Re: s390x builder outage

* Neal Gompa:

> So Michel did some excellent work to quantify exactly *how* poor the
> performance is for s390x relative to other architectures. As of last
> month, s390x is orders of magnitude worse than the next bad one
> (ppc64le): https://forge.fedoraproject.org/packaging/koji-lag-metrics/src/branch/main/reports/monthly/2026/07/report.txt

Don't the numbers show that most builds are blocked on ppc64le?

From the bottom of the page:

- `bottleneck` — builds where this arch finished last (the build was bottlenecked on it).
- `med-delay` / `tot-delay` — how long after the second-slowest arch the
  bottleneck arch finished; the extra wall-clock time it alone cost those
  builds (median per build / summed over the window).

The s390x-induced delays are higher, but the total number of
bottlenecked builds is much smaller.  Only 8.4% of the builds are
blocked on s390x.

Do we know if the delayed builds are random, or specific to certain
builders or packages?

We can speed up GCC builds for most packages if we run code generation
only once.  That should be a relatively easy win, but perhaps too small
to matter.

Thanks,
Florian

-- 
_______________________________________________
devel mailing list -- [email protected]
To unsubscribe send an email to [email protected]
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/[email protected]
Do not reply to spam, report it: https://forge.fedoraproject.org/infra/tickets/issues/new