Make snapshot timeout configurable

Also increase defaults.  In riak snapshots may be used to fold over all objects, and with delays at the receiving end, this could take significant time

include/leveled.hrl

@@ -78,7 +78,9 @@
                         source_penciller :: pid() | undefined,
                         snapshot_longrunning = true :: boolean(),
                         compression_method = native :: lz4|native,
-                        levelzero_cointoss = false :: boolean()}).
+                        levelzero_cointoss = false :: boolean(),
+                        snaptimeout_short :: pos_integer() | undefined,
+                        snaptimeout_long :: pos_integer() | undefined}).
 
 -record(iclerk_options,
                         {inker :: pid() | undefined,

src/leveled_bookie.erl

@@ -104,7 +104,6 @@
 -define(MIN_PCL_CACHE_SIZE, 400).
 -define(MAX_PCL_CACHE_SIZE, 28000). 
     % This is less than actual max - but COIN_SIDECOUNT
--define(SNAPSHOT_TIMEOUT, 300000).
 -define(CACHE_SIZE_JITTER, 25).
 -define(JOURNAL_SIZE_JITTER, 20).
 -define(ABSOLUTEMAX_JOURNALSIZE, 4000000000).
@@ -118,6 +117,8 @@
 -define(MAX_KEYCHECK_FREQUENCY, 100).
 -define(MIN_KEYCHECK_FREQUENCY, 1).
 -define(OPEN_LASTMOD_RANGE, {0, infinity}).
+-define(PCL_SNAPTIMEOUT_SHORT, 900). % 15 minutes
+-define(PCL_SNAPTIMEOUT_LONG, 43200). % 12 hours
 -define(OPTION_DEFAULTS,
             [{root_path, undefined},
                 {snapshot_bookie, undefined},
@@ -135,7 +136,9 @@
                 {compression_point, ?COMPRESSION_POINT},
                 {log_level, ?LOG_LEVEL},
                 {forced_logs, []},
-                {override_functions, []}]).
+                {override_functions, []},
+                {pcl_snapshottimeout_short, ?PCL_SNAPTIMEOUT_SHORT},
+                {pcl_snapshottimeout_long, ?PCL_SNAPTIMEOUT_LONG}]).
 
 -record(ledger_cache, {mem :: ets:tab(),
                         loader = leveled_tree:empty(?CACHE_TYPE)
@@ -328,9 +331,19 @@
             %       "P0032", "SST12", "CDB19", "SST13", "I0019"]}
             % Will log all timing points even when log_level is not set to
             % support info
-        {override_functions, list(leveled_head:appdefinable_function_tuple())}
+        {override_functions, list(leveled_head:appdefinable_function_tuple())} |
             % Provide a list of override functions that will be used for
             % user-defined tags
+        {pcl_snapshottimeout_short, pos_integer()} |
+            % Time in seconds before a snapshot that has not been shutdown is
+            % assumed to have failed, and so requires to be torndown.  The
+            % short timeout is applied to queries where long_running is set to
+            % false
+        {pcl_snapshottimeout_long, pos_integer()}
+            % Time in seconds before a snapshot that has not been shutdown is
+            % assumed to have failed, and so requires to be torndown.  The
+            % short timeout is applied to queries where long_running is set to
+            % true
         ].
 
 
@@ -1564,6 +1577,9 @@ set_options(Opts) ->
     SyncStrat = proplists:get_value(sync_strategy, Opts),
     WRP = proplists:get_value(waste_retention_period, Opts),
 
+    SnapTimeoutShort = proplists:get_value(pcl_snapshottimeout_short, Opts),
+    SnapTimeoutLong = proplists:get_value(pcl_snapshottimeout_long, Opts),
+
     AltStrategy = proplists:get_value(reload_strategy, Opts),
     ReloadStrategy = leveled_codec:inker_reload_strategy(AltStrategy),
 
