Fix and add test for when locusts fail to exit at end of iteration during stop timeout.

[cyberw]

I discovered a bug introduced in the last commit on stop-timeout. Adding a test case for it (and once you have seen that it actually fails), I will revert that commit to show that it now works :)

[codecov]

Codecov Report

Merging #1127 into master will increase coverage by 0.16%.
The diff coverage is 0%.

@@            Coverage Diff             @@
##           master    #1127      +/-   ##
==========================================
+ Coverage   74.58%   74.75%   +0.16%     
==========================================
  Files          18       18              
  Lines        1802     1806       +4     
  Branches      269      271       +2     
==========================================
+ Hits         1344     1350       +6     
+ Misses        394      389       -5     
- Partials       64       67       +3

Impacted Files	Coverage Δ
locust/core.py	85.21% <0%> (-1.51%)	⬇️
locust/stats.py	82.04% <0%> (ø)	⬆️
locust/clients.py	91.48% <0%> (ø)	⬆️
locust/runners.py	64.48% <0%> (+1.63%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 5c84591...b5c808b. Read the comment docs.

[heyman]

Ah, I see, my bad! Could we move the check from the wait() method to immediately after self.execute_next_task() instead?

[cyberw]

Ah, I see, my bad! Could we move the check from the wait() method to immediately after self.execute_next_task() instead?

You mean after line 365? That doesnt work, because it wont be called on interrupts.

[cyberw]

Ah, I see, my bad! Could we move the check from the wait() method to immediately after self.execute_next_task() instead?

You mean after line 365? That doesnt work, because it wont be called on interrupts.

Wait, I get it now, you meant move the other check. Yes, that works, and arguably makes more sense. Commiting it now...

[heyman]

Hmm, I guess we also need to to do the check before execute_next_task for the case when the stop happens while the TaskSet is executing on_start (otherwise it will execute another task before stopping).

[cyberw]

Always running one iteration when using stop-timeout should be fine imo (maybe even desirable). I can definitely add it, but personally I dont think it is worth bloating the code.

[heyman]

Yes, it does bloat the code a bit, but I do think we should do it. A test case for it would probably also be good, to make sure we don't get a regression at some point in the future.

[heyman]

Nice! LGTM?