On Thu, 2026-10-08 at 11:01 -0700, Kevin Fenzi via devel-announce wrote: > Greetings. > > As you know there have been issues around slowness and lack of > resources for s390x builds. > See https://forge.fedoraproject.org/infra/tickets/issues/13141 and > https://forge.fedoraproject.org/fesco/tickets/issues/3688 > > We have recently been working with IBM to try and solve those issues, > and I wanted > to share actions already taken and plans moving forward. > > Shorter term IBM has gathered a bunch of information about our setup > and > made some adjustments to our LPAR as well as allowing Infrastructure > to make > adjustments based on their reccommendations. This includes: > > * Our LPAR has been allocated more memory. This has allowed us to > give > more memory to builders as well as allowing us to finally move to > newer > kernels on the hypervisor ( > https://bugzilla.redhat.com/show_bug.cgi?id=2492031 ) > * The LPAR has had it's CPU weight adjusted to increase processor > time > * Some builders have had cpus added to them, allocating more from our > existing > CPU pool. > * We have setup to gather more data for IBM to review. After some > more heavy > build times this will hopefully inform them on any additional steps > we can > take with the existing LPAR/resources. > > Medium term, we are discussing the idea of more resources in another > mainframe > in another location. This discussion is ongoing and we will see if > something > can be found. > > Longer term of course we are working to improve the existing env and > increase capacity. > > So, progress is being made and hopefully there is some short term > improvements > as well as possible medium term ones on the horizion. > Looks like it does have the expected result! I crunched some numbers and did a TL;DR here https://forge.fedoraproject.org/fesco/tickets/issues/3688#issuecomment-1367964 (there is a link to a PDF report, or here to save you a redirection: https://github.com/slopfest/sandogasa/releases/download/koji-lag-store-2026-10-19/FINDINGS.pdf ) One thing that surfaced after I crunched all the data: I could have used hw_info.log to get hardware data for the builders used for specific jobs, but I'm about to be AFK for a few days and ran out of time to redo the analysis, so expect the updated data later. But that means some of the open questions listed in the findings can be answered in the next pass. Thanks nirik for all the work here! Best regards, -- _o) Michel Lind _( ) https://keyoxide.org/5dce2e7e9c3b1cffd335c1d78b229d2f7ccc04f2
-- _______________________________________________ devel-announce mailing list -- devel-announce@lists.fedoraproject.org To unsubscribe send an email to devel-announce-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel-announce@lists.fedoraproject.org Do not reply to spam, report it: https://forge.fedoraproject.org/infra/tickets/issues/new
No comments:
Post a Comment