Skip to content

Commit

Permalink
Fix SegmentReplicationPressureService to not schedule async tasks whe…
Browse files Browse the repository at this point in the history
…n remote failures are disabled (opensearch-project#10569)

* Fix SegmentReplicationPressureService to not schedule async tasks when remote failures are disabled

Today segment replication's pressure service provides a setting 'segrep.replication.time.limit' that if set to a positive value
will spawn an Async task every 30s validating if there are replicas over that limit.  If over the limit the task considers if shards should be failed remotely.
The async task is being rescheduled every 30s, even if the setting is set to 0.  This change ensures that task isn't scheduled at all and reconsiders only if
the setting is updated.

Signed-off-by: Marc Handalian <handalm@amazon.com>

* PR feedback.

Signed-off-by: Marc Handalian <handalm@amazon.com>

* Revert "PR feedback."

This reverts commit 7371161.

Signed-off-by: Marc Handalian <handalm@amazon.com>

* Add better unit test to ensure task is scheduled/unscheduled.

Signed-off-by: Marc Handalian <handalm@amazon.com>

---------

Signed-off-by: Marc Handalian <handalm@amazon.com>
  • Loading branch information
mch2 authored and rayshrey committed Mar 18, 2024
1 parent b14ec3a commit 5190107
Show file tree
Hide file tree
Showing 2 changed files with 53 additions and 3 deletions.
Original file line number Diff line number Diff line change
Expand Up @@ -98,7 +98,7 @@ public class SegmentReplicationPressureService implements Closeable {
private final SegmentReplicationStatsTracker tracker;
private final ShardStateAction shardStateAction;

private final AsyncFailStaleReplicaTask failStaleReplicaTask;
private volatile AsyncFailStaleReplicaTask failStaleReplicaTask;

@Inject
public SegmentReplicationPressureService(
Expand Down Expand Up @@ -202,6 +202,15 @@ public void setMaxAllowedStaleReplicas(double maxAllowedStaleReplicas) {

public void setReplicationTimeLimitFailReplica(TimeValue replicationTimeLimitFailReplica) {
this.replicationTimeLimitFailReplica = replicationTimeLimitFailReplica;
updateAsyncFailReplicaTask();
}

private synchronized void updateAsyncFailReplicaTask() {
try {
failStaleReplicaTask.close();
} finally {
failStaleReplicaTask = new AsyncFailStaleReplicaTask(this);
}
}

public void setReplicationTimeLimitBackpressure(TimeValue replicationTimeLimitBackpressure) {
Expand All @@ -228,13 +237,13 @@ final static class AsyncFailStaleReplicaTask extends AbstractAsyncTask {

@Override
protected boolean mustReschedule() {
return true;
return pressureService.shouldScheduleAsyncFailTask();
}

@Override
protected void runInternal() {
// Do not fail the replicas if time limit is set to 0 (i.e. disabled).
if (TimeValue.ZERO.equals(pressureService.replicationTimeLimitFailReplica) == false) {
if (pressureService.shouldScheduleAsyncFailTask()) {
final SegmentReplicationStats stats = pressureService.tracker.getStats();

// Find the shardId in node which is having stale replicas with highest current replication time.
Expand Down Expand Up @@ -302,4 +311,8 @@ public String toString() {

}

boolean shouldScheduleAsyncFailTask() {
return TimeValue.ZERO.equals(replicationTimeLimitFailReplica) == false;
}

}
Original file line number Diff line number Diff line change
Expand Up @@ -217,6 +217,8 @@ public void testFailStaleReplicaTask() throws Exception {
assertEquals(5, shardStats.getCheckpointsBehindCount());

// call the background task
assertTrue(service.getFailStaleReplicaTask().mustReschedule());
assertTrue(service.getFailStaleReplicaTask().isScheduled());
service.getFailStaleReplicaTask().runInternal();

// verify that remote shard failed method is called which fails the replica shards falling behind.
Expand Down Expand Up @@ -257,6 +259,41 @@ public void testFailStaleReplicaTaskDisabled() throws Exception {
}
}

public void testFailStaleReplicaTaskToggleOnOff() throws Exception {
final Settings settings = Settings.builder()
.put(IndexMetadata.SETTING_REPLICATION_TYPE, ReplicationType.SEGMENT)
.put(SEGMENT_REPLICATION_INDEXING_PRESSURE_ENABLED.getKey(), true)
.put(MAX_REPLICATION_TIME_BACKPRESSURE_SETTING.getKey(), TimeValue.timeValueMillis(10))
.put(MAX_REPLICATION_LIMIT_STALE_REPLICA_SETTING.getKey(), TimeValue.timeValueMillis(1))
.build();

try (ReplicationGroup shards = createGroup(1, settings, new NRTReplicationEngineFactory())) {
shards.startAll();
final IndexShard primaryShard = shards.getPrimary();
SegmentReplicationPressureService service = buildPressureService(settings, primaryShard);

// index docs in batches without refreshing
indexInBatches(5, shards, primaryShard);

// assert that replica shard is few checkpoints behind primary
Set<SegmentReplicationShardStats> replicationStats = primaryShard.getReplicationStatsForTrackedReplicas();
assertEquals(1, replicationStats.size());
SegmentReplicationShardStats shardStats = replicationStats.stream().findFirst().get();
assertEquals(5, shardStats.getCheckpointsBehindCount());

assertTrue(service.getFailStaleReplicaTask().mustReschedule());
assertTrue(service.getFailStaleReplicaTask().isScheduled());
replicateSegments(primaryShard, shards.getReplicas());

service.setReplicationTimeLimitFailReplica(TimeValue.ZERO);
assertFalse(service.getFailStaleReplicaTask().mustReschedule());
assertFalse(service.getFailStaleReplicaTask().isScheduled());
service.setReplicationTimeLimitFailReplica(TimeValue.timeValueMillis(1));
assertTrue(service.getFailStaleReplicaTask().mustReschedule());
assertTrue(service.getFailStaleReplicaTask().isScheduled());
}
}

private int indexInBatches(int count, ReplicationGroup shards, IndexShard primaryShard) throws Exception {
int totalDocs = 0;
for (int i = 0; i < count; i++) {
Expand Down

0 comments on commit 5190107

Please sign in to comment.