I'm debugging a rather difficult to diagnose build problem with edk2 in F36 rawhide. The same srpm builds fine in F35 build roots, but fails in F36 build roots with bizarre make errors. make appears to "loose" a bunch of the rules in the makefile, despite them still actually existing when I dump a copy of the makefile contents immediately before running make. It is as if it sees only partial or corrupt content for the makefile when it reads it. Most annoyingly it does NOT reproduce - locally with mock config for fedora-rawhide-x86_64 - locally with mock using 'koji mock-config' for the exact repo ID koji built with - in koji if the build happens to get placed on one of the builders with 48 vCPUs, rather than the normal 6 vCPUs I think I had 1 build where it succeeded in koji even in a 6 vCPU builder, but it fails 99% of the time in the 6 vCPU builders. Based on package differences betweeen F35 and F36 build roots, my money is currently on something changing in gcc or glibc, with a lean towards glic, but I struggle to see an obvious changelog entry that jumps out as a likely trigger. I was hoping to try doing scratch builds with specific koji repo-ids corresponding to older builds, so I could "bisect" where the problem arrived but I've hit more stumbling blocks - Koji f36-build repos older than 1 month appear to have been deleted - I get permission denied when I try on newer repos that still exist Does anyone have any tips on how to test builds in Koji with very specific RPM versions that koji has copies of, but which no longer exist in current build repos. Basically I'd like to try a build in koji with each of the glibc versions in: https://kojipkgs.fedoraproject.org/packages/glibc/2.34.9000/ to see if any one of those was the trigger. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|