Bump executorlib

[liamhuber]

No description provided.

[github-actions]

👈 Launch a binder notebook on branch pyiron/pyiron_workflow/bump-executorlib

[liamhuber]

Coverage tests hang. The same thing happens on my local machine, where it's easy to see that it hangs at integration.test_wrapped_executorlib.TestWrappedExecutorlib.test_cache. I don't currently know why.

[jan-janssen]

Coverage tests hang. The same thing happens on my local machine, where it's easy to see that it hangs at integration.test_wrapped_executorlib.TestWrappedExecutorlib.test_cache. I don't currently know why.

Do you see which of the two executors causes the hanging? NodeSingleExecutor or _CacheTestClusterExecutor ?

[jan-janssen]

One thing that might be related is that in the pydantic validation of the resource_dict, the cache_directory variable was not available before, basically because I thought it would be sufficient to only specify the cache_key. But I remember we talked about the option to add a cache_directory in the resource_dict as well, so this is added again in pyiron/executorlib#945 .

[liamhuber]

Do you see which of the two executors causes the hanging? NodeSingleExecutor or _CacheTestClusterExecutor ?

Unfortunately, no. Since it doesn't actually fail all I see is the parent test name. I will try splitting the subtest up into two separate tests and hopefully that will reveal. Answer follows at end.

One thing that might be related is that in the pydantic validation of the resource_dict, the cache_directory variable was not available before, basically because I thought it would be sufficient to only specify the cache_key. But I remember we talked about the option to add a cache_directory in the resource_dict as well, so this is added again in pyiron/executorlib#945 .

Stellar. I checked out that branch and re-ran the tests locally, and the situation is improved. I have some lingering fear because the first time I ran my tests locally they all completed successfully, but the process hung afterwards and wouldn't shut down the test suite. But I've since re-run the whole test suite and the problematic tests directly (with attempts from both the file- and function-levels) and can't reproduce the problem.

Aha, ok, splitting up and running from the function level reveals that the _CacheTestClusterExecutor gives a warning:

/Users/liamhuber/dev/miniforge3/envs/pyiron12/lib/python3.12/subprocess.py:1127: ResourceWarning: subprocess 94559 is still running
  _warn("subprocess %s is still running" % self.pid,
ResourceWarning: Enable tracemalloc to get the object allocation traceback

My gut reaction is that this is probably the thread I start to shut down the executor? In any case, I can't get it to reproduce a problem beyond the warning level.

[jan-janssen]

@liamhuber just for confirmation, the changes to executorlib fix the issue, correct? Then I merge those and release 1.9.1

[liamhuber]

Confirmed. There's a disconcerting warning, but at least on my local machine that PR gets all the tests working well. 1.9.1 would be appreciated; I'll bump to it tomorrow (pacific)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.36%. Comparing base (38009e1) to head (c23a014).
⚠️ Report is 5 commits behind head on shutdown-recursion.

Additional details and impacted files

@@                  Coverage Diff                   @@
##           shutdown-recursion     #817      +/-   ##
======================================================
- Coverage               93.46%   93.36%   -0.11%     
======================================================
  Files                      38       38              
  Lines                    3873     3873              
======================================================
- Hits                     3620     3616       -4     
- Misses                    253      257       +4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[jan-janssen]

@liamhuber I updated executorlib to 1.9.1 and all tests pass - so from my perspective it’s ready to be merged.

[liamhuber]

@liamhuber I updated executorlib to 1.9.1 and all tests pass - so from my perspective it’s ready to be merged.

Agreed 💯