LargeGatheringWrite test runs out of memory on test-azure-win2012r2-x64-3

[adamfarley]

extended.openjdk job ran on test-azure-win2012r2-x64-3 and didn't find enough memory to launch the LargeGatheringWrite unit test.

Link: https://ci.adoptium.net/job/Test_openjdk21_hs_extended.openjdk_x86-64_windows_testList_1/8/

Error message:

09:02:37  ACTION: main -- Failed. Execution failed: `main' threw exception: java.lang.OutOfMemoryError: Unable to allocate 536870912 bytes
09:02:37  REASON: User specified action: run main/othervm/timeout=480 -Xmx4G LargeGatheringWrite 

My first theory is that something else on that machine is using up all the memory during the test, as test-azure-win2012r2-x64-3 should have (according to MACHINE INFO) over 3g of available physical memory (4.3g total).

Next action: Will open an infra issue to start, to see what else is running on that machine.

Links: This issue might be related, as it identifies a number of machines that had low memory during mid-2022, but this is a separate issue for the time being as that issue doesn't mention LargeGatheringWrite.

[adamfarley]

Infra issue raised.

[smlambert]

I would rather not raise 2 issues (test and infra) for one failure. As per the test triage guidance, triage determines which repo an issue should initially be raised in, and that singular issue could then be transferred to a different repo if needed (when more info is gathered if the source of the problem would need to be fixed elsewhere).

[adamfarley]

I would rather not raise 2 issues (test and infra) for one failure. As per the test triage guidance, triage determines which repo an issue should initially be raised in, and that singular issue could then be transferred to a different repo if needed (when more info is gathered if the source of the problem would need to be fixed elsewhere).

I didn't actually know you could move issues between repositories last that. Something new every day. :)

In any case, it's clear I need to refresh myself on the triage guidance, so I'll go have a read now. Thanks for the reminder 👍

[adamfarley]

As for this issue, I'll close it and copy the information into the infra issue.