Details
Description
Nodes acquire UUIDs when they are first started up and retain them through node addition. This means that if the same base machine image is used to create all the nodes in a cluster, they will all have the same node UUID. This does not always, but can result in config corruption, rebalance hang and REST API and UI breakage; so very serious when things do go wrong.
To mitigate this we should generate a new UUID for a node when it's added to a cluster. It seems likely the vclocks of the per node keys in the config will need to be migrated to the new UUID when this maneuver happens.
Attachments
Issue Links
- blocks
-
MB-16962 3.1.4 Minor Release
- Closed