Moving infrequently failing tests into the usual exclude files#6782
Conversation
|
I am looking at the issue links under which these exclusions are being done, and we should not be referring to release triage issues such as #6421 |
smlambert
left a comment
There was a problem hiding this comment.
Should be referencing actual issues that articulate the intermittent failure (look in aqa-tests, infrastructure, and JBS existing issues that describe the problem, if an issue does not exist that describes the actual type of failure, then we should create one, some testcases may be failing in the same way, and can be grouped under the same issue)
Fair enough. Will screen them out. |
|
It seems many of the triage issues have expired job links, so I'm going through each of the failing unit tests and identifying the actual problem for each based on more recent test runs. I'll keep an eye out for failures sharing symptoms (with each other and existing issues). Plan: For each unreliable test:
Status JDK8: Complete. JDK11: Complete. - https://ci.adoptium.net/job/Grinder/16033 JDK17: Complete. - https://ci.adoptium.net/view/Test_grinder/job/Grinder/16186/
JDK21: Complete. - https://ci.adoptium.net/job/Grinder/16196/ JDK25: Grinders rerunning. - https://ci.adoptium.net/job/Grinder/16283/
|
This excludes any tests that could interrupt a release, and also helps keep our test runs green at the expense of nonblocker test executions. Plus it reduces complexity by reducing the number of exclude files, which makes maintenance easier. Addendum: All "unreliable" tests excluded in this commit have been identified as not blocking a release (nonblocker) in at least one previous release cycle. Signed-off-by: Adam Farley <adfarley@ibm.com>
Co-authored-by: Shelley Lambert <slambert@gmail.com>
Co-authored-by: Shelley Lambert <slambert@gmail.com>
086c92d to
859ed2e
Compare
Signed-off-by: Adam Farley <adfarley@ibm.com>
as they have not been seen in recent runs. Signed-off-by: Adam Farley <adfarley@ibm.com>
Where "processing" involves finding a recent reproduction or open issue, and attributing the exclusion to that issue (as opposed to a mass-triage issue that contains few details). Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
Note that many could not be reproduced in grinders, nor found to have failed in pipeline test runs. These were removed, and may need to be re-added later if they occur again. Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
Signed-off-by: Adam Farley <adfarley@ibm.com>
|
The intermittent failures for JDK8-JDK21 are complete. JDK25 has about 150 intermittently failing tests for review, so I'd like to work on that in a separate PR. JDK8-21 test runs are listed below. Once complete, I'll post a message in testing-aqavit requesting reviews.
Update: The tests are successfully excluded. |
…ium#6782) * Moving infrequently failing tests into the usual exclude files This excludes any tests that could interrupt a release, and also helps keep our test runs green at the expense of nonblocker test executions. Plus it reduces complexity by reducing the number of exclude files, which makes maintenance easier. Addendum: All "unreliable" tests excluded in this commit have been identified as not blocking a release (nonblocker) in at least one previous release cycle. Signed-off-by: Adam Farley <adfarley@ibm.com> * Update openjdk/excludes/alpine/ProblemList_openjdk8.txt Co-authored-by: Shelley Lambert <slambert@gmail.com> * Update openjdk/excludes/alpine/ProblemList_openjdk8.txt Co-authored-by: Shelley Lambert <slambert@gmail.com> * In progress * Prep for jdk8 pass Signed-off-by: Adam Farley <adfarley@ibm.com> * Removing aix unreliables for jdk8 as they have not been seen in recent runs. Signed-off-by: Adam Farley <adfarley@ibm.com> * Processing test exclusions for jdk8 intermittent tests Where "processing" involves finding a recent reproduction or open issue, and attributing the exclusion to that issue (as opposed to a mass-triage issue that contains few details). Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK8 intermittent tests with full issues linked for each one Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK11 Intermittent failures added to exclude files Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK21 Alpine Intermittent failure added to exclude file Signed-off-by: Adam Farley <adfarley@ibm.com> * Excluding EmptyFolderTest due to Windows Installer Service error Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK17 Intermittent failures merged into excluded tests Note that many could not be reproduced in grinders, nor found to have failed in pipeline test runs. These were removed, and may need to be re-added later if they occur again. Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK21 intermittent excludes and multicast exclude backporting Signed-off-by: Adam Farley <adfarley@ibm.com> * Removing exclude for closed issue Signed-off-by: Adam Farley <adfarley@ibm.com> * Removing exclude as grinder cannot reproduce Signed-off-by: Adam Farley <adfarley@ibm.com> * Adding multicast excludes for JDK25 Signed-off-by: Adam Farley <adfarley@ibm.com> --------- Signed-off-by: Adam Farley <adfarley@ibm.com> Co-authored-by: Shelley Lambert <slambert@gmail.com>
…ium#6782) * Moving infrequently failing tests into the usual exclude files This excludes any tests that could interrupt a release, and also helps keep our test runs green at the expense of nonblocker test executions. Plus it reduces complexity by reducing the number of exclude files, which makes maintenance easier. Addendum: All "unreliable" tests excluded in this commit have been identified as not blocking a release (nonblocker) in at least one previous release cycle. Signed-off-by: Adam Farley <adfarley@ibm.com> * Update openjdk/excludes/alpine/ProblemList_openjdk8.txt Co-authored-by: Shelley Lambert <slambert@gmail.com> * Update openjdk/excludes/alpine/ProblemList_openjdk8.txt Co-authored-by: Shelley Lambert <slambert@gmail.com> * In progress * Prep for jdk8 pass Signed-off-by: Adam Farley <adfarley@ibm.com> * Removing aix unreliables for jdk8 as they have not been seen in recent runs. Signed-off-by: Adam Farley <adfarley@ibm.com> * Processing test exclusions for jdk8 intermittent tests Where "processing" involves finding a recent reproduction or open issue, and attributing the exclusion to that issue (as opposed to a mass-triage issue that contains few details). Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK8 intermittent tests with full issues linked for each one Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK11 Intermittent failures added to exclude files Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK21 Alpine Intermittent failure added to exclude file Signed-off-by: Adam Farley <adfarley@ibm.com> * Excluding EmptyFolderTest due to Windows Installer Service error Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK17 Intermittent failures merged into excluded tests Note that many could not be reproduced in grinders, nor found to have failed in pipeline test runs. These were removed, and may need to be re-added later if they occur again. Signed-off-by: Adam Farley <adfarley@ibm.com> * JDK21 intermittent excludes and multicast exclude backporting Signed-off-by: Adam Farley <adfarley@ibm.com> * Removing exclude for closed issue Signed-off-by: Adam Farley <adfarley@ibm.com> * Removing exclude as grinder cannot reproduce Signed-off-by: Adam Farley <adfarley@ibm.com> * Adding multicast excludes for JDK25 Signed-off-by: Adam Farley <adfarley@ibm.com> --------- Signed-off-by: Adam Farley <adfarley@ibm.com> Co-authored-by: Shelley Lambert <slambert@gmail.com>
This excludes any tests that could interrupt a release, and also helps keep our test runs green at the expense of nonblocker test executions. Plus it reduces complexity by reducing the number of exclude files, which makes maintenance easier.
Note 1: All "unreliable" tests excluded in this commit have been identified as not blocking a release (nonblocker) in at least one previous release cycle.
Note 2: The issues related to these infrequently failing tests will be updated to include a title change and an additional tag to indicate they they are infrequent failure.
@sophia-guo - Do you think this PR can go in in parallel to that work, or do you feel we should hold fire until the related issues are updated?
Progresses (but does not resolve) #6593