Koji scratch-builds for the pam package fail inconsistently with the same changes included. taskID=76047688 for example reports a failure on aarch64, while other architectures build correctly. The report for this architecture doesn't give a hint at what's failing so I decided to trigger a new scratch-build for it. The result is available at taskID=76049589 and it succeeds. Thus, I decide to open a PR for this package, which in turn reports the failure of the scratch-build for x86_64 architecture taskID=76050708.
It might be the package itself rather than koji, but it needs some checking.
Metadata Update from @mohanboddu: - Issue tagged with: low-gain, low-trouble, ops
Metadata Update from @mohanboddu: - Issue priority set to: Waiting on Assignee (was: Needs Review)
In commit b5205aee the following change was made:
-make -C po update-gmo -make -# we do not use _smp_mflags because the build of sources in yacc/flex fails +%make_build -C po update-gmo +%make_build
I suspect very very very very strongly that this is a case of the makefiles here not being smp aware.
If you change this back to make does it work everytime? Does pam upstream say they support parallel builds?
If you still see failures after reverting that, please do reopen, but I think thats the problem.
Metadata Update from @kevin: - Issue close_status updated to: Fixed - Issue status updated to: Closed (was: Open)
Yes, you are right. Thanks for the info.
Log in to comment on this ticket.