Haskell Foundation DevOps Weekly Update, 2022-11-09


This has been an eventful week with lots of operation surprises. But I’m really happy to be completely unblocked on the looming hardware migration, and that is my focus right now!


  • Created a replacement Arm runner as part of the ongoing server migration.
  • Fixed one followup issue in the quest to make nightly tests more reliable.
    • (This showed up on last week’s list, but I had to babysit CI this week to actually get it through.)
  • Researched GitLab service errors.
    • Failure to display pipeline info
      • Fixed by bgamari with a service upgrade/restart
    • Disk space outage
      • Fixed by me with a Nix garbage collection
    • Increased rate of failures from CI runners
      • Fixed itself; i.e. the failures were intermittent and should be tracked as such
  • Got access to the last few systems I need in order to finish migrating away from end-of-life hardware!


  • Removing armv7 (32-bit Arm) tests, since they are frequently broken and only supported by end-of-life CI hardware at the moment. I made an issue tracking renewed support for the architecture.
  • Continuing preparation for the hardware migration. In particular, today I am planning the DNS changes that will be required.

Coming up

  • The actual hardware migration! It took a while to remove all the organizational blockers, but that’s done now, so I can just work and get it done.
  • A grand return to the good old days of squashing spurious failures and silencing fragile tests.