honeymoose/OpenSearch
1
0
Fork 0
mirror of https://github.com/honeymoose/OpenSearch.git synced 2025-02-27 15:39:49 +00:00
Code Issues Packages Projects Releases Wiki Activity

Set global checkpoint before open engine from store (#27972)

Browse Source 
In PR #27965, we set the global checkpoint from the translog in a store
recovery. However, we set after an engine is opened. This causes the
global checkpoint assertion in TranslogWriter violated as if we are
forced to close the engine before we set the global checkpoint. A
closing engine will close translog which in turn read the current global
checkpoint; however it is still unassigned and smaller than the initial
global checkpoint from translog.

Closes #27970
This commit is contained in:
Nhat Nguyen 2017-12-23 04:13:27 -05:00 committed by Yannick Welsch
parent 6629f4ab0d
commit 436a243e3c
1 changed files with 7 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
core/src/main/java/org/elasticsearch/index/shard/IndexShard.java
View File

@ -1331,16 +1331,19 @@ public class IndexShard extends AbstractIndexShardComponent implements IndicesCl
// we disable deletes since we allow for operations to be executed against the shard while recovering
// but we need to make sure we don't loose deletes until we are done recovering
config.setEnableGcDeletes(false);
if (openMode == EngineConfig.OpenMode.OPEN_INDEX_AND_TRANSLOG) {
// we have to set it before we open an engine and recover from the translog because
// acquiring a snapshot from the translog causes a sync which causes the global checkpoint to be pulled in,
// and an engine can be forced to close in ctor which also causes the global checkpoint to be pulled in.
globalCheckpointTracker.updateGlobalCheckpointOnReplica(Translog.readGlobalCheckpoint(translogConfig.getTranslogPath()),
"read from translog checkpoint");
}
Engine newEngine = createNewEngine(config);
verifyNotClosed();
if (openMode == EngineConfig.OpenMode.OPEN_INDEX_AND_TRANSLOG) {
// We set active because we are now writing operations to the engine; this way, if we go idle after some time and become inactive,
// we still give sync'd flush a chance to run:
active.set(true);
// we have to set it before we recover from the translog as acquring a snapshot from the translog causes a sync which
// causes the global checkpoint to be pulled in.
globalCheckpointTracker.updateGlobalCheckpointOnReplica(getEngine().getTranslog().getLastSyncedGlobalCheckpoint(),
"read from translog");
newEngine.recoverFromTranslog();
}
assertSequenceNumbersInCommit();