Alexey Goncharuk commented on IGNITE-8130:

Actually, destroy cache completes when local node finishes exchange. However, 
when we call getOrCreateCache on another node, the node may not have completed 
it's exchange yet, so getOrCreateCache is executed as if the cache is not 
destroyed yet, and results in a no-op.

Fixing this for now in the test.

> WalModeChangeCoordinatorNotAffinityNodeSelfTest#testLocalCache fails 
> sporadically in master
> -------------------------------------------------------------------------------------------
>                 Key: IGNITE-8130
>                 URL: https://issues.apache.org/jira/browse/IGNITE-8130
>             Project: Ignite
>          Issue Type: Test
>            Reporter: Alexey Goncharuk
>            Assignee: Alexey Goncharuk
>            Priority: Major
>              Labels: MakeTeamcityGreenAgain
> The reason for random failures is a loop in finally block of forAllNodes() 
> which destroys caches. This leads to a race when cache destroy request is 
> asynchronously sent to other nodes and the cache created in test is destroyed 
> by the finally block.

This message was sent by Atlassian JIRA

Reply via email to