renaissance-gauss-mix_0 hangs for 24h and times out

[adamfarley]

Summary

renaissance-gauss-mix_0 (an extended.perf renaissance test) has been hanging for 24 hours without output before timing out.

Details

JDK Versions: 17, 21, 25
Platforms: AIX, aarch64_mac, aarch64_linux.
Frequency: Approximately 50/50 (All three platforms have run again since the original failure, and passed)
Notes: There doesn't appear to be a difference between what we run (renaissance-mit-0.16.0.jar) or how we run it (command lines are identical between passes and failures), and the printenv seems near-identical between passed and failed runs (minus time stamps and job IDs).

Links

• https://ci.adoptium.net/job/Test_openjdk21_hs_extended.perf_ppc64_aix/105
• https://ci.adoptium.net/job/Test_openjdk25_hs_extended.perf_aarch64_mac/23
• https://ci.adoptium.net/job/Test_openjdk17_hs_extended.perf_aarch64_linux/240

Also, renaissance is aware of a hang that occurs in gauss, though it is infrequent and the issue is over a year old:
renaissance-benchmarks/renaissance#457

Notes

As this test seems to run for 3-12 minutes (when it passes), maybe we could set the test target timeout for 1 hour?

This doesn't solve the problem, but could limit the impact.