[
https://issues.apache.org/jira/browse/IGNITE-29271?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Nikita Amelchev updated IGNITE-29271:
-------------------------------------
Fix Version/s: 2.20
> Fix flaky ClusterMetricsSelfTest
> --------------------------------
>
> Key: IGNITE-29271
> URL: https://issues.apache.org/jira/browse/IGNITE-29271
> Project: Ignite
> Issue Type: Bug
> Reporter: Nikita Amelchev
> Assignee: Nikita Amelchev
> Priority: Minor
> Fix For: 2.20
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> {{GridDhtPartitionsExchangeFuture#onDone}} updates the {{DurationHistogram}}
> and {{CacheOperationsBlockedDurationHistogram}} metrics only after
> {{super.onDone()}}, so the test, which waits for the {{Duration}} metric to
> drop to zero (that is, for {{isDone()}}) and then reads the histograms,
> sporadically sees the previous counts. The {{CacheOperationsBlockedDuration}}
> check has the same race: {{GridCachePartitionExchangeManager}} publishes a
> new exchange future as {{lastTopologyFuture()}} before {{init()}} sets
> {{initTime}}, and until then {{currentPMEDuration(true)}} returns zero.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)
