fact that the LoopState should now be relatively stable, so it is a good point to GC and rationalise location of data. Previously on the the first base was covered. Now all three are covered through the bic_complete message. * Test all index keys have same term This works functionally, but is not optimised (the term is replicated in the index) * Summaries with same index term If the summary index all have the same index term - only the object keys need to be indexes * Simplify case statements We either match the pattern of <<Prefix:N, Suffix>> or the answer should be null * OK for M == N If M = N for the first key, it will have a suffix of <<>>. This will match (as expected) a query Start Key of the sam size, and be smaller than any query Start Key that has the same prefix. If the query Start Key does not match the prefix - it will be null - as it must be smaller than the Prefix (as other wise the query Start Key would be bigger than the Last Key). The constraint of M > N was introduced before the *_prefix_filter functions were checking the prefix, to avoid issues. Now the prefix is being checked, then M == N is ok. * Simplify Correct the test to use a binary field in the range. To avoid further issue, only apply filter when everything is a binary() type. * Add test for head_only mode When leveled is used as a tictacaae key store (in parallel mode), the keys will be head_only entries. Double check they are handled as expected like object keys * Revert previous change - must support typed buckets Add assertion to confirm worthwhile optimisation * Add support for configurable cache multiple (#375) * Mas i370 patch e (#385) Improvement to monitoring for efficiency and improved readability of logs and stats. As part of this, where possible, tried to avoid updating loop state on READ messages in leveled processes (as was the case when tracking stats within each process). No performance benefits found with change, but improved stats has helped discover other potential gains.
2022-12-18 20:18:03 +00:00
StartOpts1 =
[{root_path, RootPath},
{max_pencillercachesize, 16000},
{sync_strategy, riak_sync},
{compression_point, on_compact}
],
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{TestObject, TestSpec} = testutil:generate_testobject(),
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
ok = testutil:book_riakput(Bookie1, TestObject, TestSpec),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Add access to SQN Use book_sqn/3 or book_sqn/4 to get the SQN of an object in the store.
2019-03-13 16:21:03 +00:00
{ok, 1} = leveled_bookie:book_sqn(Bookie1,
testutil:get_bucket(TestObject),
testutil:get_key(TestObject),
?RIAK_TAG),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
ok = leveled_bookie:book_close(Bookie1),
Startup work - baffled Changes the stratup otpions to a prolist to make it easier to get environment variables as default. Tried application:start - and completely baffled as to how to get this to work.
2016-11-02 12:58:27 +00:00
StartOpts2 = [{root_path, RootPath},
Test logging of samples To prompt the log the journal size needs to be reduced
2017-11-20 15:31:31 +00:00
{max_journalsize, 50000000},
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
{max_pencillercachesize, 32000},
Improve test coverage Make compress on receipt/compaction configurable
2017-11-06 18:44:08 +00:00
{sync_strategy, testutil:sync_strategy()},
{compression_point, on_receipt}],
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
{ok, Bookie2} = leveled_bookie:book_start(StartOpts2),
Allow for run-time changes to log_level and forced_logs Will not lead to immediate run time changes in SST or CDB logs. These log settings will only change once the new files are re-written. To completely change the log level - a restart of the store is necessary with new startup options.
2018-12-11 21:59:57 +00:00
ok = leveled_bookie:book_loglevel(Bookie2, error),
ok = leveled_bookie:book_addlogs(Bookie2, ["B0015"]),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie2, TestObject),
Add access to SQN Use book_sqn/3 or book_sqn/4 to get the SQN of an object in the store.
2019-03-13 16:21:03 +00:00
{ok, 1} = leveled_bookie:book_sqn(Bookie2,
testutil:get_bucket(TestObject),
testutil:get_key(TestObject),
?RIAK_TAG),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
GenList = [2, 20002, 40002, 60002, 80002,
100002, 120002, 140002, 160002, 180002],
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
CLs = testutil:load_objects(20000, GenList, Bookie2, TestObject,
fun testutil:generate_smallobjects/2),
Allow for run-time changes to log_level and forced_logs Will not lead to immediate run time changes in SST or CDB logs. These log settings will only change once the new files are re-written. To completely change the log level - a restart of the store is necessary with new startup options.
2018-12-11 21:59:57 +00:00
{error, ["B0015"]} = leveled_bookie:book_logsettings(Bookie2),
ok = leveled_bookie:book_removelogs(Bookie2, ["B0015"]),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
CL1A = lists:nth(1, CLs),
ChkListFixed = lists:nth(length(CLs), CLs),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, CL1A),
Allow for run-time changes to log_level and forced_logs Will not lead to immediate run time changes in SST or CDB logs. These log settings will only change once the new files are re-written. To completely change the log level - a restart of the store is necessary with new startup options.
2018-12-11 21:59:57 +00:00
{error, []} = leveled_bookie:book_logsettings(Bookie2),
ok = leveled_bookie:book_loglevel(Bookie2, info),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
ObjList2A = testutil:generate_objects(5000, 2),
Shift pause out of leveled Leveled will now signal the need for a pause due to back-pressure, but not actually pause itself. The hope is that in a riak implementation this pause can be managed by the put_fsm, and so not lock the store.
2016-11-07 10:27:38 +00:00
testutil:riakload(Bookie2, ObjList2A),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
ChkList2A = lists:sublist(lists:sort(ObjList2A), 1000),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, ChkList2A),
testutil:check_forlist(Bookie2, ChkListFixed),
testutil:check_forobject(Bookie2, TestObject),
testutil:check_forlist(Bookie2, ChkList2A),
testutil:check_forlist(Bookie2, ChkListFixed),
testutil:check_forobject(Bookie2, TestObject),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
ok = leveled_bookie:book_close(Bookie2),
{ok, Bookie3} = leveled_bookie:book_start(StartOpts2),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie3, ChkList2A),
testutil:check_forobject(Bookie3, TestObject),
Add access to SQN Use book_sqn/3 or book_sqn/4 to get the SQN of an object in the store.
2019-03-13 16:21:03 +00:00
{ok, 1} = leveled_bookie:book_sqn(Bookie3,
testutil:get_bucket(TestObject),
testutil:get_key(TestObject),
?RIAK_TAG),
not_found = leveled_bookie:book_sqn(Bookie3,
testutil:get_bucket(TestObject),
testutil:get_key(TestObject),
?STD_TAG),
Test default tag
2019-03-14 00:08:01 +00:00
not_found = leveled_bookie:book_sqn(Bookie3,
testutil:get_bucket(TestObject),
testutil:get_key(TestObject)),
Test coverage - missing 1%
2017-02-26 20:52:40 +00:00
testutil:check_formissingobject(Bookie3, "Bookie1", "MissingKey0123"),
Add destruction Allow a store to be cleared out and destroyed
2016-11-21 12:34:40 +00:00
ok = leveled_bookie:book_destroy(Bookie3).
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
Add test of switching from a big journal to a little journal And sending objects over the journal size
2018-09-25 16:56:24 +01:00
bigjournal_littlejournal(_Config) ->
RootPath = testutil:reset_filestructure(),
StartOpts1 = [{root_path, RootPath},
{max_journalsize, 50000000},
{max_pencillercachesize, 32000},
{sync_strategy, testutil:sync_strategy()},
{compression_point, on_compact}],
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
ObjL1 =
testutil:generate_objects(100, 1, [],
leveled_rand:rand_bytes(10000),
fun() -> [] end, <<"B">>),
testutil:riakload(Bookie1, ObjL1),
ok = leveled_bookie:book_close(Bookie1),
StartOpts2 = lists:ukeysort(1, [{max_journalsize, 5000}|StartOpts1]),
{ok, Bookie2} = leveled_bookie:book_start(StartOpts2),
ObjL2 =
testutil:generate_objects(10, 1000, [],
leveled_rand:rand_bytes(10000),
fun() -> [] end, <<"B">>),
testutil:riakload(Bookie2, ObjL2),
testutil:check_forlist(Bookie2, ObjL1),
testutil:check_forlist(Bookie2, ObjL2),
ok = leveled_bookie:book_destroy(Bookie2).
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
bigsst_littlesst(_Config) ->
RootPath = testutil:reset_filestructure(),
StartOpts1 = [{root_path, RootPath},
{max_journalsize, 50000000},
Change cache_size in sst tests Makes results more predictable (with coin toss variations)
2019-01-29 13:40:55 +00:00
{cache_size, 500},
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
{max_pencillercachesize, 16000},
{max_sstslots, 256},
{sync_strategy, testutil:sync_strategy()},
{compression_point, on_compact}],
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
ObjL1 =
Use backwards compatible term_to_binary (#408) * Use backwards compatible term_to_binary So that where we have hashed term_to_binary output in OTP25 or earlier, that has will be matched in OTP 26. * Test reliability If all keys are put in order, the max_slots may not be used, as the driver at L0 is penciller cache size, and merge to new files (managed by the parameter) only occurs when there are overlapping files the level below
2023-10-05 10:33:20 +01:00
lists:keysort(
1,
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
testutil:generate_objects(
100000,
1,
[],
leveled_rand:rand_bytes(100),
fun() -> [] end,
<<"B">>)
Use backwards compatible term_to_binary (#408) * Use backwards compatible term_to_binary So that where we have hashed term_to_binary output in OTP25 or earlier, that has will be matched in OTP 26. * Test reliability If all keys are put in order, the max_slots may not be used, as the driver at L0 is penciller cache size, and merge to new files (managed by the parameter) only occurs when there are overlapping files the level below
2023-10-05 10:33:20 +01:00
),
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
testutil:riakload(Bookie1, ObjL1),
testutil:check_forlist(Bookie1, ObjL1),
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
timer:sleep(10000), % Wait for delete timeout
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
JFP = RootPath ++ "/ledger/ledger_files/",
{ok, FNS1} = file:list_dir(JFP),
ok = leveled_bookie:book_destroy(Bookie1),
StartOpts2 = lists:ukeysort(1, [{max_sstslots, 24}|StartOpts1]),
{ok, Bookie2} = leveled_bookie:book_start(StartOpts2),
testutil:riakload(Bookie2, ObjL1),
testutil:check_forlist(Bookie2, ObjL1),
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
timer:sleep(10000), % Wait for delete timeout
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
{ok, FNS2} = file:list_dir(JFP),
ok = leveled_bookie:book_destroy(Bookie2),
io:format("Big SST ~w files Little SST ~w files~n",
[length(FNS1), length(FNS2)]),
Change cache_size in sst tests Makes results more predictable (with coin toss variations)
2019-01-29 13:40:55 +00:00
true = length(FNS2) >= (2 * length(FNS1)).
Test sst slot configuration change Confirm it results in many more files, if the slot count reduced. Has to handle the fact that Level 0 file has unlimited slots regardless of number of slots configured
2019-01-27 22:03:55 +00:00
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
journal_compaction(_Config) ->
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
journal_compaction_tester(false, 3600),
journal_compaction_tester(false, undefined),
journal_compaction_tester(true, 3600).
journal_compaction_tester(Restart, WRP) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
RootPath = testutil:reset_filestructure(),
Startup work - baffled Changes the stratup otpions to a prolist to make it easier to get environment variables as default. Tried application:start - and completely baffled as to how to get this to work.
2016-11-02 12:58:27 +00:00
StartOpts1 = [{root_path, RootPath},
{max_journalsize, 10000000},
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
{max_run_length, 1},
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
{sync_strategy, testutil:sync_strategy()},
{waste_retention_period, WRP}],
{ok, Bookie0} = leveled_bookie:book_start(StartOpts1),
ok = leveled_bookie:book_compactjournal(Bookie0, 30000),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{TestObject, TestSpec} = testutil:generate_testobject(),
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
ok = testutil:book_riakput(Bookie0, TestObject, TestSpec),
testutil:check_forobject(Bookie0, TestObject),
Journal compaction test to cover deleted objects
2016-11-03 12:31:00 +00:00
ObjList1 = testutil:generate_objects(20000, 2),
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
testutil:riakload(Bookie0, ObjList1),
Make tets less pointless Journla compaction test wouldn't actually cause compaction
2016-10-30 20:14:11 +00:00
ChkList1 = lists:sublist(lists:sort(ObjList1), 10000),
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
testutil:check_forlist(Bookie0, ChkList1),
testutil:check_forobject(Bookie0, TestObject),
Test timeouts So that coverage testing will run.
2017-07-02 22:23:02 +01:00
{B2, K2, V2, Spec2, MD} = {"Bucket2",
"Key2",
"Value2",
Further work on system tests Another issue exposed with laziness in the using an incomplete ledger when checking for presence during compaction.
2016-10-05 18:28:31 +01:00
[],
Test timeouts So that coverage testing will run.
2017-07-02 22:23:02 +01:00
[{"MDK2", "MDV2"}]},
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{TestObject2, TestSpec2} = testutil:generate_testobject(B2, K2,
V2, Spec2, MD),
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
ok = testutil:book_riakput(Bookie0, TestObject2, TestSpec2),
ok = leveled_bookie:book_compactjournal(Bookie0, 30000),
testutil:check_forlist(Bookie0, ChkList1),
testutil:check_forobject(Bookie0, TestObject),
testutil:check_forobject(Bookie0, TestObject2),
testutil:check_forlist(Bookie0, ChkList1),
testutil:check_forobject(Bookie0, TestObject),
testutil:check_forobject(Bookie0, TestObject2),
Journal compaction test to cover deleted objects
2016-11-03 12:31:00 +00:00
%% Delete some of the objects
ObjListD = testutil:generate_objects(10000, 2),
lists:foreach(fun({_R, O, _S}) ->
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
testutil:book_riakdelete(Bookie0,
Push log update through to cdb/sst Using the cdb_options and sst_options records
2018-12-11 20:42:00 +00:00
testutil:get_bucket(O),
testutil:get_key(O),
Shift pause out of leveled Leveled will now signal the need for a pause due to back-pressure, but not actually pause itself. The hope is that in a riak implementation this pause can be managed by the put_fsm, and so not lock the store.
