-
Notifications
You must be signed in to change notification settings - Fork 26
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Weird segfaults in PodioOutput #156
Comments
From the stacktrace alone this looks like a collection is missing in the events where this is failing. Can you run with LCIO output and see if that is the case? (Not sure if we get all of them as REC and DST outputs might not write all the collections that are available). If a collection is missing in the EventStore (i.e. the Frame backing it), then the Frame will try to get the collection from FrameData and would indeed come across If you are able to find out which collection is missing in some (rare) cases the proper fix would be to make the processor always produce a collection even if it is empty. The quick fix would be to use the |
This is for
|
REC should write everything or not? Given that nothing is dropped?
|
Yes, I think it should write everything with this configuration. The anajob output does not really give us a lot here indeed. Can you get the collection name for which the thing crashes from the debugger? I think we should have debug symbols, but the actual strings might be optimized out. Maybe by running thing inside gdb until they crash and then going to the |
to quote gdb: |
Thankfully
Which is missing when there are no SimCalorimeterHits. Fixed by key4hep/k4MarlinWrapper#157. I was additionally confused because only one of the three broken events has no SimCalorimeterHits in the sim file but apparently for the other two they are deleted during reconstruction, possibly here. I am not sure how all calorimeter hits of a 200 GeV pion can be out of time but that is not relevant for this issue. |
Urgh I was quite sure that I had tested key4hep/k4MarlinWrapper#157 for all 3 files but |
Ok, its only the same stack trace but caused by a missing CalohitMCTruthLink collection ultimately caused by ConformalTracking throwing a SkipEventException in the prior event (913). Maybe PodioOutput should handle this a bit more gracefully... |
Yeah, |
I have a bunch of files that I'm reconstructing and for 3 out of 200 something Gaudi crashes with this segfault, always in the same event per file:
I have found no obvious difference with these events compared to others in the same file and I am a bit lost on how to debug further.
Should be reproducible by
The crash should happen at event 638. All three files are here: https://cernbox.cern.ch/s/oSYq2xtW01XEdTx for
pi-_89deg_1GeV
it crashes at event 41 and forpi-_40deg_200GeV
at event 914.I am a bit suspicious of this piece of podio that gets called in the process and might return 0? but my podio knowledge is too limited to know if that can happen..
The text was updated successfully, but these errors were encountered: