AllenNeuralDynamics / AllenNeuralDynamics/aind-smartspim-quantification
Matlab credentials appear to time-out
- Langage dominant
- Python
- Étoiles
- 1
- Forks
- 2
- Métriques de merge des PR
- Aucune PR mergée en 30 j
Description
I am working on a capsule to do process very large datasets using matlab, here https://codeocean.allenneuraldynamics.org/capsule/4553637/tree.
I've applied mathworks credentials to my Code Ocean credentials tab and it generally works at triggering runs. For technical reasons that have to do with the processing packaging and strange memory leaks on ubuntu, my matlab script batches the processing and splits it into some child processes.
The processing can take up to 40 hours on these machines. Since I added in the subprocess-dispatching code, the runs fail with an authentication error, shown below, when matlab tries to create a child process around the 30 hour mark. I suspect this is because the way CO is handling the authentication credentials behind-the-scenes causes them to expire after some time, probably 24 hours.
```
root/capsule/data/slap2_session/slap2/dynamic_data/acquisition_20251120_143726_DMD2-TRIAL000107.dat
{License checkout failed.
License Manager Error 4410
Invalid security token. (Status code: Unauthorized)
Troubleshoot this issue by visiting:
https://www.mathworks.com/support/lme/4410
Diagnostic Information:
Feature: Image_Toolbox
License path:
Licensing error: 4410,0.
Error in processAllTrials_Async>loadTrial (line 114)
labeled = medfilt2(meanIM(:,:,params.activityChannel), [3 3]);
Error in processAllTrials_Async (line 11)
CD = loadTrial(dr, fns{nLoad},fls(nLoad),els(nLoad),selPix,discardFrames{nLoad}, alignData{nLoad}, meanAligned(:,:,:,nLoad), motOutput(:,nLoad), roiData, params);
Error in processOneTrialBatch (line 66)
E = trialProcessor(data.dr, data.fns, data.fls, data.els, ...
}
Parallel pool using the 'Threads' profile is shutting down
```
Is this an internal bug, or is there any quick workaround?
Guide de contribution
Aucun guide de contribution indexé pour ce dépôt
Évaluation
Cette issue n'a pas encore été évaluée.