There are 2 different sorts of 'virtual' machines that Fedora uses to build s390x builds. First there are ones which are directly on an LPAR, and another set are ones which are virtual guests on a Linux VM. All of the systems should be F38.

Are the crashes just with one or two builders or across a set of them? [The crash looks like it was on one of the virtual guests. @sharkcz may be helpful on this also?

I had run the build in loop some time ago - and all - aprox 20 builds ahve failed, so maybe answer is both?

We don't have any way to grant you access to those builders. We can collect core dump or other info if you tell us what you need.

There's an ongoing issue about a maintainer test instance, but its not yet available.

cc: @sharkcz

hi! We had this issue in past, and we were allowed to connect to the builder. Tbh, I do not know how else to debug this issue.

Is there anything you can do for us to help with the issue?

Well, I offered to get the core dump and other files it mentions in the error? Is that of use to you, or wouldn't help any?

We also have the save_failed_tree plugin installed (but I have not used it in a while, so not sure if it's working).

That would let you do:

koji save-failed-tree <buildid>

on a failed build and it will tar up the entire build directory and let you download it for review.

Does the crash happen with scratch builds? You could add some debugging to the spec file and do some scratch builds to get more info?

Hi!

Thanx. yes, we already have may cras-debug hooks in build, but they are not producing anything usefull in this case.
yes, it happens also in scratch build.

koji save-failed-tree <buildid> does not seem to be doing anything. Launching new builds - both scratch and normal, jsut to see if issue maybe magically fixed itself. Its rawhide at the nd.

Hiwever - it seems that builds started to pass. Whatevcer dependence it was, it seems it fixed itself. Thanx a lot for looking into it. Clsoing