2016-11-07 10:27:38 +00:00
[])
Journal compaction test to cover deleted objects
2016-11-03 12:31:00 +00:00
end,
ObjListD),
%% Now replace all the other objects
ObjList2 = testutil:generate_objects(40000, 10002),
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
testutil:riakload(Bookie0, ObjList2),
Bookie1 =
case Restart of
true ->
ok = leveled_bookie:book_close(Bookie0),
{ok, RestartedB} = leveled_bookie:book_start(StartOpts1),
RestartedB;
false ->
Bookie0
end,
Make tets less pointless Journla compaction test wouldn't actually cause compaction
2016-10-30 20:14:11 +00:00
Inker Close nastiness Try to stop some of the potential deadlocking around Inker close and prove that snapshots at higher Manifest SQNs can be ignored
2016-11-14 19:34:11 +00:00
Add coverage for CDB delete_pending with snapshot CDB file in delete_pending state, and times out but the Inker has a snapshot open - and so the file is not deleted.
2017-11-09 12:02:03 +00:00
WasteFP = RootPath ++ "/journal/journal_files/waste",
% Start snapshot - should stop deletions
Split code for folders - introduce runner actor Introduce a dedicated module for all the different fold types. Also simplify the list of folders by deprecating those folds that should eb achieveable by fold_heads/fold_objects type folds but with smarter functions. Makes sure that the fold functiosn also have better spec coverage, and are dialyzer checked.
2017-10-17 20:39:11 +01:00
{ok, PclClone, InkClone} =
leveled_bookie:book_snapshot(Bookie1, store, undefined, false),
Add coverage for CDB delete_pending with snapshot CDB file in delete_pending state, and times out but the Inker has a snapshot open - and so the file is not deleted.
2017-11-09 12:02:03 +00:00
ok = leveled_bookie:book_compactjournal(Bookie1, 30000),
testutil:wait_for_compaction(Bookie1),
% Wait to cause delete_pending to be blocked by snapshot
% timeout on switch to delete_pending is 10s
timer:sleep(10100),
case WRP of
undefined ->
ok;
_ ->
% Check nothing got deleted
{ok, CJs} = file:list_dir(WasteFP),
true = length(CJs) == 0
end,
ok = leveled_penciller:pcl_close(PclClone),
ok = leveled_inker:ink_close(InkClone),
Allow for caching of compaction scores Potentially reduce the overheads of scoring each file on every run. The change also alters the default thresholds for compaction to favour longer runs (which will tend towards greater storage efficiency).
2020-11-27 02:35:27 +00:00
% Snapshot released so deletes should occur at next timeout
Add coverage for CDB delete_pending with snapshot CDB file in delete_pending state, and times out but the Inker has a snapshot open - and so the file is not deleted.
2017-11-09 12:02:03 +00:00
case WRP of
undefined ->
timer:sleep(10100); % wait for delete_pending timeout
Inker Close nastiness Try to stop some of the potential deadlocking around Inker close and prove that snapshots at higher Manifest SQNs can be ignored
2016-11-14 19:34:11 +00:00
% Wait 2 seconds for files to be deleted
Add coverage for CDB delete_pending with snapshot CDB file in delete_pending state, and times out but the Inker has a snapshot open - and so the file is not deleted.
2017-11-09 12:02:03 +00:00
_ ->
FindDeletedFilesFun =
fun(X, Found) ->
case Found of
true ->
Found;
false ->
{ok, Files} = file:list_dir(WasteFP),
if
length(Files) > 0 ->
io:format("Deleted files found~n"),
true;
length(Files) == 0 ->
timer:sleep(X),
false
end
end
end,
lists:foldl(FindDeletedFilesFun,
false,
[2000,2000,2000,2000,2000,2000])
end,
Inker Close nastiness Try to stop some of the potential deadlocking around Inker close and prove that snapshots at higher Manifest SQNs can be ignored
2016-11-14 19:34:11 +00:00
{ok, ClearedJournals} = file:list_dir(WasteFP),
io:format("~w ClearedJournals found~n", [length(ClearedJournals)]),
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
case is_integer(WRP) of
true ->
true = length(ClearedJournals) > 0;
false ->
true = length(ClearedJournals) == 0
end,
Make tets less pointless Journla compaction test wouldn't actually cause compaction
2016-10-30 20:14:11 +00:00
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
ChkList3 = lists:sublist(lists:sort(ObjList2), 500),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie1, ChkList3),
Inker Close nastiness Try to stop some of the potential deadlocking around Inker close and prove that snapshots at higher Manifest SQNs can be ignored
2016-11-14 19:34:11 +00:00
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
ok = leveled_bookie:book_close(Bookie1),
Add coverage for CDB delete_pending with snapshot CDB file in delete_pending state, and times out but the Inker has a snapshot open - and so the file is not deleted.
2017-11-09 12:02:03 +00:00
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
% Restart
{ok, Bookie2} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie2, TestObject),
testutil:check_forlist(Bookie2, ChkList3),
Deferred Deletion of Journals This allows for deleted journals to be retained for a period (the waste_retnetion_period). The idea being that a backup strategy can ensure that all journals are backed up, even ones created and removed from within a backup period - so that any restore pont is possible. This is also a pre-cursor to removing some of the PromptDelete complexity from the Inker Clerk - all compactions can prompt deletion as deletion is now deferred.
2016-11-14 11:17:14 +00:00
Inker Close nastiness Try to stop some of the potential deadlocking around Inker close and prove that snapshots at higher Manifest SQNs can be ignored
2016-11-14 19:34:11 +00:00
ok = leveled_bookie:book_close(Bookie2),
Deferred Deletion of Journals This allows for deleted journals to be retained for a period (the waste_retnetion_period). The idea being that a backup strategy can ensure that all journals are backed up, even ones created and removed from within a backup period - so that any restore pont is possible. This is also a pre-cursor to removing some of the PromptDelete complexity from the Inker Clerk - all compactions can prompt deletion as deletion is now deferred.
2016-11-14 11:17:14 +00:00
StartOpts2 = [{root_path, RootPath},
{max_journalsize, 10000000},
{max_run_length, 1},
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
{waste_retention_period, 1},
{sync_strategy, testutil:sync_strategy()}],
Deferred Deletion of Journals This allows for deleted journals to be retained for a period (the waste_retnetion_period). The idea being that a backup strategy can ensure that all journals are backed up, even ones created and removed from within a backup period - so that any restore pont is possible. This is also a pre-cursor to removing some of the PromptDelete complexity from the Inker Clerk - all compactions can prompt deletion as deletion is now deferred.
2016-11-14 11:17:14 +00:00
{ok, Bookie3} = leveled_bookie:book_start(StartOpts2),
ok = leveled_bookie:book_compactjournal(Bookie3, 30000),
Make check that compaction not ongoing before accepting new compaction Respond 'busy' if compaction is ongoing
2019-01-24 15:46:17 +00:00
busy = leveled_bookie:book_compactjournal(Bookie3, 30000),
Deferred Deletion of Journals This allows for deleted journals to be retained for a period (the waste_retnetion_period). The idea being that a backup strategy can ensure that all journals are backed up, even ones created and removed from within a backup period - so that any restore pont is possible. This is also a pre-cursor to removing some of the PromptDelete complexity from the Inker Clerk - all compactions can prompt deletion as deletion is now deferred.
2016-11-14 11:17:14 +00:00
testutil:wait_for_compaction(Bookie3),
ok = leveled_bookie:book_close(Bookie3),
{ok, ClearedJournalsPC} = file:list_dir(WasteFP),
io:format("~w ClearedJournals found~n", [length(ClearedJournalsPC)]),
Extend journal compaction test to cover with and without waste retention. Also makes sure that CDB files in a restarted store will respect the wast retention period set.
2017-11-08 16:18:48 +00:00
case is_integer(WRP) of
true ->
true = length(ClearedJournals) > 0;
false ->
true = length(ClearedJournals) == 0
end,
Deferred Deletion of Journals This allows for deleted journals to be retained for a period (the waste_retnetion_period). The idea being that a backup strategy can ensure that all journals are backed up, even ones created and removed from within a backup period - so that any restore pont is possible. This is also a pre-cursor to removing some of the PromptDelete complexity from the Inker Clerk - all compactions can prompt deletion as deletion is now deferred.
2016-11-14 11:17:14 +00:00
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
testutil:reset_filestructure(10000).
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
fetchput_snapshot(_Config) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
RootPath = testutil:reset_filestructure(),
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
StartOpts1 = [{root_path, RootPath},
{max_journalsize, 30000000},
Mas i449 directpromptofdeletions (#450) * Move prompt of deletions to Inker It is a series of casts, so no reason to offload this to the clerk. Simplifies potential races in shutdown * Rename * Change cache sizes In the hope of making test more consistent
2024-09-04 09:04:24 +01:00
{cache_size, 2000},
{max_pencillercachesize, 16000},
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
{sync_strategy, none}],
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{TestObject, TestSpec} = testutil:generate_testobject(),
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
ok = testutil:book_riakput(Bookie1, TestObject, TestSpec),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Load up 5000 objects
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
ObjList1 = testutil:generate_objects(5000, 2),
Shift pause out of leveled Leveled will now signal the need for a pause due to back-pressure, but not actually pause itself. The hope is that in a riak implementation this pause can be managed by the put_fsm, and so not lock the store.
2016-11-07 10:27:38 +00:00
testutil:riakload(Bookie1, ObjList1),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Now take a snapshot - check it has the same objects
Startup work - baffled Changes the stratup otpions to a prolist to make it easier to get environment variables as default. Tried application:start - and completely baffled as to how to get this to work.
2016-11-02 12:58:27 +00:00
SnapOpts1 = [{snapshot_bookie, Bookie1}],
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
{ok, SnapBookie1} = leveled_bookie:book_start(SnapOpts1),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
ChkList1 = lists:sublist(lists:sort(ObjList1), 100),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie1, ChkList1),
testutil:check_forlist(SnapBookie1, ChkList1),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
Mas d31 i413 (#415) * Allow snapshots to be reused in queries Allow for a full bookie snapshot to be re-used for multiple queries, not just KV fetches. * Reduce log noise The internal dummy tag is expected so should not prompt a log on reload * Snapshot should have same status of active db wrt head_only and head_lookup * Allow logging to specified on snapshots * Shutdown snapshot bookie is primary goes down Inker and Penciller already will shut down based on `erlang:monitor/2` * Review feedback Formatting and code readability fixes
2023-11-08 09:18:01 +00:00
compare_foldwithsnap(Bookie1, SnapBookie1, ChkList1),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Close the snapshot, check the original store still has the objects
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
ok = leveled_bookie:book_close(SnapBookie1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie1, ChkList1),
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
ok = leveled_bookie:book_close(Bookie1),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Closed initial bookies~n"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% all now closed
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Open a new store (to start with the previously loaded data)
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
{ok, Bookie2} = leveled_bookie:book_start(StartOpts1),
Startup work - baffled Changes the stratup otpions to a prolist to make it easier to get environment variables as default. Tried application:start - and completely baffled as to how to get this to work.
2016-11-02 12:58:27 +00:00
SnapOpts2 = [{snapshot_bookie, Bookie2}],
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% And take a snapshot of that store
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
{ok, SnapBookie2} = leveled_bookie:book_start(SnapOpts2),
io:format("Bookies restarted~n"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Check both the newly opened store and its snapshot have the data
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, ChkList1),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Check active bookie still contains original data~n"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(SnapBookie2, ChkList1),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Check snapshot still contains original data~n"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Generate some replacement objects, load them up - check the master
% store has the replacement objects, but the snapshot still has the old
% objects
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
ObjList2 = testutil:generate_objects(5000, 2),
Shift pause out of leveled Leveled will now signal the need for a pause due to back-pressure, but not actually pause itself. The hope is that in a riak implementation this pause can be managed by the put_fsm, and so not lock the store.
2016-11-07 10:27:38 +00:00
testutil:riakload(Bookie2, ObjList2),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Replacement objects put~n"),
ChkList2 = lists:sublist(lists:sort(ObjList2), 100),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, ChkList2),
testutil:check_forlist(SnapBookie2, ChkList1),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Checked for replacement objects in active bookie" ++
", old objects in snapshot~n"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Check out how many ledger files we now have (should just be 1)
Revert ominshambles performance refactoring To try and improve performance index entries had been removed from the Ledger Cache, and a shadow list of the LedgerCache (in SQN order) was kept to avoid gb_trees:to_list on push_mem. This did not go well. The issue was that ets does not deal with duplicate keys in the list when inserting (it will only insert one, but it is not clear which one). This has been reverted back out. The ETS parameters have been changed to [set, private]. It is not used as an iterator, and is no longer passed out of the process (the memtable_copy is sent instead). This also avoids the tab2list function being called.
2016-10-19 00:10:48 +01:00
ok = filelib:ensure_dir(RootPath ++ "/ledger/ledger_files"),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
{ok, FNsA} = file:list_dir(RootPath ++ "/ledger/ledger_files"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% generate some new objects and load them up. Check that the master store
% has the new objects, and the snapshot doesn't
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
ObjList3 = testutil:generate_objects(15000, 5002),
Shift pause out of leveled Leveled will now signal the need for a pause due to back-pressure, but not actually pause itself. The hope is that in a riak implementation this pause can be managed by the put_fsm, and so not lock the store.
2016-11-07 10:27:38 +00:00
testutil:riakload(Bookie2, ObjList3),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
ChkList3 = lists:sublist(lists:sort(ObjList3), 100),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, ChkList3),
testutil:check_formissinglist(SnapBookie2, ChkList3),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Now loads lots of new objects
Close in stages - waiting for releases (#411) * Close in stages - waiting for releases Have a consistent approach to closing the inker and the penciller - so that the close can be interrupted by releasing of snapshots. Then any unreleased snapshots are closed before shutdown - with a 10s pause to give queries a short opportunity to finish. This should address some issues, primarily seen (but very rarely) in test whereby post-rebuild destruction of parallel AAE keystores cause the crashing of aae_folds. The primary benefit is to stop an attempt to release a snapshot that has in fact already finished does not cause a crash of the database on normal stop. this was primarily an issue when shutdown is delayed by an ongoing journal compaction job. * Boost default test budget for EQC * Update test to use correct type * Update following review Avoid filtering out exited PIDs when closing snapshots by catching the exit exception when the Pid is down
2023-10-03 18:30:40 +01:00
GenList = [20002, 40002, 60002, 80002, 100002, 120002, 140002, 160002],
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
CLs2 = testutil:load_objects(20000, GenList, Bookie2, TestObject,
fun testutil:generate_smallobjects/2),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Loaded significant numbers of new objects~n"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, lists:nth(length(CLs2), CLs2)),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Checked active bookie has new objects~n"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Start a second snapshot, which should have the new objects, whilst the
% previous snapshot still doesn't
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
{ok, SnapBookie3} = leveled_bookie:book_start(SnapOpts2),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(SnapBookie3, lists:nth(length(CLs2), CLs2)),
testutil:check_formissinglist(SnapBookie2, ChkList3),
testutil:check_formissinglist(SnapBookie2, lists:nth(length(CLs2), CLs2)),
Revert ominshambles performance refactoring To try and improve performance index entries had been removed from the Ledger Cache, and a shadow list of the LedgerCache (in SQN order) was kept to avoid gb_trees:to_list on push_mem. This did not go well. The issue was that ets does not deal with duplicate keys in the list when inserting (it will only insert one, but it is not clear which one). This has been reverted back out. The ETS parameters have been changed to [set, private]. It is not used as an iterator, and is no longer passed out of the process (the memtable_copy is sent instead). This also avoids the tab2list function being called.
