QA (public) » openQA Project (public)

kazhua:~/devel/openQA # openqa-clone-job  --skip-deps --skip-download --host https://openqa.suse.de --clone-children https://openqa.suse.de/tests/5444256
Created job #5449971: sle-15-SP3-Online-x86_64-Build145.1-wireguard_client@64bit -> https://openqa.suse.de/t5449971
kazhua:~/devel/openQA # openqa-clone-job  --skip-deps --skip-download --host https://openqa.suse.de --clone-children https://openqa.suse.de/tests/5443990
Created job #5449972: sle-15-SP3-Online-x86_64-Build145.1-wireguard_server@64bit -> https://openqa.suse.de/t5449972

Doesn't look like it actually works. As I wasn't sure which one was the parent job, I cloned from both.
But none of the calls spawned two new jobs.

none, is the parent. The real parent is the HDD creation job: https://openqa.suse.de/tests/5443775

Only that one provides the complete cluster. Of course you don't want to run 40 jobs, you just want the "wireguard server+client". But openQA does not know what part of a cluster you want. That's making it hard. But of course it should be feasible to detect the siblings. Let me try to bring others into the discussion.

That's what I feared.

Of course I only want the two multimachine jobs and neither regenerate the hdd image nor run the other 40 children of that parent job.

Also it would be nice if it would be implemented in a way that allows downloading the hdd image to another openQA instance and running the multimachine jobs there without the need of having to re-run the parent job on the target instance again.

Doesn't look like it actually works. But none of the calls spawned two new jobs.

Considering you've specified --skip-deps that shouldn't be a surprise, though.

As I wasn't sure which one was the parent job, I cloned from both.

see #81859#note-7

I've just tested the following:

script/openqa-clone-job --dir /hdd/openqa-devel/openqa/share/factory --show-progress --apikey … --apisecret … --host http://localhost:9526 --clone-children https://openqa.suse.de/tests/5443990

It worked fine, both parallel jobs have been cloned. Of course the create HDD job has been cloned as well. This can be avoided via --skip-chained-deps, e.g.:

script/openqa-clone-job --dir /hdd/openqa-devel/openqa/share/factory --show-progress --apikey … --apisecret … --host http://localhost:9526 --skip-chained-deps --clone-children https://openqa.suse.de/tests/5443990

Also it would be nice if it would be implemented in a way that allows downloading the hdd image to another openQA instance

To my knowledge that's something that's really not implemented so far. Of course you can simply run openqa-clone-job on the target web UI. Obviously that would require some daemon on the target web UI to do the download. Not sure whether it is worth implementing that.

mkittler wrote:

[...]
It worked fine, both parallel jobs have been cloned. Of course the create HDD job has been cloned as well. This can be avoided via --skip-chained-deps

Do you think something like

openqa-clone-job --skip-chained-deps --clone-children https://openqa.example.org/tests/1234

can be more prominently featured in the help text of openqa-clone-job or the docs? Can you do that?

That's great.

@dheidler can you confirm this helps? Do you have something else in mind that could help?

Looks good.