-
Type:
Bug
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
-
Replication
-
ALL
-
(copied to CRM)
As seen in https://jira.mongodb.org/browse/HELP-62301, the _flushRoutingTableCacheUpdatesWithWriteConcern ({w:majority, wtimeout: 0}) command fails to block indefinitely (which is expected if the wtimeout: 0 as per the docs) on a shard primary in case the shard is undergoing network issues and is in a state where the shard's replica set has no primary.
An error status of WriteConcernFailed error is returned which is not retriable (by design). If a retriable (such as HostNotFound or HostUnreachable) error had been returned, the _flushRoutingTableCacheUpdatesWithWriteConcern would have been retried until it was successful (once the shard's replica set was healthy) and the resharding operation would not have failed.
(Note: this was investigated on v6.0)