2016-10-19 00:10:48 +01:00
testutil:check_forlist(Bookie2, ChkList2),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(SnapBookie3, ChkList2),
testutil:check_forlist(SnapBookie2, ChkList1),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
io:format("Started new snapshot and check for new objects~n"),
Mas d31 i413 (#415) * Allow snapshots to be reused in queries Allow for a full bookie snapshot to be re-used for multiple queries, not just KV fetches. * Reduce log noise The internal dummy tag is expected so should not prompt a log on reload * Snapshot should have same status of active db wrt head_only and head_lookup * Allow logging to specified on snapshots * Shutdown snapshot bookie is primary goes down Inker and Penciller already will shut down based on `erlang:monitor/2` * Review feedback Formatting and code readability fixes
2023-11-08 09:18:01 +00:00
compare_foldwithsnap(Bookie2, SnapBookie3, ChkList3),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Load yet more objects, these are replacement objects for the last load
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
CLs3 = testutil:load_objects(20000, GenList, Bookie2, TestObject,
fun testutil:generate_smallobjects/2),
testutil:check_forlist(Bookie2, lists:nth(length(CLs3), CLs3)),
testutil:check_forlist(Bookie2, lists:nth(1, CLs3)),
Broken snapshot test The test confirming that deleting sft files wer eheld open whilst snapshots were registered was actually broken. This test has now been fixed, as well as the logic in registring snapshots which had used ledger_sqn mistakenly rather than manifest_sqn.
2016-10-21 11:38:30 +01:00
io:format("Starting 15s sleep in which snap2 should block deletion~n"),
timer:sleep(15000),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% There should be lots of ledger files, as we have replaced the objects
% which has created new files, but the old files are still in demand from
% the snapshot
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
{ok, FNsB} = file:list_dir(RootPath ++ "/ledger/ledger_files"),
ok = leveled_bookie:book_close(SnapBookie2),
Broken snapshot test The test confirming that deleting sft files wer eheld open whilst snapshots were registered was actually broken. This test has now been fixed, as well as the logic in registring snapshots which had used ledger_sqn mistakenly rather than manifest_sqn.
2016-10-21 11:38:30 +01:00
io:format("Starting 15s sleep as snap2 close should unblock deletion~n"),
timer:sleep(15000),
io:format("Pause for deletion has ended~n"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% So the pause here is to allow for delete pendings to take effect after the
% closing of the snapshot
% Now check that any deletions haven't impacted the availability of data
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, lists:nth(length(CLs3), CLs3)),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
% Close the other snapshot, and pause - after the pause there should be a
% reduction in the number of ledger files due to the deletes
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
ok = leveled_bookie:book_close(SnapBookie3),
Broken snapshot test The test confirming that deleting sft files wer eheld open whilst snapshots were registered was actually broken. This test has now been fixed, as well as the logic in registring snapshots which had used ledger_sqn mistakenly rather than manifest_sqn.
2016-10-21 11:38:30 +01:00
io:format("Starting 15s sleep as snap3 close should unblock deletion~n"),
timer:sleep(15000),
io:format("Pause for deletion has ended~n"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forlist(Bookie2, lists:nth(length(CLs3), CLs3)),
testutil:check_forlist(Bookie2, lists:nth(1, CLs3)),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
{ok, FNsC} = file:list_dir(RootPath ++ "/ledger/ledger_files"),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
io:format("FNsA ~w FNsB ~w FNsC ~w~n",
[length(FNsA), length(FNsB), length(FNsC)]),
Add more complex snapshot test This exposed another off-by-one error on startup. This commit also includes an unsafe change to reply early from a rolling CDB file (with lots of objects writing the hash table can take too long). This is bad, but will be resolved through a refactor of the manifest writing: essentially we deferred writing of the manifest update which was an unnecessary performance optimisation. If instead we wait on this, the process is made substantially simpler, and it is safer to perform the roll of the complete CDB journal asynchronously. If the manifest update takes too long, an append-only log may be used instead.
2016-10-07 10:04:48 +01:00
true = length(FNsB) > length(FNsA),
Add comments to test Also loads some more objects to potentially make results clearer
2018-05-17 18:20:16 +01:00
true = length(FNsB) > length(FNsC),
% smaller due to replacements and files deleting
% This is dependent on the sleep though (yuk)
Iterator support Add iterator support, used initially only for retrieving bucket statistics. The iterator is supported by exporting a function, and when the function is claled it will take a snapshot of the ledger, run the iterator and hten close the snapshot. This required a numbe rof underlying changes, in particular to get key comparison to work as "expected". The code had previously misunderstood how comparison worked between Erlang terms, and in particular did not account for tuple length being compared first by size of the tuple (and not just by each element in order).
2016-10-12 17:12:49 +01:00
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{B1Size, B1Count} = testutil:check_bucket_stats(Bookie2, "Bucket1"),
Iterator support Add iterator support, used initially only for retrieving bucket statistics. The iterator is supported by exporting a function, and when the function is claled it will take a snapshot of the ledger, run the iterator and hten close the snapshot. This required a numbe rof underlying changes, in particular to get key comparison to work as "expected". The code had previously misunderstood how comparison worked between Erlang terms, and in particular did not account for tuple length being compared first by size of the tuple (and not just by each element in order).
2016-10-12 17:12:49 +01:00
true = B1Size > 0,
true = B1Count == 1,
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{B1Size, B1Count} = testutil:check_bucket_stats(Bookie2, "Bucket1"),
{BSize, BCount} = testutil:check_bucket_stats(Bookie2, "Bucket"),
Iterator support Add iterator support, used initially only for retrieving bucket statistics. The iterator is supported by exporting a function, and when the function is claled it will take a snapshot of the ledger, run the iterator and hten close the snapshot. This required a numbe rof underlying changes, in particular to get key comparison to work as "expected". The code had previously misunderstood how comparison worked between Erlang terms, and in particular did not account for tuple length being compared first by size of the tuple (and not just by each element in order).
2016-10-12 17:12:49 +01:00
true = BSize > 0,
Close in stages - waiting for releases (#411) * Close in stages - waiting for releases Have a consistent approach to closing the inker and the penciller - so that the close can be interrupted by releasing of snapshots. Then any unreleased snapshots are closed before shutdown - with a 10s pause to give queries a short opportunity to finish. This should address some issues, primarily seen (but very rarely) in test whereby post-rebuild destruction of parallel AAE keystores cause the crashing of aae_folds. The primary benefit is to stop an attempt to release a snapshot that has in fact already finished does not cause a crash of the database on normal stop. this was primarily an issue when shutdown is delayed by an ongoing journal compaction job. * Boost default test budget for EQC * Update test to use correct type * Update following review Avoid filtering out exited PIDs when closing snapshots by catching the exit exception when the Pid is down
2023-10-03 18:30:40 +01:00
true = BCount == 180000,
io:format("Shutdown with overhanging snapshot~n"),
Iterator support Add iterator support, used initially only for retrieving bucket statistics. The iterator is supported by exporting a function, and when the function is claled it will take a snapshot of the ledger, run the iterator and hten close the snapshot. This required a numbe rof underlying changes, in particular to get key comparison to work as "expected". The code had previously misunderstood how comparison worked between Erlang terms, and in particular did not account for tuple length being compared first by size of the tuple (and not just by each element in order).
2016-10-12 17:12:49 +01:00
Close in stages - waiting for releases (#411) * Close in stages - waiting for releases Have a consistent approach to closing the inker and the penciller - so that the close can be interrupted by releasing of snapshots. Then any unreleased snapshots are closed before shutdown - with a 10s pause to give queries a short opportunity to finish. This should address some issues, primarily seen (but very rarely) in test whereby post-rebuild destruction of parallel AAE keystores cause the crashing of aae_folds. The primary benefit is to stop an attempt to release a snapshot that has in fact already finished does not cause a crash of the database on normal stop. this was primarily an issue when shutdown is delayed by an ongoing journal compaction job. * Boost default test budget for EQC * Update test to use correct type * Update following review Avoid filtering out exited PIDs when closing snapshots by catching the exit exception when the Pid is down
2023-10-03 18:30:40 +01:00
{ok, SnpPCL1, SnpJrnl1} =
leveled_bookie:book_snapshot(Bookie2, store, undefined, true),
{ok, SnpPCL2, SnpJrnl2} =
leveled_bookie:book_snapshot(Bookie2, store, undefined, true),
TestPid = self(),
spawn(
fun() ->
ok = leveled_bookie:book_close(Bookie2),
TestPid ! ok
end),
timer:sleep(5000),
ok = leveled_penciller:pcl_close(SnpPCL1),
ok = leveled_inker:ink_close(SnpJrnl1),
true = is_process_alive(SnpPCL2),
true = is_process_alive(SnpJrnl2),
io:format("Time for close to complete is 2 * 10s~n"),
io:format("Both Inker and Penciller will have snapshot delay~n"),
receive ok -> ok end,
false = is_process_alive(SnpPCL2),
false = is_process_alive(SnpJrnl2),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:reset_filestructure().
Add initial end-to-end common tests These tests highlighted some logical issues when scanning over databases on startup, so fixes are wrapped in here.
2016-10-05 09:54:53 +01:00
Mas d31 i413 (#415) * Allow snapshots to be reused in queries Allow for a full bookie snapshot to be re-used for multiple queries, not just KV fetches. * Reduce log noise The internal dummy tag is expected so should not prompt a log on reload * Snapshot should have same status of active db wrt head_only and head_lookup * Allow logging to specified on snapshots * Shutdown snapshot bookie is primary goes down Inker and Penciller already will shut down based on `erlang:monitor/2` * Review feedback Formatting and code readability fixes
2023-11-08 09:18:01 +00:00
compare_foldwithsnap(Bookie, SnapBookie, ChkList) ->
HeadFoldFun = fun(B, K, _Hd, Acc) -> [{B, K}|Acc] end,
KeyFoldFun = fun(B, K, Acc) -> [{B, K}|Acc] end,
{async, HeadFoldDB} =
leveled_bookie:book_headfold(
Bookie, ?RIAK_TAG, {HeadFoldFun, []}, true, false, false
),
{async, HeadFoldSnap} =
leveled_bookie:book_headfold(
SnapBookie, ?RIAK_TAG, {HeadFoldFun, []}, true, false, false
),
true = HeadFoldDB() == HeadFoldSnap(),
testutil:check_forlist(SnapBookie, ChkList),
{async, KeyFoldSnap} =
leveled_bookie:book_keylist(
SnapBookie, ?RIAK_TAG, {KeyFoldFun, []}
),
true = HeadFoldSnap() == KeyFoldSnap().
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
load_and_count(_Config) ->
% Use artificially small files, and the load keys, counting they're all
% present
Handle L0 cache being full A test thta will cause leveled to crash due to a low cache size being set - but protect against this (as well as the general scenario of the cache being full). There could be a potential case where a L0 file present (post pending) without work backlog being set. In this case we want to roll the level zero to memory, but don't accept the cache update if the L0 cache is already full.
