Uploaded image for project: 'Performance and Scale for AI Platforms'
  1. Performance and Scale for AI Platforms
  2. PSAP-504

Create release-informing CI job to verify driver-toolkit contents

XMLWordPrintable

    • PSAP Sprint 208, PSAP Sprint 209, PSAP Sprint 210, PSAP Sprint 211, PSAP Sprint 213, PSAP Sprint 214, PSAP Sprint 215, PSAP Sprint 216, PSAP Sprint 217
    • undefined

      As more customers are depending on the Driver Toolkit, it will become urgent that we never ship a DTK with the wrong packages installed. We can create a periodic job that checks this every ~12 hours, and this can be a release "informing" job. Later on we may want to make it a release gating job.

      The test should be rather simple:

      • pull the DTK for this release/nightly build and save the contents of /etc/driver-toolkit-release.json.
      • run a debug shell (or a pod?) on a worker node, and check the kernel version with uname or /proc/sys/kernel/osrelease.
      • in a debug shell also check RHEL version by inspecting /etc/os-release
      • If the kernel / rhel version matches the driver-toolkit, we pass.

      A discussion on how to set up a release informing job is in this slack thread:
      https://coreos.slack.com/archives/CBN38N3MW/p1618579015229200

      Potential extensions of this testing include:

      • Verify that the driver-toolkit also contains the right version of the real time kernel packages
      • Build the simple-kmod on top of the driver-toolkit
      • Test that the imagestream is correct after a cluster upgrade

              dagray@redhat.com David Gray
              dagray@redhat.com David Gray
              Votes:
              0 Vote for this issue
              Watchers:
              4 Start watching this issue

                Created:
                Updated:
                Resolved: