Details
-
Bug
-
Resolution: Fixed
-
Major
-
7.6.0
-
Untriaged
-
0
-
Unknown
Description
At present, we do not authorize new nodes to join the cluster until after we have obtained the rebalance locks. In the event a long DDL/DML is executing, this can cause the joining node(s) to be in a failure loop for an extended period of time. These fast failures also cause the restart delay to max out, delaying the time until the rebalance can ultimately proceed.
Attachments
For Gerrit Dashboard: MB-59792 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
201474,2 | MB-59792: add authorized nodes prior to acquiring rebalance lock; -= ping servlet from admin server | trinity | cbas-core | Status: MERGED | +2 | +1 |
201545,5 | MB-59792: add authorized nodes prior to acquiring rebalance lock; -= ping servlet from admin server | neo | cbas-core | Status: MERGED | +2 | +1 |
201660,2 | MB-59792: resolve nodeIds to hostnames for rebalance failures in driver | neo | cbas | Status: MERGED | +2 | +1 |