2019-01-14 12:27:51 +00:00
load_and_count(50000000, 2500, 28000),
Develop 3.1 d30update (#386) * Mas i370 patch d (#383) * Refactor penciller memory In high-volume tests on large key-count clusters, so significant variation in the P0031 time has been seen: TimeBucket PatchA a.0ms_to_1ms 18554 b.1ms_to_2ms 51778 c.2ms_to_3ms 696 d.3ms_to_5ms 220 e.5ms_to_8ms 59 f.8ms_to_13ms 40 g.13ms_to_21ms 364 h.21ms_to_34ms 277 i.34ms_to_55ms 34 j.55ms_to_89ms 17 k.89ms_to_144ms 21 l.144ms_to_233ms 31 m.233ms_to_377ms 45 n.377ms_to_610ms 52 o.610ms_to_987ms 59 p.987ms_to_1597ms 55 q.1597ms_to_2684ms 54 r.2684ms_to_4281ms 29 s.4281ms_to_6965ms 7 t.6295ms_to_11246ms 1 It is unclear why this varies so much. The time to add to the cache appears to be minimal (but perhaps there is an issue with timing points in the code), whereas the time to add to the index is much more significant and variable. There is also variable time when the memory is rolled (although the actual activity here appears to be minimal. The refactoring here is two-fold: - tidy and simplify by keeping LoopState managed within handle_call, and add more helpful dialyzer specs; - change the update to the index to be a simple extension of a list, rather than any conversion. This alternative version of the pmem index in unit test is orders of magnitude faster to add - and is the same order of magnitude to check. Anticipation is that it may be more efficient in terms of memory changes. * Compress SST index Reduces the size of the leveled_sst index with two changes: 1 - Where there is a common prefix of tuple elements (e.g. Bucket) across the whole leveled_sst file - only the non-common part is indexed, and a function is used to compare. 2 - There is less "indexing" of the index i.e. only 1 in 16 keys are passed into the gb_trees part instead of 1 in 4 * Immediate hibernate Reasons for delay in hibernate were not clear. Straight after creation the process will not be in receipt of messages (must wait for the manifest to be updated), so better to hibernate now. This also means the log PC023 provides more accurate information. * Refactor BIC This patch avoids the following: - repeated replacement of the same element in the BIC (via get_kvrange), by checking presence via GET before sing SET - Stops re-reading of all elements to discover high modified date Also there appears to have been a bug where a missing HMD for the file is required to add to the cache. However, now the cache may be erased without erasing the HMD. This means that the cache can never be rebuilt * Use correct size in test results erts_debug:flat_size/1 returns size in words (i.e. 8 bytes on 64-bit CPU) not bytes * Don't change summary record As it is persisted as part of the file write, any change to the summary record cannot be rolled back * Clerk to prompt L0 write Simplifies the logic if the clerk request work for the penciller prompts L0 writes as well as Manifest changes. The advantage now is that if the penciller memory is full, and PUT load stops, the clerk should still be able to prompt persistence. the penciller can therefore make use of dead time this way * Add push on journal compact If there has been a backlog, followed by a quiet period - there may be a large ledger cache left unpushed. Journal compaction events are about once per hour, so the performance overhead of a false push should be minimal, with the advantage of clearing any backlog before load starts again. This is only relevant to riak users with very off/full batch type workloads. * Extend tests To more consistently trigger all overload scenarios * Fix range keys smaller than prefix Can't make end key an empty binary in this case, as it may be bigger than any keys within the range, but will appear to be smaller. Unit tests and ct tests added to expose the potential issue * Tidy-up - Remove penciller logs which are no longer called - Get pclerk to only wait MIN_TIMEOUT after doing work, in case there is a backlog - Remove update_levelzero_cache function as it is unique to handle_call of push_mem, and simple enough to be inline - Alight testutil slow offer with standard slow offer used * Tidy-up Remove pre-otp20 references. Reinstate the check that the starting pid is still active, this was added to tidy up shutdown. Resolve failure to run on otp20 due to `-if` sttaement * Tidy up Using null rather then {null, Key} is potentially clearer as it is not a concern what they Key is in this case, and removes a comparison step from the leveled_codec:endkey_passed/2 function. There were issues with coverage in eunit tests as the leveled_pclerk shut down. This prompted a general tidy of leveled_pclerk (remove passing of LoopState into internal functions, and add dialyzer specs. * Remove R16 relic * Further testing another issue The StartKey must always be less than or equal to the prefix when the first N characters are stripped, but this is not true of the EndKey (for the query) which does not have to be between the FirstKey and the LastKey. If the EndKey query does not match it must be greater than the Prefix (as otherwise it would not have been greater than the FirstKey - so set to null. * Fix unit test Unit test had a typo - and result interpretation had a misunderstanding. * Code and spec tidy Also look to the cover the situation when the FirstKey is the same as the Prefix with tests. This is, in theory, not an issue as it is the EndKey for each sublist which is indexed in leveled_tree. However, guard against it mapping to null here, just in case there are dangers lurking (note that tests will still pass without `M > N` guard in place. * Hibernate on BIC complete There are three situations when the BIC becomes complete: - In a file created as part of a merge the BIS is learned in the merge - After startup, files below L1 learn the block cache through reads that happen to read the block, eventually the while cache will be read, unless... - Either before/after the cache is complete, it can get whiped by a timeout after a get_sqn request (e.g. as prompted by a journal compaction) ... it will then be re-filled of the back of get/get-range requests. In all these situations we want to hibernate after the BIC is fill - to reflect the fact that the LoopState should now be relatively stable, so it is a good point to GC and rationalise location of data. Previously on the the first base was covered. Now all three are covered through the bic_complete message. * Test all index keys have same term This works functionally, but is not optimised (the term is replicated in the index) * Summaries with same index term If the summary index all have the same index term - only the object keys need to be indexes * Simplify case statements We either match the pattern of <<Prefix:N, Suffix>> or the answer should be null * OK for M == N If M = N for the first key, it will have a suffix of <<>>. This will match (as expected) a query Start Key of the sam size, and be smaller than any query Start Key that has the same prefix. If the query Start Key does not match the prefix - it will be null - as it must be smaller than the Prefix (as other wise the query Start Key would be bigger than the Last Key). The constraint of M > N was introduced before the *_prefix_filter functions were checking the prefix, to avoid issues. Now the prefix is being checked, then M == N is ok. * Simplify Correct the test to use a binary field in the range. To avoid further issue, only apply filter when everything is a binary() type. * Add test for head_only mode When leveled is used as a tictacaae key store (in parallel mode), the keys will be head_only entries. Double check they are handled as expected like object keys * Revert previous change - must support typed buckets Add assertion to confirm worthwhile optimisation * Add support for configurable cache multiple (#375) * Mas i370 patch e (#385) Improvement to monitoring for efficiency and improved readability of logs and stats. As part of this, where possible, tried to avoid updating loop state on READ messages in leveled processes (as was the case when tracking stats within each process). No performance benefits found with change, but improved stats has helped discover other potential gains.
2022-12-18 20:18:03 +00:00
load_and_count(200000000, 50, 200000),
load_and_count(50000000, 1000, 5000).
Handle L0 cache being full A test thta will cause leveled to crash due to a low cache size being set - but protect against this (as well as the general scenario of the cache being full). There could be a potential case where a L0 file present (post pending) without work backlog being set. In this case we want to roll the level zero to memory, but don't accept the cache update if the L0 cache is already full.
2019-01-14 12:27:51 +00:00
load_and_count(JournalSize, BookiesMemSize, PencillerMemSize) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
RootPath = testutil:reset_filestructure(),
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
StartOpts1 = [{root_path, RootPath},
Handle L0 cache being full A test thta will cause leveled to crash due to a low cache size being set - but protect against this (as well as the general scenario of the cache being full). There could be a potential case where a L0 file present (post pending) without work backlog being set. In this case we want to roll the level zero to memory, but don't accept the cache update if the L0 cache is already full.
2019-01-14 12:27:51 +00:00
{max_journalsize, JournalSize},
{cache_size, BookiesMemSize},
{max_pencillercachesize, PencillerMemSize},
Develop 3.1 d30update (#386) * Mas i370 patch d (#383) * Refactor penciller memory In high-volume tests on large key-count clusters, so significant variation in the P0031 time has been seen: TimeBucket PatchA a.0ms_to_1ms 18554 b.1ms_to_2ms 51778 c.2ms_to_3ms 696 d.3ms_to_5ms 220 e.5ms_to_8ms 59 f.8ms_to_13ms 40 g.13ms_to_21ms 364 h.21ms_to_34ms 277 i.34ms_to_55ms 34 j.55ms_to_89ms 17 k.89ms_to_144ms 21 l.144ms_to_233ms 31 m.233ms_to_377ms 45 n.377ms_to_610ms 52 o.610ms_to_987ms 59 p.987ms_to_1597ms 55 q.1597ms_to_2684ms 54 r.2684ms_to_4281ms 29 s.4281ms_to_6965ms 7 t.6295ms_to_11246ms 1 It is unclear why this varies so much. The time to add to the cache appears to be minimal (but perhaps there is an issue with timing points in the code), whereas the time to add to the index is much more significant and variable. There is also variable time when the memory is rolled (although the actual activity here appears to be minimal. The refactoring here is two-fold: - tidy and simplify by keeping LoopState managed within handle_call, and add more helpful dialyzer specs; - change the update to the index to be a simple extension of a list, rather than any conversion. This alternative version of the pmem index in unit test is orders of magnitude faster to add - and is the same order of magnitude to check. Anticipation is that it may be more efficient in terms of memory changes. * Compress SST index Reduces the size of the leveled_sst index with two changes: 1 - Where there is a common prefix of tuple elements (e.g. Bucket) across the whole leveled_sst file - only the non-common part is indexed, and a function is used to compare. 2 - There is less "indexing" of the index i.e. only 1 in 16 keys are passed into the gb_trees part instead of 1 in 4 * Immediate hibernate Reasons for delay in hibernate were not clear. Straight after creation the process will not be in receipt of messages (must wait for the manifest to be updated), so better to hibernate now. This also means the log PC023 provides more accurate information. * Refactor BIC This patch avoids the following: - repeated replacement of the same element in the BIC (via get_kvrange), by checking presence via GET before sing SET - Stops re-reading of all elements to discover high modified date Also there appears to have been a bug where a missing HMD for the file is required to add to the cache. However, now the cache may be erased without erasing the HMD. This means that the cache can never be rebuilt * Use correct size in test results erts_debug:flat_size/1 returns size in words (i.e. 8 bytes on 64-bit CPU) not bytes * Don't change summary record As it is persisted as part of the file write, any change to the summary record cannot be rolled back * Clerk to prompt L0 write Simplifies the logic if the clerk request work for the penciller prompts L0 writes as well as Manifest changes. The advantage now is that if the penciller memory is full, and PUT load stops, the clerk should still be able to prompt persistence. the penciller can therefore make use of dead time this way * Add push on journal compact If there has been a backlog, followed by a quiet period - there may be a large ledger cache left unpushed. Journal compaction events are about once per hour, so the performance overhead of a false push should be minimal, with the advantage of clearing any backlog before load starts again. This is only relevant to riak users with very off/full batch type workloads. * Extend tests To more consistently trigger all overload scenarios * Fix range keys smaller than prefix Can't make end key an empty binary in this case, as it may be bigger than any keys within the range, but will appear to be smaller. Unit tests and ct tests added to expose the potential issue * Tidy-up - Remove penciller logs which are no longer called - Get pclerk to only wait MIN_TIMEOUT after doing work, in case there is a backlog - Remove update_levelzero_cache function as it is unique to handle_call of push_mem, and simple enough to be inline - Alight testutil slow offer with standard slow offer used * Tidy-up Remove pre-otp20 references. Reinstate the check that the starting pid is still active, this was added to tidy up shutdown. Resolve failure to run on otp20 due to `-if` sttaement * Tidy up Using null rather then {null, Key} is potentially clearer as it is not a concern what they Key is in this case, and removes a comparison step from the leveled_codec:endkey_passed/2 function. There were issues with coverage in eunit tests as the leveled_pclerk shut down. This prompted a general tidy of leveled_pclerk (remove passing of LoopState into internal functions, and add dialyzer specs. * Remove R16 relic * Further testing another issue The StartKey must always be less than or equal to the prefix when the first N characters are stripped, but this is not true of the EndKey (for the query) which does not have to be between the FirstKey and the LastKey. If the EndKey query does not match it must be greater than the Prefix (as otherwise it would not have been greater than the FirstKey - so set to null. * Fix unit test Unit test had a typo - and result interpretation had a misunderstanding. * Code and spec tidy Also look to the cover the situation when the FirstKey is the same as the Prefix with tests. This is, in theory, not an issue as it is the EndKey for each sublist which is indexed in leveled_tree. However, guard against it mapping to null here, just in case there are dangers lurking (note that tests will still pass without `M > N` guard in place. * Hibernate on BIC complete There are three situations when the BIC becomes complete: - In a file created as part of a merge the BIS is learned in the merge - After startup, files below L1 learn the block cache through reads that happen to read the block, eventually the while cache will be read, unless... - Either before/after the cache is complete, it can get whiped by a timeout after a get_sqn request (e.g. as prompted by a journal compaction) ... it will then be re-filled of the back of get/get-range requests. In all these situations we want to hibernate after the BIC is fill - to reflect the fact that the LoopState should now be relatively stable, so it is a good point to GC and rationalise location of data. Previously on the the first base was covered. Now all three are covered through the bic_complete message. * Test all index keys have same term This works functionally, but is not optimised (the term is replicated in the index) * Summaries with same index term If the summary index all have the same index term - only the object keys need to be indexes * Simplify case statements We either match the pattern of <<Prefix:N, Suffix>> or the answer should be null * OK for M == N If M = N for the first key, it will have a suffix of <<>>. This will match (as expected) a query Start Key of the sam size, and be smaller than any query Start Key that has the same prefix. If the query Start Key does not match the prefix - it will be null - as it must be smaller than the Prefix (as other wise the query Start Key would be bigger than the Last Key). The constraint of M > N was introduced before the *_prefix_filter functions were checking the prefix, to avoid issues. Now the prefix is being checked, then M == N is ok. * Simplify Correct the test to use a binary field in the range. To avoid further issue, only apply filter when everything is a binary() type. * Add test for head_only mode When leveled is used as a tictacaae key store (in parallel mode), the keys will be head_only entries. Double check they are handled as expected like object keys * Revert previous change - must support typed buckets Add assertion to confirm worthwhile optimisation * Add support for configurable cache multiple (#375) * Mas i370 patch e (#385) Improvement to monitoring for efficiency and improved readability of logs and stats. As part of this, where possible, tried to avoid updating loop state on READ messages in leveled processes (as was the case when tracking stats within each process). No performance benefits found with change, but improved stats has helped discover other potential gains.
2022-12-18 20:18:03 +00:00
{sync_strategy, testutil:sync_strategy()},
{stats_logfrequency, 5},
{stats_probability, 80}],
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{TestObject, TestSpec} = testutil:generate_testobject(),
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
ok = testutil:book_riakput(Bookie1, TestObject, TestSpec),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
io:format("Loading initial small objects~n"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
G1 = fun testutil:generate_smallobjects/2,
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
lists:foldl(fun(_X, Acc) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:load_objects(5000,
[Acc + 2],
Bookie1,
TestObject,
G1),
Search range failure Discovered a bug with search ranges in leveled_tree - this was uncovered by an intermittently fialing 19.3 test. Test case added and bug fixed. It was due to a fialure to use end_key passed causing issues with particular manifests and full bucket ranges.
2017-10-24 13:19:30 +01:00
{_S, Count} =
testutil:check_bucket_stats(Bookie1, "Bucket"),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
if
Acc + 5000 == Count ->
ok
end,
Acc + 5000 end,
0,
lists:seq(1, 20)),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
io:format("Loading larger compressible objects~n"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
G2 = fun testutil:generate_compressibleobjects/2,
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
lists:foldl(fun(_X, Acc) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:load_objects(5000,
[Acc + 2],
Bookie1,
TestObject,
G2),
Search range failure Discovered a bug with search ranges in leveled_tree - this was uncovered by an intermittently fialing 19.3 test. Test case added and bug fixed. It was due to a fialure to use end_key passed causing issues with particular manifests and full bucket ranges.
2017-10-24 13:19:30 +01:00
{_S, Count} =
testutil:check_bucket_stats(Bookie1, "Bucket"),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
if
Acc + 5000 == Count ->
ok
end,
Acc + 5000 end,
100000,
lists:seq(1, 20)),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
io:format("Replacing small objects~n"),
lists:foldl(fun(_X, Acc) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:load_objects(5000,
[Acc + 2],
Bookie1,
TestObject,
G1),
Search range failure Discovered a bug with search ranges in leveled_tree - this was uncovered by an intermittently fialing 19.3 test. Test case added and bug fixed. It was due to a fialure to use end_key passed causing issues with particular manifests and full bucket ranges.
