Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [sumo-dev] possible multi QThread safety violation on single sink loopdetector write

I am using sumo option --threads.

No multi-threading in the runner script.


Von: sumo-dev-bounces@xxxxxxxxxxx <sumo-dev-bounces@xxxxxxxxxxx> im Auftrag von Jakob Erdmann <namdre.sumo@xxxxxxxxx>
Gesendet: Donnerstag, 24. September 2020 07:52:02
An: sumo developer discussions
Betreff: Re: [sumo-dev] possible multi QThread safety violation on single sink loopdetector write
 
Are you using sumo option --threads or is your python script multi-threaded?

Am Mi., 23. Sept. 2020 um 18:25 Uhr schrieb Richter Gerald <Gerald.Richter@xxxxxxxxx>:
Hi again,

ran another test
after splitting up the iIL_detector output into separate files (1 for
each detector) again:
* python3 shell, 4 threads: all fine

best,
Gerald

On 23.09.20 17:51, Richter Gerald wrote:
> Hi,
>
> I got a scenario and execution scripts using a traci runner
> where threads were set to 4.
> While rebuilding and configuring this, I set loopdetector output file
> identical
> for all the instantE1 detectors.
>
> running the runner.py script from an ipython3 shell
> repeatedly for different seeds,
> apparently there is a thread race going on,
> because traci server crashes indeterministically with
> FatalTraCIError: connection closed by SUMO
>
> * this did not occur before the single loopdetector output file
> * it ceases to be a problem when threads are set to 1
>
> and debugging into this state gives a lot of pouring QThread messages.
>
> so right now I am not sure if it is the thread-twitchyness of ipython,
> or something else... testing with
> * python3 shell, 4 threads: crashes
> * python3 shell, 1 threads: all fine
>
> is this a known phenomenon?
>
> cheers,
> Gerald
>
> _______________________________________________
> sumo-dev mailing list
> sumo-dev@xxxxxxxxxxx
> To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-dev
_______________________________________________
sumo-dev mailing list
sumo-dev@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-dev

Back to the top