Marc Handalian

@Poojita-Raj So the failed tests in SegmentReplicationIndexShardTests are because we are using LuceneTestCase's `newFSDirectory`, which on close runs a [checkIndex](https://github.com/apache/lucene/blob/main/lucene/test-framework/src/java/org/apache/lucene/tests/store/MockDirectoryWrapper.java#L909) that will run various tests to identify corruption. We aren't...

I think we can actually look at disabling fsyncs entirely on replicas if we are triggering a commit/fsync when the shard closes. That would make this implementation much simpler by...

@Rishikesh1159 I don't think you need to change the visibility of the replication tracker. `SegmentReplicationTargetService` invokes `IndexShard#shouldProcessCheckpoint`, you can add a check there?

@tan31989 Thanks for your interest in this issue. > Hi, > > I am giving my thoughts on this issue. Ideally, feature flags should be allowed to toggle, but with...

I think 2 is the best option given we want this as a best effort. I also wouldn't be worried about it delaying shard promotion right now, we can set...

> I think the issue is that when we stop the replication for CCR, we close and reopen the index to reload the Engine. The index close is being handled...

I've added a commit to this ensuring cancelling primary allocation succeeds and that the replica is promoted & primary recreated as a replica. In testing that I found we were...

@dreamer-89 Yeah this should not be failing the replica, it would catch up to the new cp after the current replication event completes. I think this is happening bc we...

Have opened #4182 to cover moving this to allocationID over node. I have not been able to repro after applying this change but I think we should leave this open...

closing this one bc I haven't seen it since, please reopen if needed.