-
Type: Bug
-
Resolution: Done
-
Priority: Major - P3
-
None
-
Affects Version/s: 2.6.1
-
Component/s: Sharding
-
ALL
-
-
Sharding 2019-02-25
-
(copied to CRM)
Shard dies with a stack trace pointing to DummyRunner class destructor.
During chunk migration, if running repairDatabase and compact, it will cause the old collection pointer to be invalid. You will get a different pointer when you call the getCollection method. The current workaround for this is to stop the balancer whenever user runs these commands.
- depends on
-
SERVER-15647 Extent Manager must be Thread-safe
- Closed
-
SERVER-15633 ShardingState::doRefreshMetadata should take a more granular lock than exclusive db
- Closed