2017-10-24 13:19:30 +01:00
{_S, Count} =
testutil:check_bucket_stats(Bookie1, "Bucket"),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
if
Count == 200000 ->
ok
end,
Acc + 5000 end,
0,
lists:seq(1, 20)),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
io:format("Loading more small objects~n"),
Develop 3.1 d30update (#386) * Mas i370 patch d (#383) * Refactor penciller memory In high-volume tests on large key-count clusters, so significant variation in the P0031 time has been seen: TimeBucket PatchA a.0ms_to_1ms 18554 b.1ms_to_2ms 51778 c.2ms_to_3ms 696 d.3ms_to_5ms 220 e.5ms_to_8ms 59 f.8ms_to_13ms 40 g.13ms_to_21ms 364 h.21ms_to_34ms 277 i.34ms_to_55ms 34 j.55ms_to_89ms 17 k.89ms_to_144ms 21 l.144ms_to_233ms 31 m.233ms_to_377ms 45 n.377ms_to_610ms 52 o.610ms_to_987ms 59 p.987ms_to_1597ms 55 q.1597ms_to_2684ms 54 r.2684ms_to_4281ms 29 s.4281ms_to_6965ms 7 t.6295ms_to_11246ms 1 It is unclear why this varies so much. The time to add to the cache appears to be minimal (but perhaps there is an issue with timing points in the code), whereas the time to add to the index is much more significant and variable. There is also variable time when the memory is rolled (although the actual activity here appears to be minimal. The refactoring here is two-fold: - tidy and simplify by keeping LoopState managed within handle_call, and add more helpful dialyzer specs; - change the update to the index to be a simple extension of a list, rather than any conversion. This alternative version of the pmem index in unit test is orders of magnitude faster to add - and is the same order of magnitude to check. Anticipation is that it may be more efficient in terms of memory changes. * Compress SST index Reduces the size of the leveled_sst index with two changes: 1 - Where there is a common prefix of tuple elements (e.g. Bucket) across the whole leveled_sst file - only the non-common part is indexed, and a function is used to compare. 2 - There is less "indexing" of the index i.e. only 1 in 16 keys are passed into the gb_trees part instead of 1 in 4 * Immediate hibernate Reasons for delay in hibernate were not clear. Straight after creation the process will not be in receipt of messages (must wait for the manifest to be updated), so better to hibernate now. This also means the log PC023 provides more accurate information. * Refactor BIC This patch avoids the following: - repeated replacement of the same element in the BIC (via get_kvrange), by checking presence via GET before sing SET - Stops re-reading of all elements to discover high modified date Also there appears to have been a bug where a missing HMD for the file is required to add to the cache. However, now the cache may be erased without erasing the HMD. This means that the cache can never be rebuilt * Use correct size in test results erts_debug:flat_size/1 returns size in words (i.e. 8 bytes on 64-bit CPU) not bytes * Don't change summary record As it is persisted as part of the file write, any change to the summary record cannot be rolled back * Clerk to prompt L0 write Simplifies the logic if the clerk request work for the penciller prompts L0 writes as well as Manifest changes. The advantage now is that if the penciller memory is full, and PUT load stops, the clerk should still be able to prompt persistence. the penciller can therefore make use of dead time this way * Add push on journal compact If there has been a backlog, followed by a quiet period - there may be a large ledger cache left unpushed. Journal compaction events are about once per hour, so the performance overhead of a false push should be minimal, with the advantage of clearing any backlog before load starts again. This is only relevant to riak users with very off/full batch type workloads. * Extend tests To more consistently trigger all overload scenarios * Fix range keys smaller than prefix Can't make end key an empty binary in this case, as it may be bigger than any keys within the range, but will appear to be smaller. Unit tests and ct tests added to expose the potential issue * Tidy-up - Remove penciller logs which are no longer called - Get pclerk to only wait MIN_TIMEOUT after doing work, in case there is a backlog - Remove update_levelzero_cache function as it is unique to handle_call of push_mem, and simple enough to be inline - Alight testutil slow offer with standard slow offer used * Tidy-up Remove pre-otp20 references. Reinstate the check that the starting pid is still active, this was added to tidy up shutdown. Resolve failure to run on otp20 due to `-if` sttaement * Tidy up Using null rather then {null, Key} is potentially clearer as it is not a concern what they Key is in this case, and removes a comparison step from the leveled_codec:endkey_passed/2 function. There were issues with coverage in eunit tests as the leveled_pclerk shut down. This prompted a general tidy of leveled_pclerk (remove passing of LoopState into internal functions, and add dialyzer specs. * Remove R16 relic * Further testing another issue The StartKey must always be less than or equal to the prefix when the first N characters are stripped, but this is not true of the EndKey (for the query) which does not have to be between the FirstKey and the LastKey. If the EndKey query does not match it must be greater than the Prefix (as otherwise it would not have been greater than the FirstKey - so set to null. * Fix unit test Unit test had a typo - and result interpretation had a misunderstanding. * Code and spec tidy Also look to the cover the situation when the FirstKey is the same as the Prefix with tests. This is, in theory, not an issue as it is the EndKey for each sublist which is indexed in leveled_tree. However, guard against it mapping to null here, just in case there are dangers lurking (note that tests will still pass without `M > N` guard in place. * Hibernate on BIC complete There are three situations when the BIC becomes complete: - In a file created as part of a merge the BIS is learned in the merge - After startup, files below L1 learn the block cache through reads that happen to read the block, eventually the while cache will be read, unless... - Either before/after the cache is complete, it can get whiped by a timeout after a get_sqn request (e.g. as prompted by a journal compaction) ... it will then be re-filled of the back of get/get-range requests. In all these situations we want to hibernate after the BIC is fill - to reflect the fact that the LoopState should now be relatively stable, so it is a good point to GC and rationalise location of data. Previously on the the first base was covered. Now all three are covered through the bic_complete message. * Test all index keys have same term This works functionally, but is not optimised (the term is replicated in the index) * Summaries with same index term If the summary index all have the same index term - only the object keys need to be indexes * Simplify case statements We either match the pattern of <<Prefix:N, Suffix>> or the answer should be null * OK for M == N If M = N for the first key, it will have a suffix of <<>>. This will match (as expected) a query Start Key of the sam size, and be smaller than any query Start Key that has the same prefix. If the query Start Key does not match the prefix - it will be null - as it must be smaller than the Prefix (as other wise the query Start Key would be bigger than the Last Key). The constraint of M > N was introduced before the *_prefix_filter functions were checking the prefix, to avoid issues. Now the prefix is being checked, then M == N is ok. * Simplify Correct the test to use a binary field in the range. To avoid further issue, only apply filter when everything is a binary() type. * Add test for head_only mode When leveled is used as a tictacaae key store (in parallel mode), the keys will be head_only entries. Double check they are handled as expected like object keys * Revert previous change - must support typed buckets Add assertion to confirm worthwhile optimisation * Add support for configurable cache multiple (#375) * Mas i370 patch e (#385) Improvement to monitoring for efficiency and improved readability of logs and stats. As part of this, where possible, tried to avoid updating loop state on READ messages in leveled processes (as was the case when tracking stats within each process). No performance benefits found with change, but improved stats has helped discover other potential gains.
2022-12-18 20:18:03 +00:00
io:format("Now with unused snapshot so deletions are blocked~n"),
{ok, PclClone, null} =
leveled_bookie:book_snapshot(Bookie1, ledger, undefined, true),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
lists:foldl(fun(_X, Acc) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:load_objects(5000,
[Acc + 2],
Bookie1,
TestObject,
G2),
Search range failure Discovered a bug with search ranges in leveled_tree - this was uncovered by an intermittently fialing 19.3 test. Test case added and bug fixed. It was due to a fialure to use end_key passed causing issues with particular manifests and full bucket ranges.
2017-10-24 13:19:30 +01:00
{_S, Count} =
testutil:check_bucket_stats(Bookie1, "Bucket"),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
if
Acc + 5000 == Count ->
ok
end,
Acc + 5000 end,
200000,
lists:seq(1, 20)),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Develop 3.1 d30update (#386) * Mas i370 patch d (#383) * Refactor penciller memory In high-volume tests on large key-count clusters, so significant variation in the P0031 time has been seen: TimeBucket PatchA a.0ms_to_1ms 18554 b.1ms_to_2ms 51778 c.2ms_to_3ms 696 d.3ms_to_5ms 220 e.5ms_to_8ms 59 f.8ms_to_13ms 40 g.13ms_to_21ms 364 h.21ms_to_34ms 277 i.34ms_to_55ms 34 j.55ms_to_89ms 17 k.89ms_to_144ms 21 l.144ms_to_233ms 31 m.233ms_to_377ms 45 n.377ms_to_610ms 52 o.610ms_to_987ms 59 p.987ms_to_1597ms 55 q.1597ms_to_2684ms 54 r.2684ms_to_4281ms 29 s.4281ms_to_6965ms 7 t.6295ms_to_11246ms 1 It is unclear why this varies so much. The time to add to the cache appears to be minimal (but perhaps there is an issue with timing points in the code), whereas the time to add to the index is much more significant and variable. There is also variable time when the memory is rolled (although the actual activity here appears to be minimal. The refactoring here is two-fold: - tidy and simplify by keeping LoopState managed within handle_call, and add more helpful dialyzer specs; - change the update to the index to be a simple extension of a list, rather than any conversion. This alternative version of the pmem index in unit test is orders of magnitude faster to add - and is the same order of magnitude to check. Anticipation is that it may be more efficient in terms of memory changes. * Compress SST index Reduces the size of the leveled_sst index with two changes: 1 - Where there is a common prefix of tuple elements (e.g. Bucket) across the whole leveled_sst file - only the non-common part is indexed, and a function is used to compare. 2 - There is less "indexing" of the index i.e. only 1 in 16 keys are passed into the gb_trees part instead of 1 in 4 * Immediate hibernate Reasons for delay in hibernate were not clear. Straight after creation the process will not be in receipt of messages (must wait for the manifest to be updated), so better to hibernate now. This also means the log PC023 provides more accurate information. * Refactor BIC This patch avoids the following: - repeated replacement of the same element in the BIC (via get_kvrange), by checking presence via GET before sing SET - Stops re-reading of all elements to discover high modified date Also there appears to have been a bug where a missing HMD for the file is required to add to the cache. However, now the cache may be erased without erasing the HMD. This means that the cache can never be rebuilt * Use correct size in test results erts_debug:flat_size/1 returns size in words (i.e. 8 bytes on 64-bit CPU) not bytes * Don't change summary record As it is persisted as part of the file write, any change to the summary record cannot be rolled back * Clerk to prompt L0 write Simplifies the logic if the clerk request work for the penciller prompts L0 writes as well as Manifest changes. The advantage now is that if the penciller memory is full, and PUT load stops, the clerk should still be able to prompt persistence. the penciller can therefore make use of dead time this way * Add push on journal compact If there has been a backlog, followed by a quiet period - there may be a large ledger cache left unpushed. Journal compaction events are about once per hour, so the performance overhead of a false push should be minimal, with the advantage of clearing any backlog before load starts again. This is only relevant to riak users with very off/full batch type workloads. * Extend tests To more consistently trigger all overload scenarios * Fix range keys smaller than prefix Can't make end key an empty binary in this case, as it may be bigger than any keys within the range, but will appear to be smaller. Unit tests and ct tests added to expose the potential issue * Tidy-up - Remove penciller logs which are no longer called - Get pclerk to only wait MIN_TIMEOUT after doing work, in case there is a backlog - Remove update_levelzero_cache function as it is unique to handle_call of push_mem, and simple enough to be inline - Alight testutil slow offer with standard slow offer used * Tidy-up Remove pre-otp20 references. Reinstate the check that the starting pid is still active, this was added to tidy up shutdown. Resolve failure to run on otp20 due to `-if` sttaement * Tidy up Using null rather then {null, Key} is potentially clearer as it is not a concern what they Key is in this case, and removes a comparison step from the leveled_codec:endkey_passed/2 function. There were issues with coverage in eunit tests as the leveled_pclerk shut down. This prompted a general tidy of leveled_pclerk (remove passing of LoopState into internal functions, and add dialyzer specs. * Remove R16 relic * Further testing another issue The StartKey must always be less than or equal to the prefix when the first N characters are stripped, but this is not true of the EndKey (for the query) which does not have to be between the FirstKey and the LastKey. If the EndKey query does not match it must be greater than the Prefix (as otherwise it would not have been greater than the FirstKey - so set to null. * Fix unit test Unit test had a typo - and result interpretation had a misunderstanding. * Code and spec tidy Also look to the cover the situation when the FirstKey is the same as the Prefix with tests. This is, in theory, not an issue as it is the EndKey for each sublist which is indexed in leveled_tree. However, guard against it mapping to null here, just in case there are dangers lurking (note that tests will still pass without `M > N` guard in place. * Hibernate on BIC complete There are three situations when the BIC becomes complete: - In a file created as part of a merge the BIS is learned in the merge - After startup, files below L1 learn the block cache through reads that happen to read the block, eventually the while cache will be read, unless... - Either before/after the cache is complete, it can get whiped by a timeout after a get_sqn request (e.g. as prompted by a journal compaction) ... it will then be re-filled of the back of get/get-range requests. In all these situations we want to hibernate after the BIC is fill - to reflect the fact that the LoopState should now be relatively stable, so it is a good point to GC and rationalise location of data. Previously on the the first base was covered. Now all three are covered through the bic_complete message. * Test all index keys have same term This works functionally, but is not optimised (the term is replicated in the index) * Summaries with same index term If the summary index all have the same index term - only the object keys need to be indexes * Simplify case statements We either match the pattern of <<Prefix:N, Suffix>> or the answer should be null * OK for M == N If M = N for the first key, it will have a suffix of <<>>. This will match (as expected) a query Start Key of the sam size, and be smaller than any query Start Key that has the same prefix. If the query Start Key does not match the prefix - it will be null - as it must be smaller than the Prefix (as other wise the query Start Key would be bigger than the Last Key). The constraint of M > N was introduced before the *_prefix_filter functions were checking the prefix, to avoid issues. Now the prefix is being checked, then M == N is ok. * Simplify Correct the test to use a binary field in the range. To avoid further issue, only apply filter when everything is a binary() type. * Add test for head_only mode When leveled is used as a tictacaae key store (in parallel mode), the keys will be head_only entries. Double check they are handled as expected like object keys * Revert previous change - must support typed buckets Add assertion to confirm worthwhile optimisation * Add support for configurable cache multiple (#375) * Mas i370 patch e (#385) Improvement to monitoring for efficiency and improved readability of logs and stats. As part of this, where possible, tried to avoid updating loop state on READ messages in leveled processes (as was the case when tracking stats within each process). No performance benefits found with change, but improved stats has helped discover other potential gains.
