ss77892 opened a new pull request, #11908: URL: https://github.com/apache/ozone/pull/11908
## What changes were proposed in this pull request? `TestQueryNode#testStaleNodesCount` stopped two datanodes fire-and-forget on a single-thread executor and immediately started a 10 s wait for both to become STALE or DEAD. That timeout therefore had to absorb both datanode stop times plus the 3 s stale interval. When stopping the datanodes took longer than about 7 s on a loaded CI host, the wait timed out. The background execution served no purpose, since nothing else in the test runs concurrently with the shutdown. It also leaked the executor (never shut down) and swallowed any exception thrown by `shutdownHddsDatanode`, which would surface only as a confusing timeout. This change: - Stops the two datanodes synchronously, so the waits start only after the datanodes are down and measure only SCM's stale/dead detection. This matches how every other `shutdownHddsDatanode` caller in `integration-test` sequences it. If the stop takes longer than the 6 s dead interval, the first wait still passes because it accepts STALE + DEAD. - Raises the DEAD wait from 4 s to 10 s. With the 3 s stale and 6 s dead intervals, DEAD follows STALE by about 3 s, which left only about 1 s of margin. `waitFor` returns as soon as the condition holds, so the passing path is not slower. ## What is the link to the Apache JIRA https://issues.apache.org/jira/browse/HDDS-17651 ## How was this patch tested? 50 repeating runs of the test. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
