-
Enhancement
-
Resolution: Duplicate
-
Major
-
None
-
11.0.3.Final, 12.0.0.Final
-
None
-
Documentation (Ref Guide, User Guide, etc.), User Experience
After a cluster is brought down with "shutdown cluster" command the restart need to have all the known nodes back to prevent from any data lost.
If there is an issue bringing the nodes back it should be possible to trigger the state-transfer and bring the cluster back online.
It should have two options
- keep the data
- clear the cache and start empty
This would cause a loss of data if not all segments are available for a cache.
In this case a WARN message should be logged to notify the user that data is lost.
It will be possible to reset the state by removing the data/*.state files, but in this case another restart is needed and more data is lost as the first node will use the (incomplete) store and other local stores are not used to recover.
Reset the state when having all remaining nodes up will keep more data as some segments are still complete or have at least the primary or one backup owner.
Best case if if less nodes than numOwner are missed, in that case the data is completely available and con be sucessfully rebalanced.
- duplicates
-
ISPN-14418 CLI/REST command to recover an incomplete cluster if it can not recover after 'shutdown cluster'
- Resolved
- incorporates
-
ISPN-12224 Cluster in a confusing state after restarted from graceful shutdown - no hint for waiting on complete restarted
- Resolved
- is duplicated by
-
ISPN-6967 Add a possibility to manually start the caches after a graceful shutdown if nodes are missing
- Closed
- relates to
-
JDG-3967 Cluster in a confusing state after restarted from graceful shutdown - no hint for waiting on complete restarted
- Verified