2022-12-18 20:18:03 +00:00
ok = leveled_penciller:pcl_close(PclClone),
{_S, 300000} = testutil:check_bucket_stats(Bookie1, "Bucket"),
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
ok = leveled_bookie:book_close(Bookie1),
{ok, Bookie2} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{_, 300000} = testutil:check_bucket_stats(Bookie2, "Bucket"),
Close in stages - waiting for releases (#411) * Close in stages - waiting for releases Have a consistent approach to closing the inker and the penciller - so that the close can be interrupted by releasing of snapshots. Then any unreleased snapshots are closed before shutdown - with a 10s pause to give queries a short opportunity to finish. This should address some issues, primarily seen (but very rarely) in test whereby post-rebuild destruction of parallel AAE keystores cause the crashing of aae_folds. The primary benefit is to stop an attempt to release a snapshot that has in fact already finished does not cause a crash of the database on normal stop. this was primarily an issue when shutdown is delayed by an ongoing journal compaction job. * Boost default test budget for EQC * Update test to use correct type * Update following review Avoid filtering out exited PIDs when closing snapshots by catching the exit exception when the Pid is down
2023-10-03 18:30:40 +01:00
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
ok = leveled_bookie:book_close(Bookie2),
Close in stages - waiting for releases (#411) * Close in stages - waiting for releases Have a consistent approach to closing the inker and the penciller - so that the close can be interrupted by releasing of snapshots. Then any unreleased snapshots are closed before shutdown - with a 10s pause to give queries a short opportunity to finish. This should address some issues, primarily seen (but very rarely) in test whereby post-rebuild destruction of parallel AAE keystores cause the crashing of aae_folds. The primary benefit is to stop an attempt to release a snapshot that has in fact already finished does not cause a crash of the database on normal stop. this was primarily an issue when shutdown is delayed by an ongoing journal compaction job. * Boost default test budget for EQC * Update test to use correct type * Update following review Avoid filtering out exited PIDs when closing snapshots by catching the exit exception when the Pid is down
2023-10-03 18:30:40 +01:00
GC manifest files when numbers skipped Otherwise list of old files perpetually grows
2019-05-23 10:16:15 +01:00
ManifestFP =
leveled_pmanifest:filepath(filename:join(RootPath, ?LEDGER_FP),
manifest),
IsManifest = fun(FN) -> filename:extension(FN) == ".man" end,
{ok, RawManList} = file:list_dir(ManifestFP),
ManList = lists:filter(IsManifest, RawManList),
io:format("Length of manifest file list ~w~n", [length(ManList)]),
true = length(ManList) =< 5,
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:reset_filestructure().
Load and Count test This test exposed two bugs: - Yet another set of off-by-one errors (really stupidly scanning the Manifest from Level 1 not Level 0) - The return of an old issue related to scanning the journal on load whereby we fail to go back to the previous file before the current SQN
2016-10-13 17:51:47 +01:00
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
load_and_count_withdelete(_Config) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
RootPath = testutil:reset_filestructure(),
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
StartOpts1 = [{root_path, RootPath},
{max_journalsize, 50000000},
{sync_strategy, testutil:sync_strategy()}],
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{TestObject, TestSpec} = testutil:generate_testobject(),
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
ok = testutil:book_riakput(Bookie1, TestObject, TestSpec),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
io:format("Loading initial small objects~n"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
G1 = fun testutil:generate_smallobjects/2,
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
lists:foldl(fun(_X, Acc) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:load_objects(5000,
[Acc + 2],
Bookie1,
TestObject,
G1),
{_S, Count} = testutil:check_bucket_stats(Bookie1,
"Bucket"),
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
if
Acc + 5000 == Count ->
ok
end,
Acc + 5000 end,
0,
lists:seq(1, 20)),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_forobject(Bookie1, TestObject),
Push log update through to cdb/sst Using the cdb_options and sst_options records
2018-12-11 20:42:00 +00:00
{BucketD, KeyD} =
{testutil:get_bucket(TestObject), testutil:get_key(TestObject)},
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{_, 1} = testutil:check_bucket_stats(Bookie1, BucketD),
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
ok = testutil:book_riakdelete(Bookie1, BucketD, KeyD, []),
not_found = testutil:book_riakget(Bookie1, BucketD, KeyD),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{_, 0} = testutil:check_bucket_stats(Bookie1, BucketD),
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
io:format("Loading larger compressible objects~n"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
G2 = fun testutil:generate_compressibleobjects/2,
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
lists:foldl(fun(_X, Acc) ->
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:load_objects(5000,
[Acc + 2],
Bookie1,
no_check,
G2),
{_S, Count} = testutil:check_bucket_stats(Bookie1,
"Bucket"),
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
if
Acc + 5000 == Count ->
ok
end,
Acc + 5000 end,
100000,
lists:seq(1, 20)),
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
not_found = testutil:book_riakget(Bookie1, BucketD, KeyD),
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
ok = leveled_bookie:book_close(Bookie1),
Close in stages - waiting for releases (#411) * Close in stages - waiting for releases Have a consistent approach to closing the inker and the penciller - so that the close can be interrupted by releasing of snapshots. Then any unreleased snapshots are closed before shutdown - with a 10s pause to give queries a short opportunity to finish. This should address some issues, primarily seen (but very rarely) in test whereby post-rebuild destruction of parallel AAE keystores cause the crashing of aae_folds. The primary benefit is to stop an attempt to release a snapshot that has in fact already finished does not cause a crash of the database on normal stop. this was primarily an issue when shutdown is delayed by an ongoing journal compaction job. * Boost default test budget for EQC * Update test to use correct type * Update following review Avoid filtering out exited PIDs when closing snapshots by catching the exit exception when the Pid is down
2023-10-03 18:30:40 +01:00
Delete Add functionality to delete keys. No tombstone reaping yet.
2016-10-16 15:41:09 +01:00
{ok, Bookie2} = leveled_bookie:book_start(StartOpts1),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:check_formissingobject(Bookie2, BucketD, KeyD),
Test coverage - missing 1%
2017-02-26 20:52:40 +00:00
testutil:check_formissingobject(Bookie2, "Bookie1", "MissingKey0123"),
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
{_BSize, 0} = testutil:check_bucket_stats(Bookie2, BucketD),
ok = leveled_bookie:book_close(Bookie2),
Close in stages - waiting for releases (#411) * Close in stages - waiting for releases Have a consistent approach to closing the inker and the penciller - so that the close can be interrupted by releasing of snapshots. Then any unreleased snapshots are closed before shutdown - with a 10s pause to give queries a short opportunity to finish. This should address some issues, primarily seen (but very rarely) in test whereby post-rebuild destruction of parallel AAE keystores cause the crashing of aae_folds. The primary benefit is to stop an attempt to release a snapshot that has in fact already finished does not cause a crash of the database on normal stop. this was primarily an issue when shutdown is delayed by an ongoing journal compaction job. * Boost default test budget for EQC * Update test to use correct type * Update following review Avoid filtering out exited PIDs when closing snapshots by catching the exit exception when the Pid is down
2023-10-03 18:30:40 +01:00
Support for 2i query part1 Added basic support for 2i query. This involved some refactoring of the test code to share functions between suites. There is sill a need for a Part 2 as no tests currently cover removal of index entries.
2016-10-18 01:59:03 +01:00
testutil:reset_filestructure().
Add initial system test Add some initial system tests. This highlighted issues: - That files deleted by compaction would be left orphaned and not close, and would not in fact delete (now deleted by closure only) - There was an issue on stratup that the first few keys in each journal would not be re-loaded into the ledger
2016-10-03 23:34:28 +01:00
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
Recovery strategy testing Test added for the "retain" recovery strategy. This strategy makes sure a full history of index changes is made so that if the Ledger is wiped out, the Ledger cna be fully rebuilt from the Journal. This exposed two journal compaction problems - The BestRun selected did not have the source files correctly sorted in order before compaction - The compaction process incorrectly dealt with the KeyDelta object left after a compaction - i.e. compacting twice the same key caused that key history to be lost. These issues have now been corrected.
2016-10-27 00:57:19 +01:00
space_clear_ondelete(_Config) ->
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
RootPath = testutil:reset_filestructure(),
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
StartOpts1 = [{root_path, RootPath},
Change default compaction settings Need to allow specific settings to be passed into unit tests. Also, too much journal compaction may lead to intermittent failures on the basic_SUITE space_clear_on_delete test. think this is because there are less “deletes” to reload in on startup to trigger the cascade down and clear up?
2017-06-02 08:37:57 +01:00
{max_journalsize, 10000000},
Pass through sync_strategy Allow to switch for Riak to use o_sync as the sync flag rather than sync
2016-11-25 17:41:08 +00:00
{sync_strategy, testutil:sync_strategy()}],
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
{ok, Book1} = leveled_bookie:book_start(StartOpts1),
G2 = fun testutil:generate_compressibleobjects/2,
testutil:load_objects(20000,
[uuid, uuid, uuid, uuid],
Book1,
no_check,
G2),
Fix bad accumulator in test Test takes a long time due to sleep (still need to work on that), but also FoldKeysFun uses ++ rather than [{B, K}|Acc] to extend the list. Order of magnitude speed-up for these queries by changing the way this accumulates
2018-05-01 22:30:28 +01:00
FoldKeysFun = fun(B, K, Acc) -> [{B, K}|Acc] end,
Provide a top level API for folds As the fold functions have been added to get_runner in an ad hoc way, naturally, given the ongoing development of levelEd to support Riak, it was difficult for a new user (in this case Quviq) to see what folds are supported, and with what arguments, and expectations. This PR is for discussion. It is one of many ways to group, spec, and document the fold functions. A test is also added for coverage of range queries.
2018-08-23 10:27:43 +01:00
{async, F1} = leveled_bookie:book_keylist(Book1, o_rkv, {FoldKeysFun, []}),
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
SW1 = os:timestamp(),
KL1 = F1(),
ok = case length(KL1) of
80000 ->
io:format("Key list took ~w microseconds for 80K keys~n",
[timer:now_diff(os:timestamp(), SW1)]),
ok
end,
timer:sleep(10000), % Allow for any L0 file to be rolled
{ok, FNsA_L} = file:list_dir(RootPath ++ "/ledger/ledger_files"),
{ok, FNsA_J} = file:list_dir(RootPath ++ "/journal/journal_files"),
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
io:format("FNsA - Bookie created ~w journal files and ~w ledger files~n",
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
[length(FNsA_J), length(FNsA_L)]),
Provide a top level API for folds As the fold functions have been added to get_runner in an ad hoc way, naturally, given the ongoing development of levelEd to support Riak, it was difficult for a new user (in this case Quviq) to see what folds are supported, and with what arguments, and expectations. This PR is for discussion. It is one of many ways to group, spec, and document the fold functions. A test is also added for coverage of range queries.
2018-08-23 10:27:43 +01:00
Registering and releasing of Journal snapshots Added a test of journal compaction with a registered snapshot and it showed that the deleting of files did not correctly check the list of registerd snapshots. Corrected.
2016-11-04 15:56:57 +00:00
% Get an iterator to lock the inker during compaction
Switch to binary format Riak object Initial change to try and test assuming that leveled received the binary format of Riak objects (and parses that for metadata).
2016-11-28 22:26:09 +00:00
FoldObjectsFun = fun(B, K, ObjBin, Acc) ->
[{B, K, erlang:phash2(ObjBin)}|Acc] end,
Provide a top level API for folds As the fold functions have been added to get_runner in an ad hoc way, naturally, given the ongoing development of levelEd to support Riak, it was difficult for a new user (in this case Quviq) to see what folds are supported, and with what arguments, and expectations. This PR is for discussion. It is one of many ways to group, spec, and document the fold functions. A test is also added for coverage of range queries.
2018-08-23 10:27:43 +01:00
{async, HTreeF1} = leveled_bookie:book_objectfold(Book1,
?RIAK_TAG,
{FoldObjectsFun, []},
false),
Fix bad accumulator in test Test takes a long time due to sleep (still need to work on that), but also FoldKeysFun uses ++ rather than [{B, K}|Acc] to extend the list. Order of magnitude speed-up for these queries by changing the way this accumulates
2018-05-01 22:30:28 +01:00
% This query does not Snap PreFold - and so will not prevent
% pending deletes from prompting actual deletes
Provide a top level API for folds As the fold functions have been added to get_runner in an ad hoc way, naturally, given the ongoing development of levelEd to support Riak, it was difficult for a new user (in this case Quviq) to see what folds are supported, and with what arguments, and expectations. This PR is for discussion. It is one of many ways to group, spec, and document the fold functions. A test is also added for coverage of range queries.
2018-08-23 10:27:43 +01:00
{async, KF1} = leveled_bookie:book_keylist(Book1, o_rkv, {FoldKeysFun, []}),
Fix bad accumulator in test Test takes a long time due to sleep (still need to work on that), but also FoldKeysFun uses ++ rather than [{B, K}|Acc] to extend the list. Order of magnitude speed-up for these queries by changing the way this accumulates
2018-05-01 22:30:28 +01:00
% This query does Snap PreFold, and so will prevent deletes from
% the ledger
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
Registering and releasing of Journal snapshots Added a test of journal compaction with a registered snapshot and it showed that the deleting of files did not correctly check the list of registerd snapshots. Corrected.
2016-11-04 15:56:57 +00:00
% Delete the keys
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
SW2 = os:timestamp(),
lists:foreach(fun({Bucket, Key}) ->
Shift pause out of leveled Leveled will now signal the need for a pause due to back-pressure, but not actually pause itself. The hope is that in a riak implementation this pause can be managed by the put_fsm, and so not lock the store.
2016-11-07 10:27:38 +00:00
testutil:book_riakdelete(Book1,
Clean API of Riak-specific Methods Clena the API of Riak specific methods, and also resolve timing issue in simple_server unit test. Previously this would end up with missing data (and a lower sequence number after start) because of the penciller_clerk timeout being relatively large in the context of this test. Now the timeout has bene reduced the L0 slot is cleared by the time of the close. To make sure an extra sleep has been added as a precaution to avoid any intermittent issues.
