Fedora Account System
Red Hat Associate
Red Hat Customer
On s390x, building bup fails on s390x when we get to doing the python tests. When asked in Matrix, Neal Gompa suggested making a ticket and forwarding it to Dan to see if he could figure out why it does fail there, but nowhere else. Reproducible: Always
Created attachment 2161701 [details] Build log
Hi Steve, do you have more info about the failure(s)? Like whether it came with new upstream release (previous were OK), are tests newly run during the build, are all Fedora releases affected, etc. Any details could be helpful.
(In reply to Dan Horák from comment #2) > Hi Steve, > do you have more info about the failure(s)? Like whether it came with new > upstream release (previous were OK), are tests newly run during the build, > are all Fedora releases affected, etc. Any details could be helpful. Do note this is a package that is an unretirement from the early python3 days, so no recent information I'm afraid. As far as the other releases go, unfortunately no because if something doesn't build in rawhide, I can't build it for other releases. But I'm trying a test build in copr right now to see what it looks like in other branches: https://copr.fedorainfracloud.org/coprs/farchord/s390x-test/builds/
OK for the record, there is a public Fedora s390x system for development/debugging listed at https://fedoraproject.org/wiki/Test_Machine_Resources_For_Package_Maintainers if you want to look closer
interesting, I get a successful build (with tests enabled) on F-44
How do you build to those custom workers though?
If you mean the "test systems", then you get a shell access (with sudo), so you can do local debugging, run "fedpkg local" or "fedpkg mockbuild", etc.
There seem to be some randomness in the test results or some dependency on the runtime environment :-(
Now I am getting test failures also on F-44, so it's rather flaky tests than a Python issue ...
Huh...