action #137771
closedopenQA Project - coordination #112862: [saga][epic] Future ideas for easy multi-machine handling: MM-tests as first-class citizens
openQA Project - coordination #111929: [epic] Stable multi-machine tests covering multiple physical workers
Configure o3 ppc64le multi-machine worker size:M
0%
Description
Motivation¶
There are scenarios scheduled like https://openqa.opensuse.org/tests/latest?arch=ppc64le&distri=opensuse&flavor=DVD&machine=ppc64le&test=nfs_server&version=Tumbleweed which do not find a matching worker. Right now we only have kerosene.qe.nue2.suse.org which apparently is not configured for multi-machine, e.g. has no tap devices and hence also no "tap" in /etc/openqa/workers.ini
Acceptance criteria¶
- AC1: o3 ppc64le multi-machine tests are again successfully executed
Suggestions¶
- Run https://github.com/os-autoinst/os-autoinst/blob/master/script/os-autoinst-setup-multi-machine on kerosene and see what's missing
- Manually trigger multi-machine tests matching that worker and see if they can work
- After verification add "tap" in /etc/openqa/workers.ini on kerosene.qe.nue2.suse.org
Updated by okurz 5 months ago
- Related to action #150869: Ensure multi-machine tests work on aarch64-o3 (or another but single machine only) size:M added
Updated by mkittler 3 months ago
Since there are no production jobs to disrupt I just enabled the tap worker class (after running the setup script and rebooting). This should give existing jobs like https://openqa.opensuse.org/tests/3878822 (https://openqa.opensuse.org/tests/latest?arch=ppc64le&distri=opensuse&flavor=DVD&machine=ppc64le&test=nfs_client&version=Tumbleweed) a chance to run. (These jobs are so far just stayed scheduled or perhaps were cancelled at some point.)
Updated by mkittler 3 months ago · Edited
The tests were failing: https://openqa.opensuse.org/tests/3878798#dependencies
So I removed the tap worker class again and will try a different scenario.
EDIT: I created wicked_basic_ref/…_sut jobs for PowerPC via openqa-clone-job --skip-chained-deps --parental-inheritance --within-instance https://openqa.opensuse.org/tests/3883054 _GROUP=0 BUILD=20240121-tap-test TEST+=-tap-test WORKER_CLASS=qemu_ppc64le,tap_poo137771 ARCH=ppc64le HDD_1=opensuse-Tumbleweed-ppc64le-20240121-textmode@ppc64le.qcow2 ASSET_1= ASSET_2= ASSET_LIBVIRT= ASSET_VIRTUALBOX= IS0=openSUSE-Tumbleweed-DVD-ppc64le-Snapshot20240121-Media.iso MIRROR_HTTP=http://openqa.opensuse.org/assets/repo/openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121 MIRROR_HTTPS=https://openqa.opensuse.org/assets/repo/openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121 MIRROR_PREFIX=http://openqa.opensuse.org/assets/repo REPO_0=openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121 REPO_1=openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121 REPO_2=openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121-debuginfo REPO_3=openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121-source REPO_OSS=openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121 REPO_OSS_DEBUGINFO=openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121-debuginfo REPO_OSS_SOURCE=openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121-source SUSEMIRROR=http://openqa.opensuse.org/assets/repo/openSUSE-Tumbleweed-oss-ppc64_ppc64le-Snapshot20240121 OFW=1 QEMUCPUS= (based on https://openqa.opensuse.org/tests/3883054 and https://openqa.opensuse.org/tests/3882774). Let's see whether that way of creating a PowerPC job from an x86_64 one works at all and that I've caught all arch-specific variables.
If not I could just add
- wicked_basic_sut
- wicked_basic_ref
on https://openqa.opensuse.org/admin/job_templates/4 and wait for the next build.
Updated by mkittler 3 months ago · Edited
- Status changed from In Progress to Feedback
Looks like the basic wicket test scenario actually worked: https://openqa.opensuse.org/tests/3884317
So I suppose the tap setup generally works and I added the tap worker class back. (The other scenario likely just failed due to a discrepancy between needles and what was actually happening when booting, see https://openqa.opensuse.org/tests/3878798.)
Updated by mkittler 3 months ago · Edited
Considering the scenario worked in my testing I added
- wicked_basic_sut
- wicked_basic_ref
on https://openqa.opensuse.org/admin/job_templates/4. I'll wait for the next Tumbleweed build and check whether it works.
EDIT: One actually has to create a PR for this: https://github.com/os-autoinst/opensuse-jobgroups/pull/420
Updated by okurz 3 months ago
- Status changed from Resolved to In Progress
But the tests mentioned in the description that are currently scheduled should be able to run: https://openqa.opensuse.org/tests/latest?arch=ppc64le&distri=opensuse&flavor=DVD&machine=ppc64le&test=nfs_server&version=Tumbleweed
Updated by mkittler 3 months ago
- Status changed from In Progress to Resolved
It looks like too many PowerPC workers are currently busy and there are only 2 free slots. That's just not enough for this 3-machine scenario. The worker class is shown correctly in the worker's table, though. So I don't think there's anything left to do here.
Note that the basic wicket tests already ran (successfully, see https://openqa.opensuse.org/tests/3886504#dependencies) because my manual edits to the job group YAML are actually still in place. That shows that tap-based tests are generally picked up and work.