2016-11-07 10:11:57 +00:00
Bucket,
Key,
[])
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
end,
KL1),
io:format("Deletion took ~w microseconds for 80K keys~n",
[timer:now_diff(os:timestamp(), SW2)]),
Registering and releasing of Journal snapshots Added a test of journal compaction with a registered snapshot and it showed that the deleting of files did not correctly check the list of registerd snapshots. Corrected.
2016-11-04 15:56:57 +00:00
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
ok = leveled_bookie:book_compactjournal(Book1, 30000),
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
F = fun leveled_bookie:book_islastcompactionpending/1,
lists:foldl(fun(X, Pending) ->
case Pending of
false ->
false;
true ->
io:format("Loop ~w waiting for journal "
++ "compaction to complete~n", [X]),
timer:sleep(20000),
F(Book1)
end end,
true,
lists:seq(1, 15)),
Registering and releasing of Journal snapshots Added a test of journal compaction with a registered snapshot and it showed that the deleting of files did not correctly check the list of registerd snapshots. Corrected.
2016-11-04 15:56:57 +00:00
io:format("Waiting for journal deletes - blocked~n"),
timer:sleep(20000),
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
Fix bad accumulator in test Test takes a long time due to sleep (still need to work on that), but also FoldKeysFun uses ++ rather than [{B, K}|Acc] to extend the list. Order of magnitude speed-up for these queries by changing the way this accumulates
2018-05-01 22:30:28 +01:00
io:format("Sleep over - Fold Objects query ~n"),
% for this query snapshot is made at fold time, and so the results are
% empty
Split code for folders - introduce runner actor Introduce a dedicated module for all the different fold types. Also simplify the list of folders by deprecating those folds that should eb achieveable by fold_heads/fold_objects type folds but with smarter functions. Makes sure that the fold functiosn also have better spec coverage, and are dialyzer checked.
2017-10-17 20:39:11 +01:00
true = length(HTreeF1()) == 0,
Fix bad accumulator in test Test takes a long time due to sleep (still need to work on that), but also FoldKeysFun uses ++ rather than [{B, K}|Acc] to extend the list. Order of magnitude speed-up for these queries by changing the way this accumulates
2018-05-01 22:30:28 +01:00
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
% This query uses a genuine async fold on a snasphot made at request time
Fix bad accumulator in test Test takes a long time due to sleep (still need to work on that), but also FoldKeysFun uses ++ rather than [{B, K}|Acc] to extend the list. Order of magnitude speed-up for these queries by changing the way this accumulates
2018-05-01 22:30:28 +01:00
% and so the results should be non-empty
io:format("Now Query 2 - Fold Keys query~n"),
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
true = length(KF1()) == 80000,
Registering and releasing of Journal snapshots Added a test of journal compaction with a registered snapshot and it showed that the deleting of files did not correctly check the list of registerd snapshots. Corrected.
2016-11-04 15:56:57 +00:00
io:format("Waiting for journal deletes - unblocked~n"),
timer:sleep(20000),
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
{ok, FNsB_L} = file:list_dir(RootPath ++ "/ledger/ledger_files"),
{ok, FNsB_J} = file:list_dir(RootPath ++ "/journal/journal_files"),
Registering and releasing of Journal snapshots Added a test of journal compaction with a registered snapshot and it showed that the deleting of files did not correctly check the list of registerd snapshots. Corrected.
2016-11-04 15:56:57 +00:00
{ok, FNsB_PC} = file:list_dir(RootPath
++ "/journal/journal_files/post_compact"),
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
PointB_Journals = length(FNsB_J) + length(FNsB_PC),
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
io:format("FNsB - Bookie has ~w journal files and ~w ledger files " ++
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
"after deletes~n",
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
[PointB_Journals, length(FNsB_L)]),
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
Provide a top level API for folds As the fold functions have been added to get_runner in an ad hoc way, naturally, given the ongoing development of levelEd to support Riak, it was difficult for a new user (in this case Quviq) to see what folds are supported, and with what arguments, and expectations. This PR is for discussion. It is one of many ways to group, spec, and document the fold functions. A test is also added for coverage of range queries.
2018-08-23 10:27:43 +01:00
{async, F2} = leveled_bookie:book_keylist(Book1, o_rkv, {FoldKeysFun, []}),
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
SW3 = os:timestamp(),
KL2 = F2(),
ok = case length(KL2) of
0 ->
io:format("Key list took ~w microseconds for no keys~n",
[timer:now_diff(os:timestamp(), SW3)]),
ok
end,
ok = leveled_bookie:book_close(Book1),
{ok, Book2} = leveled_bookie:book_start(StartOpts1),
Provide a top level API for folds As the fold functions have been added to get_runner in an ad hoc way, naturally, given the ongoing development of levelEd to support Riak, it was difficult for a new user (in this case Quviq) to see what folds are supported, and with what arguments, and expectations. This PR is for discussion. It is one of many ways to group, spec, and document the fold functions. A test is also added for coverage of range queries.
2018-08-23 10:27:43 +01:00
{async, F3} = leveled_bookie:book_keylist(Book2, o_rkv, {FoldKeysFun, []}),
Basement Tombstones Further progress towards the tidying up of basement tombstones in the Ledger, with support added for key-listing to help with testing (and as a potentially required feature). The test is incomplete, but committing at this stage as the last commit broke some tests (within the test code). There are some outstanding questions about the handling of tombstones in the Journal during compaction. There exists a condition whereby values could return if a recent journal is compacted and tombstones are removed (as they are no longer present), but older journals have not been compacted. Now on stop/start - if the Ledger is wiped the removal of the keys will be forgotten but the original PUTs would still remain. The safest thing maybe to have rule that tombstones are never deleted from the Inker's Journal - and accept the build-up of garbage. Or there could be an addition to the compaction process that checks back through all the inker files to check that the Key of a tombstone is not present in the past, before it is removed in the compaction.
2016-10-23 22:45:43 +01:00
SW4 = os:timestamp(),
KL3 = F3(),
ok = case length(KL3) of
0 ->
io:format("Key list took ~w microseconds for no keys~n",
[timer:now_diff(os:timestamp(), SW4)]),
ok
end,
ok = leveled_bookie:book_close(Book2),
{ok, FNsC_L} = file:list_dir(RootPath ++ "/ledger/ledger_files"),
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
io:format("FNsC - Bookie has ~w ledger files " ++
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
"after close~n", [length(FNsC_L)]),
{ok, Book3} = leveled_bookie:book_start(StartOpts1),
io:format("This should cause a final ledger merge event~n"),
io:format("Will require the penciller to resolve the issue of creating" ++
" an empty file as all keys compact on merge~n"),
Roll in fix for intermittently failing test As descibed in https://github.com/martinsumner/leveled/issues/92 Only the first fix was made. Just to eb safe - archiving means renaming to another file with a different extension. Assumption is that renamed files cna be manually reaped if necessary.
2017-09-27 23:52:49 +01:00
CheckFun =
fun(X, FileCount) ->
case FileCount of
0 ->
0;
_ ->
timer:sleep(X),
{ok, NewFC} =
file:list_dir(RootPath ++ "/ledger/ledger_files"),
io:format("Looping with ledger file count ~w~n",
[length(NewFC)]),
length(strip_nonsst(NewFC))
end
end,
FC = lists:foldl(CheckFun, infinity, [2000, 3000, 5000, 8000]),
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
ok = leveled_bookie:book_close(Book3),
Roll in fix for intermittently failing test As descibed in https://github.com/martinsumner/leveled/issues/92 Only the first fix was made. Just to eb safe - archiving means renaming to another file with a different extension. Assumption is that renamed files cna be manually reaped if necessary.
2017-09-27 23:52:49 +01:00
case FC of
0 ->
ok;
_ ->
{ok, Book4} = leveled_bookie:book_start(StartOpts1),
lists:foldl(CheckFun, infinity, [2000, 3000, 5000, 8000]),
leveled_bookie:book_close(Book4)
end,
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
{ok, FNsD_L} = file:list_dir(RootPath ++ "/ledger/ledger_files"),
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
io:format("FNsD - Bookie has ~w ledger files " ++
Roll in fix for intermittently failing test As descibed in https://github.com/martinsumner/leveled/issues/92 Only the first fix was made. Just to eb safe - archiving means renaming to another file with a different extension. Assumption is that renamed files cna be manually reaped if necessary.
2017-09-27 23:52:49 +01:00
"after second close~n", [length(strip_nonsst(FNsD_L))]),
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
lists:foreach(fun(FN) ->
Change default compaction settings Need to allow specific settings to be passed into unit tests. Also, too much journal compaction may lead to intermittent failures on the basic_SUITE space_clear_on_delete test. think this is because there are less “deletes” to reload in on startup to trigger the cascade down and clear up?
2017-06-02 08:37:57 +01:00
io:format("FNsD - Ledger file is ~s~n", [FN])
Correct async fold fold objects which snaps in the fold was implemented incorrectly - it took information from the LedgeCache at the point of the request, not at the point of the fold. So the LedgerCache SQN may have been surpassed in the Penciller by the time the fold was called.
2017-04-17 23:01:55 +01:00
end,
FNsD_L),
CDB - switch to gen_fsm The CDB file management server has distinct states, and was growing case logic to prevent certain messages from being handled in ceratin states, and to handle different messages differently. So this has now been converted to a gen_fsm. As part of resolving this, the space_clear_ondelete test has been completed, and completing this revealed that the Penciller could not cope with a change which emptied the ledger. So a series of changes has been handled to allow it to smoothly progress to an empty manifest.
2016-10-26 20:39:16 +01:00
true = PointB_Journals < length(FNsA_J),
Roll in fix for intermittently failing test As descibed in https://github.com/martinsumner/leveled/issues/92 Only the first fix was made. Just to eb safe - archiving means renaming to another file with a different extension. Assumption is that renamed files cna be manually reaped if necessary.
2017-09-27 23:52:49 +01:00
true = length(strip_nonsst(FNsD_L)) < length(strip_nonsst(FNsA_L)),
true = length(strip_nonsst(FNsD_L)) < length(strip_nonsst(FNsB_L)),
Mas d31 i410looptoclose (#421) * Mas i410 looptoclose (#420) * Stop waiting full SHUTDOWN_PAUSE If there is a snapshot outstanding at shutdown time, there was a wait of SHUTDOWN_PAUSE to give the snapshot time to close down. This causes an issue in kv_index_tictactree when rebuilds complete, when an exchange was in flight at the point the rebuild completed - the aae_controller will become blocked for the full shutdown pause, whilst it waits for the replaced key store to be closed. This change is to loop within the shutdown pause, so that if the snapshot supporting the exchange is closed, the paused bookie can close more quickly (unblocking the controller). Without this fix, there are intermittent issues in kv_index_tictactree's mockvnode_SUITE tests. * Address test reliability Be a bit clearer with waiting round seconds, Was intermittently failing on QR4 previously (but QR5 1s later was always OK). * Update iterator_SUITE.erl * Refine test assertion At Stage C there might be 0 files left, in which case equality with Stage D result is ok.
2023-11-10 15:04:47 +00:00
true = length(strip_nonsst(FNsD_L)) =< length(strip_nonsst(FNsC_L)),
Roll in fix for intermittently failing test As descibed in https://github.com/martinsumner/leveled/issues/92 Only the first fix was made. Just to eb safe - archiving means renaming to another file with a different extension. Assumption is that renamed files cna be manually reaped if necessary.
2017-09-27 23:52:49 +01:00
true = length(strip_nonsst(FNsD_L)) == 0.
Add test for is_empty Bucket listing didn't care if keys were active - now does.
2017-05-23 11:59:44 +01:00
Roll in fix for intermittently failing test As descibed in https://github.com/martinsumner/leveled/issues/92 Only the first fix was made. Just to eb safe - archiving means renaming to another file with a different extension. Assumption is that renamed files cna be manually reaped if necessary.
2017-09-27 23:52:49 +01:00
strip_nonsst(FileList) ->
SSTOnlyFun =
fun(FN, Acc) ->
case filename:extension(FN) of
".sst" ->
[FN|Acc];
_ ->
Acc
end
end,
lists:foldl(SSTOnlyFun, [], FileList).
Add test for is_empty Bucket listing didn't care if keys were active - now does.
2017-05-23 11:59:44 +01:00
is_empty_test(_Config) ->
RootPath = testutil:reset_filestructure(),
StartOpts1 = [{root_path, RootPath},
{sync_strategy, testutil:sync_strategy()}],
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
{B1, K1, V1, Spec, MD} = {term_to_binary("Bucket1"),
term_to_binary("Key1"),
"Value1",
[],
[{"MDK1", "MDV1"}]},
{TestObject1, TestSpec1} =
testutil:generate_testobject(B1, K1, V1, Spec, MD),
{B1, K2, V2, Spec, MD} = {term_to_binary("Bucket1"),
term_to_binary("Key2"),
"Value2",
[],
[{"MDK1", "MDV1"}]},
{TestObject2, TestSpec2} =
testutil:generate_testobject(B1, K2, V2, Spec, MD),
{B2, K3, V3, Spec, MD} = {term_to_binary("Bucket2"),
term_to_binary("Key3"),
"Value3",
[],
[{"MDK1", "MDV1"}]},
{TestObject3, TestSpec3} =
testutil:generate_testobject(B2, K3, V3, Spec, MD),
ok = testutil:book_riakput(Bookie1, TestObject1, TestSpec1),
ok = testutil:book_riakput(Bookie1, TestObject2, TestSpec2),
ok = testutil:book_riakput(Bookie1, TestObject3, TestSpec3),
FoldBucketsFun = fun(B, Acc) -> sets:add_element(B, Acc) end,
Switch from binary_bucketlist Allow for bucket listing of non-binary buckets (integer buckets, buckets with ascii strings)
2018-09-01 10:39:23 +01:00
BucketListQuery = {bucket_list,
Add test for is_empty Bucket listing didn't care if keys were active - now does.
