honeymoose
/
OpenSearch
mirror of https://github.com/honeymoose/OpenSearch.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

Remove Needless Context Switch From Snapshot Finalization (#56871) (#59443) 

No need to do any switch to the `SNAPSHOT` pool here, the blob store
repo handles all its writes async on the `SNAPSHOT` pool so we're just
needlessly context-switching to enqueue those tasks there.
Also cleaned up the source only repository (the only override to `finalizeSnapshot`)
to make it clear that no IO is happening there and we don't need to run it on the
`SNAPSHOT` pool either.
This commit is contained in:
Armin Braun 2020-07-13 20:11:07 +02:00 committed by GitHub
parent 31be3a3645
commit bde92fc5fc
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 47 additions and 50 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

97
server/src/main/java/org/elasticsearch/snapshots/SnapshotsService.java
View File

@ -1074,27 +1074,24 @@ public class SnapshotsService extends AbstractLifecycleComponent implements Clus
removeSnapshotFromClusterState(entry.snapshot(), new SnapshotException(snapshot, "Aborted on initialization"), null); removeSnapshotFromClusterState(entry.snapshot(), new SnapshotException(snapshot, "Aborted on initialization"), null);
return; return;
} }
threadPool.executor(ThreadPool.Names.SNAPSHOT).execute(new AbstractRunnable() { try {
@Override final String failure = entry.failure();
protected void doRun() { logger.trace("[{}] finalizing snapshot in repository, state: [{}], failure[{}]", snapshot, entry.state(), failure);
final Repository repository = repositoriesService.repository(snapshot.getRepository()); ArrayList<SnapshotShardFailure> shardFailures = new ArrayList<>();
final String failure = entry.failure(); for (ObjectObjectCursor<ShardId, ShardSnapshotStatus> shardStatus : entry.shards()) {
logger.trace("[{}] finalizing snapshot in repository, state: [{}], failure[{}]", snapshot, entry.state(), failure); ShardId shardId = shardStatus.key;
ArrayList<SnapshotShardFailure> shardFailures = new ArrayList<>(); ShardSnapshotStatus status = shardStatus.value;
for (ObjectObjectCursor<ShardId, ShardSnapshotStatus> shardStatus : entry.shards()) { final ShardState state = status.state();
ShardId shardId = shardStatus.key; if (state.failed()) {
ShardSnapshotStatus status = shardStatus.value; shardFailures.add(new SnapshotShardFailure(status.nodeId(), shardId, status.reason()));
final ShardState state = status.state(); } else if (state.completed() == false) {
if (state.failed()) { shardFailures.add(new SnapshotShardFailure(status.nodeId(), shardId, "skipped"));
shardFailures.add(new SnapshotShardFailure(status.nodeId(), shardId, status.reason())); } else {
} else if (state.completed() == false) { assert state == ShardState.SUCCESS;
shardFailures.add(new SnapshotShardFailure(status.nodeId(), shardId, "skipped"));
} else {
assert state == ShardState.SUCCESS;
}
} }
final ShardGenerations shardGenerations = buildGenerations(entry, metadata); }
repository.finalizeSnapshot( final ShardGenerations shardGenerations = buildGenerations(entry, metadata);
repositoriesService.repository(snapshot.getRepository()).finalizeSnapshot(
snapshot.getSnapshotId(), snapshot.getSnapshotId(),
shardGenerations, shardGenerations,
entry.startTime(), entry.startTime(),
@ -1106,38 +1103,38 @@ public class SnapshotsService extends AbstractLifecycleComponent implements Clus
metadataForSnapshot(entry, metadata), metadataForSnapshot(entry, metadata),
entry.userMetadata(), entry.userMetadata(),
entry.version(), entry.version(),
state -> stateWithoutSnapshot(state, snapshot), state -> stateWithoutSnapshot(state, snapshot),
ActionListener.wrap(result -> { ActionListener.wrap(result -> {
final List<ActionListener<Tuple<RepositoryData, SnapshotInfo>>> completionListeners = final List<ActionListener<Tuple<RepositoryData, SnapshotInfo>>> completionListeners =
snapshotCompletionListeners.remove(snapshot); snapshotCompletionListeners.remove(snapshot);
if (completionListeners != null) { if (completionListeners != null) {
try { try {
ActionListener.onResponse(completionListeners, result); ActionListener.onResponse(completionListeners, result);
} catch (Exception e) { } catch (Exception e) {
logger.warn("Failed to notify listeners", e); logger.warn("Failed to notify listeners", e);
}
} }
endingSnapshots.remove(snapshot); }
logger.info("snapshot [{}] completed with state [{}]", snapshot, result.v2().state()); endingSnapshots.remove(snapshot);
}, this::onFailure)); logger.info("snapshot [{}] completed with state [{}]", snapshot, result.v2().state());
} }, e -> handleFinalizationFailure(e, entry)));
} catch (Exception e) {
handleFinalizationFailure(e, entry);
}
}
@Override private void handleFinalizationFailure(Exception e, SnapshotsInProgress.Entry entry) {
public void onFailure(final Exception e) { Snapshot snapshot = entry.snapshot();
Snapshot snapshot = entry.snapshot(); if (ExceptionsHelper.unwrap(e, NotMasterException.class, FailedToCommitClusterStateException.class) != null) {
if (ExceptionsHelper.unwrap(e, NotMasterException.class, FailedToCommitClusterStateException.class) != null) { // Failure due to not being master any more, don't try to remove snapshot from cluster state the next master
// Failure due to not being master any more, don't try to remove snapshot from cluster state the next master // will try ending this snapshot again
// will try ending this snapshot again logger.debug(() -> new ParameterizedMessage(
logger.debug(() -> new ParameterizedMessage( "[{}] failed to update cluster state during snapshot finalization", snapshot), e);
"[{}] failed to update cluster state during snapshot finalization", snapshot), e); failSnapshotCompletionListeners(snapshot,
failSnapshotCompletionListeners(snapshot, new SnapshotException(snapshot, "Failed to update cluster state during snapshot finalization", e));
new SnapshotException(snapshot, "Failed to update cluster state during snapshot finalization", e)); } else {
} else { logger.warn(() -> new ParameterizedMessage("[{}] failed to finalize snapshot", snapshot), e);
logger.warn(() -> new ParameterizedMessage("[{}] failed to finalize snapshot", snapshot), e); removeSnapshotFromClusterState(snapshot, e, null);
removeSnapshotFromClusterState(snapshot, e, null); }
}
}
});
} }
private static ClusterState stateWithoutSnapshot(ClusterState state, Snapshot snapshot) { private static ClusterState stateWithoutSnapshot(ClusterState state, Snapshot snapshot) {