[Bug]: PABOTEXECUTIONPOOLID not correctly set

[azdludovic]

What happened?

Hello!

Since version 5.2.0 (probably commit d509a0a, but i'm not familiar enough with you codebase), PABOTEXECUTIONPOOLID seem to have change its behavior:
My interpretation is PABOTEXECUTIONPOOLID is set to integer representing in which "process pool" the current test is running. Thus, we use it to compute value for shared ressources (like system port, tmp file etc).

Now, the behavior seem different, and I'm not sure what PABOTEXECUTIONPOOLID means.

Here is an example output of "correct/previous" behavior:

2026-04-09 07:24:54.477057 [PID:102] [3] [ID:5] still running Test Suites.testA after 15.0 seconds
2026-04-09 07:25:00.890282 [PID:110] [2] [ID:2] PASSED Test Suites.test1 in 21.4 seconds
2026-04-09 07:25:00.890890 [PID:886] [2] [ID:6] EXECUTING Test Suites.test3

What I see: test1 finishes in "pool" 2, test3 starts in pool 2

With 5.2.2 I got this result:

2026-04-09 07:24:54.477057 [PID:102] [3] [ID:5] still running Test Suites.testA after 15.0 seconds
2026-04-09 07:25:00.890282 [PID:110] [2] [ID:2] PASSED Test Suites.test1 in 21.4 seconds
2026-04-09 07:25:00.890890 [PID:886] [0] [ID:6] EXECUTING Test Suites.test3

What I see: test1 finishes in "pool" 2, test3 starts in pool 0 ? (while pool 0 is still running another test)

Steps to reproduce

1. Run a few tests in parallel and print PABOTEXECUTIONPOOLID
2. The PABOTEXECUTIONPOOLID won't represent correct "pool id"

Version

5.2.2

[joonaskuisma]

Hi and thanks for bug report!

Actually, I have been noticing this same issue earlier when I was creating new upcoming features, but I supposed that issue was mainly cosmetic because nobody has been reported that earlier. Sorry about that assumption.

There's a fix already in my development branch: joonaskuisma@d1085e2
And I will try to release next version pabot 5.3.0 hopefully soon.

[azdludovic]

No worries, we reverted to 5.1.0 for now ! I guess we should use the resources files rather than that for this, but the POOLID trick is neat and easy is for us!

[azdludovic]

And thanks for the good work, we use your tool daily !

Just a though : it would be great to have the pool id in the Suite Metadata to have it in the log.html !

[joonaskuisma]

Hi, and nice to hear that! 🙂

Actually, a solution for that is being planned, most likely in pabot version 5.3.0. Suite metadata would otherwise be a good option, but it doesn’t work with the --testlevelsplit option. Since Robot Framework does not (yet) support test-specific metadata, the plan is to solve the issue by using pabot’s own automatically added tags. You can find an example of this approach at least on Slack: https://robotframework.slack.com/archives/C7HKR2L6L/p1772283992668759