OHDSI Home | Forums | Wiki | Github

Cohort generation not working

Developers
,
#1

We have 3 omop cdm schemas each having their own results schema. In one we are able to generate a cohort successfully. The other two are not working. There is no error in the ui or logs.

We hit the generate button in the atlas ui and it goes into a red pending state. We can see the job associated with the generation starts and completes. Nothing else happens. If we refresh the page it will go back to state prior to hitting the generate button.

I tried running the exported sql against the database directy. It ran ok and inserted records into the results database.

I’m hoping someone can point out a way to move forward with debugging.

Capture
Capture1003×199 6.2 KB

#2

Hmm, if i was in this position, (no log messages, no errors) I’d have to just get a IDE to execute the code in a local debug session. The entry point for generation is here, which is where you can start debugging.

#3

Thanks for the reply. The issue only happens in a production environment and I can’t export the data to local. I could add some logging lines around the code you mentioned though. Maybe that will show what is wrong.

#4

I would start by checking permissions across all results schemas to ensure that your service account has proper access across all 3 of them.

btw, what database are you running?

#5

Hi. It was permissions. I didn’t realize you have to authorize for new cdm schemas. Thanks.

1 Like
#6

So, looks like there may be a bug there then: that you were able to click on the ‘generate’ button next to the source when you didn’t have permissions…the button should be disabled if you don’t have access to the source. We’ll look into that behavior, but I thought I saw (at least as of the 2.8 release) if security is enabled but you don’t have access to a source, the generate button is disabled.

#7

Hi. You don’t need to investigate. I’ve modified the code to disable authorization so I must have missed a place. In our system once you authenticate all users should have access to see all data since we have multiple instances each having the subset for a particular research group.

t