Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

ReplicationTracker.markAllocationIdAsInSync may hang if allocation is cancelled #30316

Merged
merged 2 commits into from
May 2, 2018
Merged

ReplicationTracker.markAllocationIdAsInSync may hang if allocation is cancelled #30316

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
503e932
fix inSync on cancelled allocation
bleskes May 1, 2018
cef12b1
spaces
bleskes May 2, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 10 additions & 0 deletions server/src/main/java/org/elasticsearch/index/seqno/ReplicationTracker.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -339,6 +339,11 @@ private boolean invariant() {
"shard copy " + entry.getKey() + " is in-sync but not tracked";
}

// all pending in sync shards are tracked
for (String aId : pendingInSync) {
assert checkpoints.get(aId) != null : "aId [" + aId + "] is pending in sync but isn't tracked";
}

return true;
}

Expand Down Expand Up @@ -521,6 +526,9 @@ public synchronized void updateFromMaster(final long applyingClusterStateVersion
checkpoints.put(initializingId, new CheckpointState(localCheckpoint, globalCheckpoint, inSync, inSync));
}
}
if (removedEntries) {
pendingInSync.removeIf(aId -> checkpoints.containsKey(aId) == false);
}
} else {
for (String initializingId : initializingAllocationIds) {
if (shardAllocationId.equals(initializingId) == false) {
Expand Down Expand Up @@ -549,6 +557,8 @@ public synchronized void updateFromMaster(final long applyingClusterStateVersion
replicationGroup = calculateReplicationGroup();
if (primaryMode && removedEntries) {
updateGlobalCheckpointOnPrimary();
// notify any waiter for local checkpoint advancement to recheck that their shard is still being tracked.
notifyAllWaiters();
}
}
assert invariant();
Expand Down
24 changes: 17 additions & 7 deletions server/src/test/java/org/elasticsearch/index/seqno/ReplicationTrackerTests.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -305,7 +305,8 @@ public void testWaitForAllocationIdToBeInSync() throws Exception {
final AllocationId inSyncAllocationId = AllocationId.newInitializing();
final AllocationId trackingAllocationId = AllocationId.newInitializing();
final ReplicationTracker tracker = newTracker(inSyncAllocationId);
tracker.updateFromMaster(randomNonNegativeLong(), Collections.singleton(inSyncAllocationId.getId()),
final long clusterStateVersion = randomNonNegativeLong();
tracker.updateFromMaster(clusterStateVersion, Collections.singleton(inSyncAllocationId.getId()),
routingTable(Collections.singleton(trackingAllocationId), inSyncAllocationId), emptySet());
tracker.activatePrimaryMode(globalCheckpoint);
final Thread thread = new Thread(() -> {
Expand Down Expand Up @@ -336,13 +337,22 @@ public void testWaitForAllocationIdToBeInSync() throws Exception {
assertBusy(() -> assertTrue(tracker.pendingInSync.contains(trackingAllocationId.getId())));
}

tracker.updateLocalCheckpoint(trackingAllocationId.getId(), randomIntBetween(globalCheckpoint, 64));
// synchronize with the waiting thread to mark that it is complete
barrier.await();
assertTrue(complete.get());
assertTrue(tracker.getTrackedLocalCheckpointForShard(trackingAllocationId.getId()).inSync);
if (randomBoolean()) {
// normal path, shard catches up
tracker.updateLocalCheckpoint(trackingAllocationId.getId(), randomIntBetween(globalCheckpoint, 64));
// synchronize with the waiting thread to mark that it is complete
barrier.await();
assertTrue(complete.get());
assertTrue(tracker.getTrackedLocalCheckpointForShard(trackingAllocationId.getId()).inSync);
} else {
// master changes its mind and cancels the allocation
tracker.updateFromMaster(clusterStateVersion + 1, Collections.singleton(inSyncAllocationId.getId()),
routingTable(emptySet(), inSyncAllocationId), emptySet());
barrier.await();
assertTrue(complete.get());
assertNull(tracker.getTrackedLocalCheckpointForShard(trackingAllocationId.getId()));
}
assertFalse(tracker.pendingInSync.contains(trackingAllocationId.getId()));

thread.join();
}

Expand Down