action #169351
closed
[BCI] Switch PPC test runs to CC-compliant ppc workers
Added by ph03nix about 2 months ago.
Updated 23 days ago.
Description
Due to new Common Criteria datacenter compliance rules (#168295 and #165282) some PPC workers will no longer work for some time. We need to check, if we can use qemu_ppc64le or emulated ppc64le workers instead for the BCI test runs.
See https://progress.opensuse.org/issues/169387#note-4
See e.g. https://openqa.suse.de/tests/15856207 for an emulated ppc test run.
We need to see if those emulated workers perform as expected. Especially if we run into some timeout issues or if the overall job time needs to be adjusted.
Acceptance criteria¶
- AC1: BCI tests will use PPC workers within the CC zone for BCI tests
- Status changed from Workable to In Progress
- Assignee set to mgrossu
If this works fine, there might be some timeouts to be changed, cause ppc emulated on intel is very slow.
I also don't expect the same stability as current real powerKVM jobs, therefore we can reduce the number of jobs that are run and keep it to the bare minimum (e.g. running ppc jobs only for latest SP in the test matrix).
- Subject changed from [BCI] Switch PPC test runs to emulated PPC to [BCI] Switch PPC test runs to CC-compliant ppc workers
- Description updated (diff)
HMC means powerVM, so, treat it as as bare-metal. It is not possible to boot qcow2 on HMC. If we want to go this path, then we need to have system installation first, which could be overkill.
My take on this for now is:
- try with
ppc64le-emu, which is not ideal, but is the only workaround with the least work required.
- reduce the number of jobs to the bare-minimum. Let's keep ppc coverage on the latest SP only.
Once we have the mitigations in place (wireguard tunnel in NUE2 machines), we might be able to restore all the powerKVM tests we have today.
For example, the jobs that we have to create the HDDs: https://openqa.suse.de/tests/15842540 publish the following asset: SLES-15-SP6-ppc64le-containers.qcow2..
So, you can't just take this and boot it on hmc backend.
Note: It was agreed with FX to release the current round of containers due to release pressure, if only ppc is failing. This is a one-time cause because there are some pressing CVEs to be released. We still need to fix the test runs.
Note for myself: https://suse.slack.com/archives/C07D6ACD39P/p1731512814200339
- Related to action #169885: [BCI] Fix several issues regarding working on emulated ppc workers added
- Priority changed from High to Normal
- Status changed from In Progress to Feedback
- Status changed from Feedback to Resolved
Also available in: Atom
PDF