2017-05-23 11:59:44 +01:00
?RIAK_TAG,
{FoldBucketsFun, sets:new()}},
{async, BL} = leveled_bookie:book_returnfolder(Bookie1, BucketListQuery),
true = sets:size(BL()) == 2,
ok = leveled_bookie:book_put(Bookie1, B2, K3, delete, [], ?RIAK_TAG),
{async, BLpd1} = leveled_bookie:book_returnfolder(Bookie1, BucketListQuery),
true = sets:size(BLpd1()) == 1,
ok = leveled_bookie:book_put(Bookie1, B1, K2, delete, [], ?RIAK_TAG),
{async, BLpd2} = leveled_bookie:book_returnfolder(Bookie1, BucketListQuery),
true = sets:size(BLpd2()) == 1,
ok = leveled_bookie:book_put(Bookie1, B1, K1, delete, [], ?RIAK_TAG),
{async, BLpd3} = leveled_bookie:book_returnfolder(Bookie1, BucketListQuery),
true = sets:size(BLpd3()) == 0,
ok = leveled_bookie:book_close(Bookie1).
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
Test to expose startup issue See #194
2018-10-03 18:29:20 +01:00
remove_journal_test(_Config) ->
RootPath = testutil:reset_filestructure(),
StartOpts1 = [{root_path, RootPath},
{max_pencillercachesize, 16000},
{sync_strategy, testutil:sync_strategy()},
{compression_point, on_compact}],
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
GenList = [1, 20001, 40001, 60001],
CLs = testutil:load_objects(20000, GenList, Bookie1, no_check,
fun testutil:generate_smallobjects/2),
CheckList1 = lists:sublist(lists:nth(1, CLs), 100, 1000),
CheckList2 = lists:sublist(lists:nth(2, CLs), 100, 1000),
CheckList3 = lists:sublist(lists:nth(3, CLs), 100, 1000),
CheckList4 = lists:sublist(lists:nth(4, CLs), 100, 1000),
testutil:check_forlist(Bookie1, CheckList1),
testutil:check_forlist(Bookie1, CheckList2),
testutil:check_forlist(Bookie1, CheckList3),
testutil:check_forlist(Bookie1, CheckList4),
ok = leveled_bookie:book_close(Bookie1),
leveled_inker:clean_testdir(RootPath ++ "/journal"),
{ok, Bookie2} = leveled_bookie:book_start(StartOpts1),
% If we're not careful here new data will be added, and we
% won't be able to read it
[NewCheckList] =
testutil:load_objects(1000, [80001], Bookie2, no_check,
fun testutil:generate_smallobjects/2),
ok = leveled_bookie:book_close(Bookie2),
{ok, Bookie3} = leveled_bookie:book_start(StartOpts1),
testutil:check_forlist(Bookie3, NewCheckList),
ok = leveled_bookie:book_destroy(Bookie3).
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
many_put_fetch_switchcompression(_Config) ->
Add ZSTD compression (#430) * Add support for zstd and split compression Add support for using zstd as an alternative to native, lz4. Upgrade lz4 to v1.9.4 (with ARM enhancements). Allow for split compression algorithms - i.e. use native on journal, but lz4 on ledger. * Switch to AdRoll zstd Development appears to be active and ongoing. No issues running on different linux flavours. * Use realistic bucket name * Update README.md * Switch branch * Add comment following review
2024-01-23 16:25:03 +00:00
{T0, ok} =
timer:tc(fun many_put_fetch_switchcompression_tester/1, [native]),
{T1, ok} =
timer:tc(fun many_put_fetch_switchcompression_tester/1, [lz4]),
{T2, ok} =
timer:tc(fun many_put_fetch_switchcompression_tester/1, [zstd]),
io:format("Test timings native=~w lz4=~w, zstd=~w", [T0, T1, T2]).
many_put_fetch_switchcompression_tester(CompressionMethod) ->
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
RootPath = testutil:reset_filestructure(),
StartOpts1 = [{root_path, RootPath},
{max_pencillercachesize, 16000},
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
{max_journalobjectcount, 30000},
{compression_level, 3},
{sync_strategy, testutil:sync_strategy()},
Add ZSTD compression (#430) * Add support for zstd and split compression Add support for using zstd as an alternative to native, lz4. Upgrade lz4 to v1.9.4 (with ARM enhancements). Allow for split compression algorithms - i.e. use native on journal, but lz4 on ledger. * Switch to AdRoll zstd Development appears to be active and ongoing. No issues running on different linux flavours. * Use realistic bucket name * Update README.md * Switch branch * Add comment following review
2024-01-23 16:25:03 +00:00
{compression_method, native},
{ledger_compression, none}],
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
StartOpts2 = [{root_path, RootPath},
{max_pencillercachesize, 24000},
{max_journalobjectcount, 30000},
{sync_strategy, testutil:sync_strategy()},
Add ZSTD compression (#430) * Add support for zstd and split compression Add support for using zstd as an alternative to native, lz4. Upgrade lz4 to v1.9.4 (with ARM enhancements). Allow for split compression algorithms - i.e. use native on journal, but lz4 on ledger. * Switch to AdRoll zstd Development appears to be active and ongoing. No issues running on different linux flavours. * Use realistic bucket name * Update README.md * Switch branch * Add comment following review
2024-01-23 16:25:03 +00:00
{compression_method, CompressionMethod},
{ledger_compression, as_store}],
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
StartOpts3 = [{root_path, RootPath},
{max_pencillercachesize, 16000},
{max_journalobjectcount, 30000},
{sync_strategy, testutil:sync_strategy()},
Add ZSTD compression (#430) * Add support for zstd and split compression Add support for using zstd as an alternative to native, lz4. Upgrade lz4 to v1.9.4 (with ARM enhancements). Allow for split compression algorithms - i.e. use native on journal, but lz4 on ledger. * Switch to AdRoll zstd Development appears to be active and ongoing. No issues running on different linux flavours. * Use realistic bucket name * Update README.md * Switch branch * Add comment following review
2024-01-23 16:25:03 +00:00
{compression_method, none},
{ledger_compression, as_store}],
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
{TestObject, TestSpec} = testutil:generate_testobject(),
ok = testutil:book_riakput(Bookie1, TestObject, TestSpec),
testutil:check_forobject(Bookie1, TestObject),
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
CL1s =
testutil:load_objects(
40000,
[2, 40002],
Bookie1,
TestObject,
fun testutil:generate_smallobjects/2),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie1, CL) end, CL1s),
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
ok = leveled_bookie:book_close(Bookie1),
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
%% Change compression method -> lz4
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
{ok, Bookie2} = leveled_bookie:book_start(StartOpts2),
testutil:check_forobject(Bookie2, TestObject),
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie2, CL) end, CL1s),
CL2s =
testutil:load_objects(
40000,
[80002, 120002],
Bookie2,
TestObject,
fun testutil:generate_smallobjects/2),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie2, CL) end, CL2s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie2, CL) end, CL1s),
Make compression algorithm an option Compression can be switched between LZ4 and zlib (native). The setting to determine if compression should happen on receipt is now a macro definition in leveled_codec.
2017-11-06 15:54:58 +00:00
ok = leveled_bookie:book_close(Bookie2),
%% Change method back again
{ok, Bookie3} = leveled_bookie:book_start(StartOpts1),
testutil:check_formissingobject(Bookie3, "Bookie1", "MissingKey0123"),
Mas d31 i416 (#418) * Add compression controls (#417) * Add compression controls Add configuration options to allow for a compression algorithm of `none` to disable compression altogether. Also an option to change the point in the LSM tree when compression is applied. * Handle configurable defaults consistently Move them into leveled.hrl. This forces double-definitions to be resolved. There are some other constants in leveled_bookie that are relevant outside of leveled_bookie. These are all now in the non-configurable startup defaults section. * Clarify referred-to default is OTP not leveled * Update leveled_bookie.erl Handle xref issue with eunit include
2023-11-07 14:58:43 +00:00
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie3, CL) end, CL2s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie3, CL) end, CL1s),
CL3s =
testutil:load_objects(
40000,
[160002, 200002],
Bookie3,
TestObject,
fun testutil:generate_smallobjects/2,
30000
),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie3, CL) end, CL3s),
ok = leveled_bookie:book_close(Bookie3),
% Change method to no compression
{ok, Bookie4} = leveled_bookie:book_start(StartOpts3),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie4, CL) end, CL2s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie4, CL) end, CL1s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie4, CL) end, CL3s),
CL4s =
testutil:load_objects(
40000,
[240002, 280002],
Bookie4,
TestObject,
fun testutil:generate_smallobjects/2
),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie4, CL) end, CL3s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie4, CL) end, CL4s),
testutil:delete_some_objects(Bookie4, lists:flatten(CL3s), 60000),
CL5s =
testutil:load_objects(
40000,
[320002, 360002],
Bookie4,
TestObject,
fun testutil:generate_smallobjects/2
),
ok = leveled_bookie:book_compactjournal(Bookie4, 30000),
testutil:wait_for_compaction(Bookie4),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie4, CL) end, CL4s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie4, CL) end, CL5s),
ok = leveled_bookie:book_close(Bookie4),
%% Change compression method -> lz4
{ok, Bookie5} = leveled_bookie:book_start(StartOpts2),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie5, CL) end, CL1s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie5, CL) end, CL4s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie5, CL) end, CL5s),
ok = leveled_bookie:book_close(Bookie5),
%% Change compression method -> native
{ok, Bookie6} = leveled_bookie:book_start(StartOpts1),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie6, CL) end, CL1s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie6, CL) end, CL4s),
lists:foreach(
fun(CL) -> ok = testutil:check_forlist(Bookie6, CL) end, CL5s),
ok = leveled_bookie:book_destroy(Bookie6).
Add ct test Which passes
2018-09-27 11:32:42 +01:00
safereaderror_startup(_Config) ->
RootPath = testutil:reset_filestructure(),
StartOpts1 = [{root_path, RootPath},
{compression_point, on_compact},
{max_journalsize, 1000}, {cache_size, 2060}],
{ok, Bookie1} = leveled_bookie:book_plainstart(StartOpts1),
B1 = <<98, 117, 99, 107, 101, 116, 51>>,
K1 =
<<38, 50, 201, 47, 167, 125, 57, 232, 84, 38, 14, 114, 24, 62,
12, 74>>,
Obj1 =
<<87, 150, 217, 230, 4, 81, 170, 68, 181, 224, 60, 232, 4, 74,
159, 12, 156, 56, 194, 181, 18, 158, 195, 207, 106, 191, 80,
111, 100, 81, 252, 248>>,
Obj2 =
<<86, 201, 253, 149, 213, 10, 32, 166, 33, 136, 42, 79, 103, 250,
139, 95, 42, 143, 161, 3, 185, 74, 149, 226, 232, 214, 183, 64,
69, 56, 167, 78>>,
ok = leveled_bookie:book_put(Bookie1, B1, K1, Obj1, []),
ok = leveled_bookie:book_put(Bookie1, B1, K1, Obj2, []),
exit(Bookie1, kill),
{ok, Bookie2} = leveled_bookie:book_start(StartOpts1),
{ok, ReadBack} = leveled_bookie:book_get(Bookie2, B1, K1),
io:format("Read back ~w", [ReadBack]),
true = ReadBack == Obj2,
Add performance/profiling test (#424) * Add performance/profiling test Add test to perf_SUITE to do performance tests and also profile different activities in leveled. This can then be used to highlight functions with unexpectedly high execution times, and prove the impact of changes. Switch between riak_ctperf and riak_fullperf to change from standard test (with profile option) to full-scale performance test * Change shape of default perfTest * Change fullPerf Change the fullPerf test to run more tests, but with fewer keys. Given that RS of 512 is being pushed in Riak, 2M objects is till a 300M+ object cluster. 10M >> 1B. so these are still reasonable sizes to test. A profilePerf test also added to generate all the profiles base don 2M objects. * Extend test Queries where previously all returning a large number of index entries - changes made to make number of entries per query more realistic. Also an update process added to show difference between loading and rotating keys. * Relabel as AAE fold * Test v5 Test mini-queries - where generally a small number of entries are returned * Default to ctperf
2023-12-19 11:56:03 +00:00
ok = leveled_bookie:book_close(Bookie2).
bigpcl_bucketlist(_Config) ->
%% https://github.com/martinsumner/leveled/issues/326
%% In OTP 22+ there appear to be issues with anonymous functions which
%% have a reference to loop state, requiring a copy of all the loop state
%% to be made when returning the function.
%% This test creates alarge loop state on the leveled_penciller to prove
%% this.
%% The problem can be resolved simply by renaming the element of the loop
%% state using within the anonymous function.
RootPath = testutil:reset_filestructure(),
BucketCount = 500,
ObjectCount = 100,
StartOpts1 = [{root_path, RootPath},
{max_journalsize, 50000000},
{cache_size, 4000},
{max_pencillercachesize, 128000},
{max_sstslots, 256},
{sync_strategy, testutil:sync_strategy()}],
{ok, Bookie1} = leveled_bookie:book_start(StartOpts1),
BucketList =
lists:map(fun(I) -> list_to_binary(integer_to_list(I)) end,
lists:seq(1, BucketCount)),
MapFun =
fun(B) ->
testutil:generate_objects(ObjectCount, 1, [],
leveled_rand:rand_bytes(100),
fun() -> [] end,
B)
end,
ObjLofL = lists:map(MapFun, BucketList),
lists:foreach(fun(ObjL) -> testutil:riakload(Bookie1, ObjL) end, ObjLofL),
BucketFold =
fun(B, _K, _V, Acc) ->
case sets:is_element(B, Acc) of
true ->
Acc;
false ->
sets:add_element(B, Acc)
end
end,
FBAccT = {BucketFold, sets:new()},
{async, BucketFolder1} =
leveled_bookie:book_headfold(Bookie1,
?RIAK_TAG,
{bucket_list, BucketList},
FBAccT,
false, false, false),
{FoldTime1, BucketList1} = timer:tc(BucketFolder1, []),
true = BucketCount == sets:size(BucketList1),
ok = leveled_bookie:book_close(Bookie1),
{ok, Bookie2} = leveled_bookie:book_start(StartOpts1),
{async, BucketFolder2} =
leveled_bookie:book_headfold(Bookie2,
?RIAK_TAG,
{bucket_list, BucketList},
FBAccT,
false, false, false),
{FoldTime2, BucketList2} = timer:tc(BucketFolder2, []),
true = BucketCount == sets:size(BucketList2),
io:format("Fold pre-close ~w ms post-close ~w ms~n",
[FoldTime1 div 1000, FoldTime2 div 1000]),
true = FoldTime1 < 10 * FoldTime2,
%% The fold in-memory should be the same order of magnitude of response
%% time as the fold post-persistence
ok = leveled_bookie:book_destroy(Bookie2).
Reference in a new issue Copy permalink