[Bug] TableChangeWatcher will fire CreatePartitionEvent
after restart although the partition has been created
#333
Labels
Milestone
Search before asking
Fluss version
main
Minimal reproduce step
Create a partition table, and restart coordinator server
What doesn't meet your expectations?
It'll fire
CreatePartitionEvent
andCoordinatorEventProcessor
will process it again which cause the state mess.Anything else?
We use
TablePathChangeListener
which implementsCuratorCacheListener
to listen zk node changes, but found during restarting, it'll load all nodes into cache which will fire methodevent(Type type, ChildData oldData, ChildData newData)
. We haven't figured it out and mistak it as a new node is created which cause theCreatePartitionEvent
fired for an already existing zk node..I'd like to suggest to rewrite method
void initialized()
to know the cache is initialized and only process the events after initialized..Are you willing to submit a PR?
The text was updated successfully, but these errors were encountered: