NSAmelchev opened a new pull request, #13811: URL: https://github.com/apache/ignite/pull/13811
`GridDhtPartitionsExchangeFuture#onDone` updates the `DurationHistogram` and `CacheOperationsBlockedDurationHistogram` metrics after `super.onDone()`, so a thread that has already observed `isDone()` (the `Duration` metric drops to zero) can still read the previous histogram counts. `GridCachePartitionExchangeManager` publishes a new exchange future as `lastTopologyFuture()` before `init()` sets `initTime`, so `CacheOperationsBlockedDuration` (`currentPMEDuration(true)`) can still be zero right after the switch. `ClusterMetricsSelfTest` read these metrics immediately after the events it waited for and failed sporadically. The test now waits for the expected metric values with `waitForCondition`. Also fixed `ThinClientPartitionAwarenessResourceReleaseTest#testResourcesReleasedAfterCacheDestroyed`. The test takes the client's affinity version before late affinity assignment for the joined node completes, so the cache destroy and the affinity switch together increase the minor version by two, while the test waits for exactly one. The test now calls `awaitPartitionMapExchange()` after the grids start, as other `ThinClientPartitionAwareness*Test` classes do. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
