SEZ9 commented on issue #13246: URL: https://github.com/apache/seatunnel/issues/13246#issuecomment-6844864186
Thanks for the detailed report and the OpenJDK references. The proposed fix (full-name matching of `G1 Refine#` followed by digits via a precompiled pattern, used in both filtering paths, with the wait, classloader and other application-thread checks unchanged) sounds reasonable and low-risk. I'll take a look at #13247. Since the detector stays name based, other JVM-owned GC/runtime workers started between snapshots could cause the same false positive. It would help to note in the PR why the fix is scoped to `G1 Refine#` only, and to include the `SeaTunnelContainerThreadExemptionTest` cases you described. Once hosted CI has run, could you link the failing run(s) showing `G1 Refine#6` / `G1 Refine#5` so reviewers can confirm the symptom? Also, does the failure reproduce on dev without the separate Python image-preparation fix? That would help confirm this is purely a harness classification issue. <!-- streview-comment:2321 --> -- 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]
