Repository navigation
AIX host build failing around 90% of the time #2566
Description
test-osuosl-aix72-ppc64_be-2 is failing a lot. A quick look suggests build failures, not test failures.
https://ci.nodejs.org/computer/test-osuosl-aix72-ppc64_be-2/builds
Only two of the last 20 builds have passed. (One is in progress.)
Activity
Trott commentedon Mar 11, 2021
@nodejs/platform-aix
sxa commentedon Mar 11, 2021
There's a big mix of failures too so it's not just one issue :-(
Of the once since the last success, 35566 and 35572 seem to be after it was running through the tests
AshCripps commentedon Mar 12, 2021
Ive taken the host offline to see if I can try and investigate its failures - it seems to just crash part way through the build
Trott commentedon Mar 12, 2021
It looks like it's still online? (Or maybe you put it back online?)
https://ci.nodejs.org/computer/test-osuosl-aix72-ppc64_be-2/builds
AshCripps commentedon Mar 12, 2021
Ive stripped it off the label for now and am running test builds, I managed to get a build working locally but even jenkins build crashed on the V8 step (hangs for over 10 mins then jenkins kills the build) I currently have a local build running to see if it gets past this step (been stuck for over a half hour now) if not Ill keep the label removed and get support to have a look at the machine.
AshCripps commentedon Mar 16, 2021
machine "fixed" and added back to the pool. Turns out it was a parallelisation issue and running a build on that machine with anything bigger than
JOBS=3caused it to hang on a V8 instruction. Ive set this machine to 3 and removed the hardcoded value from the job which means the other machines will use the value of 6 added in their startup scripts. I will open a PR to explore movingJOBSto a machine specific variable for AIX