Details
-
Bug
-
Resolution: Fixed
-
Medium
-
5.6, 5.10
-
Severity 2 - Major
-
Description
Add a configurable 'time to survive after split brain' parameter for clusters. This will be used to pause the cluster safety job if a likely split brain is detected, ie if the cluster safety job has run on this instance before (meaning it's not the first run after startup), but the cluster safety number cannot be found in the cache.
In this scenario, the cluster safety job will wait for the configured time-to-survive, to allow the nodes to rejoin. After this time has passed, if the node still cannot find the cluster safety number in the cache, the node will panic.
Attachments
Issue Links
- supersedes
-
CONFSERVER-40685 Whole cluster can panic if one node is doing extended GC
- Closed
-
CONFSERVER-41644 Data Centre - don't trigger cluster panic when node is merging
- Closed
- mentioned in
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...
-
Page Loading...