-
Type: Improvement
-
Resolution: Done
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: Replication
-
None
Periodically, we should rerun the logic in ReplSetImpl::getMemberToSyncTo(). The ping times may have changed since last check, or the oplog staleness within the cluster may have changed such that a different node is now much more up-to-date.
Currently we only do this check when the cluster state changes.
- depends on
-
SERVER-6476 Setting slaveDelay on a sync target should cause syncing members to find a new target.
- Closed
- is depended on by
-
SERVER-4927 Slaves stops replog sync if another slaves used fsyncLock
- Closed
-
SERVER-6137 When disk is full on any secondary, the whole set of secondaries freezes until the full node is stopped
- Closed
- is duplicated by
-
SERVER-6882 Automatically reevaluate sync sources periodically
- Closed
- is related to
-
SERVER-4750 Secondary syncs to another secondary that has slaveDelay > 0
- Closed
- related to
-
SERVER-8235 too-frequent sync source changes causes node to fall behind
- Closed