On Mon, Dec 23, 2024 at 5:38 PM Santiago Vila <sanv...@debian.org> wrote: > This build failure happens randomly, with different probability depending on > the build machine. Sure, there were some failures of NNG on s390x for some time. But then as upstream noted (as I remember) v1.9.0 fixed flaky issues. This is seconded with the successful build of it. Later with the Mbed-TLS transition it failed four times in a row on the same test. I do not see in your statistics how many builds were tried, ie the failing percent and not its plain number. But then yesterday I could build it in my s390x environment and it was built for the first time on the buildd as well. Thus checking the possibilities of a flaky test seems lower: first it fails four times in a row, then in a later time it builds twice out of two. Let's say I will try more build tests.
> I would hope that we still don't want packages which FTBFS randomly > in the buildds. A single successful build does not mean that the > randomness has disappeared. Please check above, four failures in a row in the past, present is two successes out of two (100% failure vs 0% failure in a later time) might indicate another issue. > If you need a VM to reproduce, please contact me privately and I will > gladly provide one. If it's an s390x VM please get it by asking for it. Please send the instructions to me in private. Thanks for the fruit at DebConf. > Alternatively, you can take a look at the build logs I've collected > in the last weeks: I see all of these are on amd64 architecture. Do you have such failure rate information for s390x architecture? > Note: I'm just reopening the bug because I believe it's not fixed. > The Release Managers have always the final say about the RC status. OK, I will disable the mentioned tests in the next few days. Regards, Laszlo/GCS
