Opened 4 years ago
Last modified 3 years ago
#6048 new defect
FMI testing of models without states behaves in a bizarre way
| Reported by: | Francesco Casella | Owned by: | Martin Sjölund |
|---|---|---|---|
| Priority: | blocker | Milestone: | 1.19.0 |
| Component: | Testing Framework | Version: | |
| Keywords: | Cc: | Lennart Ochel, Adrian Pop |
Description
Please check this Jenkins report. About 200 FMI test cases went from almost instantaneous to slightly above 50 seconds. About another 200 went the other way round.
The timeout of these tests is set to 50 s, so this may hold a clue.
Apparently, what all these tests have in common is that the models have no continuous states. I guess this causes something to go wrong somewhere.
Change History (5)
follow-up: 2 comment:1 by , 4 years ago
comment:2 by , 4 years ago
Replying to Karim.Abdelhak:
Isn't this just the usual multi thread problem? Some tests have to wait for others to finish and therefore get this artificial high computation time. It switches because it is a little fragile and seems random which machine finishes first and gets assigned new tasks as far as i know.
Aha. Do you mean our FMUs are not multi-threaded, so they can't run on multi-cpu Jenkins tests?
But why does it only happen if there are no states?
Isn't this just the usual multi thread problem? Some tests have to wait for others to finish and therefore get this artificial high computation time. It switches because it is a little fragile and seems random which machine finishes first and gets assigned new tasks as far as i know.
I thought it is a thing for some time now, but you are right we should probably fix it.