@@ -1613,6 +1629,8 @@ set_options(Opts) ->
         #penciller_options{root_path = LedgerFP,
                             max_inmemory_tablesize = PCLL0CacheSize,
                             levelzero_cointoss = true,
+                            snaptimeout_short = SnapTimeoutShort,
+                            snaptimeout_long = SnapTimeoutLong,
                             sst_options =
                                 #sst_options{press_method = CompressionMethod,
                                             log_options=leveled_log:get_opts()}}

src/leveled_penciller.erl

@@ -236,8 +236,6 @@
 -define(COIN_SIDECOUNT, 5).
 -define(SLOW_FETCH, 20000).
 -define(ITERATOR_SCANWIDTH, 4).
--define(SNAPSHOT_TIMEOUT_LONG, 3600).
--define(SNAPSHOT_TIMEOUT_SHORT, 600).
 -define(TIMING_SAMPLECOUNTDOWN, 10000).
 -define(TIMING_SAMPLESIZE, 100).
 -define(OPEN_LASTMOD_RANGE, {0, infinity}).
@@ -270,6 +268,9 @@
                 timings = no_timing :: pcl_timings(),
                 timings_countdown = 0 :: integer(),
 
+                snaptimeout_short :: pos_integer()|undefined,
+                snaptimeout_long :: pos_integer()|undefined,
+
                 sst_options = #sst_options{} :: #sst_options{}}).
 
 -record(pcl_timings, 
@@ -633,7 +634,8 @@ init([LogOpts, PCLopts]) ->
             LongRunning = PCLopts#penciller_options.snapshot_longrunning,
             %% monitor the bookie, and close the snapshot when bookie
             %% exits
-	    BookieMonitor =  erlang:monitor(process, PCLopts#penciller_options.bookies_pid),
+	        BookieMonitor = 
+                erlang:monitor(process, PCLopts#penciller_options.bookies_pid),
 
             {ok, State} = pcl_registersnapshot(SrcPenciller, 
                                                 self(), 
@@ -793,24 +795,24 @@ handle_call({fetch_keys,
     end;
 handle_call(get_startup_sqn, _From, State) ->
     {reply, State#state.persisted_sqn, State};
-handle_call({register_snapshot, Snapshot, Query, BookiesMem, LR}, _From, State) ->
+handle_call({register_snapshot, Snapshot, Query, BookiesMem, LongRunning},
+                                                            _From, State) ->
     % Register and load a snapshot
     %
     % For setup of the snapshot to be efficient should pass a query
     % of (StartKey, EndKey) - this will avoid a fully copy of the penciller's
     % memory being required to be trasnferred to the clone.  However, this
     % will not be a valid clone for fetch
-    Timeout = 
-        case LR of
-            true ->
-                ?SNAPSHOT_TIMEOUT_LONG;
-            false ->
-                ?SNAPSHOT_TIMEOUT_SHORT
-        end,
     
-    Manifest0 = leveled_pmanifest:add_snapshot(State#state.manifest,
+    TimeO = 
-                                                Snapshot,
+        case LongRunning of
-                                                Timeout),
+            true ->
+                State#state.snaptimeout_long;
+            false ->
+                State#state.snaptimeout_short
+        end,
+    Manifest0 =
+        leveled_pmanifest:add_snapshot(State#state.manifest, Snapshot, TimeO),
 
     {BookieIncrTree, BookieIdx, MinSQN, MaxSQN} = BookiesMem,
     LM1Cache =
@@ -1091,6 +1093,9 @@ start_from_file(PCLopts) ->
     MaxTableSize = PCLopts#penciller_options.max_inmemory_tablesize,
     OptsSST = PCLopts#penciller_options.sst_options,
 
+    SnapTimeoutShort = PCLopts#penciller_options.snaptimeout_short,
+    SnapTimeoutLong = PCLopts#penciller_options.snaptimeout_long,
+    
     {ok, MergeClerk} = leveled_pclerk:clerk_new(self(), RootPath, OptsSST),
     
     CoinToss = PCLopts#penciller_options.levelzero_cointoss,
@@ -1103,6 +1108,8 @@ start_from_file(PCLopts) ->
                         levelzero_maxcachesize = MaxTableSize,
                         levelzero_cointoss = CoinToss,
                         levelzero_index = leveled_pmem:new_index(),
+                        snaptimeout_short = SnapTimeoutShort,
+                        snaptimeout_long = SnapTimeoutLong,
                         sst_options = OptsSST},
     
     %% Open manifest