breeze test parallel

[Bowrna]

relates to: #23538
closes: #23538

---

^ Add meaningful description above

Read the Pull Request Guidelines for more information.
In case of fundamental code change, Airflow Improvement Proposal (AIP) is needed.
In case of a new dependency, check compliance with the ASF 3rd Party License Policy.
In case of backwards incompatible changes please leave a note in a newsfragement file, named {pr_number}.significant.rst, in newsfragments.

[joppevos]

Nice work!

[potiuk]

I think it needs quite a bit more changes :).

More details in #23538 (comment)

[potiuk]

Hey @Bowrna the #24236 is merged - feel free to rebase and work on "parallelising" :)

[Bowrna]

@potiuk I have rebased the changes in main and currently, I have changes to run the test in parallel and the changes you have added include showing the limited progress.

I need to add the option to show the limited progress in parallel tests alone or both parallel and usual test flow?

[potiuk]

I think the "limited" output should be always enabled when parallell is used

[potiuk]

I just opened a PR to remove the option in Helm tests here: #26144

[eladkal]

@Bowrna there are conflicts to resolve

[potiuk]

I think we can remove limit_progress_output parameter altogether (from both helm_tests and testing). Once we get parallel running, the limit_progess_output should be always enabled when run in parallel and disabled when not, so we do not need another flag there.

[potiuk]

Hey @Bowrna - see https://github.com/apache/airflow/actions/runs/2983790915 :

When you have a bug in the workflow - you will see exclamation mark and detailed explanation :

Invalid workflow file: .github/workflows/ci.yml#L1007
The workflow is not valid. .github/workflows/ci.yml (Line: 1007, Col: 9): 'run' is already defined .github/workflows/ci.yml (Line: 1008, Col: 9): 'env' is already defined

[Bowrna]

@potiuk I am facing issues in CI workflow in breeze parallel tests
https://github.com/apache/airflow/runs/8204814169?check_suite_focus=true

But I am not sure why it's failing and can't figure out much from the error logs. Can you help me how I could understand this part better?

I tried running it on my machine and it failed in different steps with error below

Bind for 0.0.0.0:28080 failed: port is already allocated

[Bowrna]

@potiuk I am facing issues in CI workflow in breeze parallel tests https://github.com/apache/airflow/runs/8204814169?check_suite_focus=true

But I am not sure why it's failing and can't figure out much from the error logs. Can you help me how I could understand this part better?

I tried running it on my machine and it failed in different steps with error below

Bind for 0.0.0.0:28080 failed: port is already allocated

@potiuk when you get time, please give some directions in resolving this issue. Thanks

[potiuk]

@potiuk when you get time, please give some directions in resolving this issue. Thanks

The thing here is that you should override (like it is in the original code) the port number for each "tests" command. Each of the tests run a command which is run here needs a different port number when run in paralllel and in the original code I am dynamically generating the port number so that they are different. This is a bit "arcane" so if you feel like this might be too difficult, I can take over that one?

[Bowrna]

@potiuk when you get time, please give some directions in resolving this issue. Thanks

The thing here is that you should override (like it is in the original code) the port number for each "tests" command. Each of the tests run a command which is run here needs a different port number when run in paralllel and in the original code I am dynamically generating the port number so that they are different. This is a bit "arcane" so if you feel like this might be too difficult, I can take over that one?

Yes @potiuk I think you can take over that part and I will watch how you are doing the dynamic allocation of port number. So that I can take over something like that next time and do it

[potiuk]

I think #26612 will likely get it obsolete :).
I've done much more there - like debugging and tracking what's going on + some optimisations. It's close to merge, just testing last case. But it turned out to be quite a bit more complex eventually

[Bowrna]

@potiuk Oh, I see! I didn't notice the changes you are making in that PR. Let me check it now.

[eladkal]

@Bowrna do we still need this PR?

[potiuk]

Ah no. It's implemented already :)