Closed nr23730 closed 1 month ago
Thank you for the detailed error report.
The problem seems not to be related to the last docker image, though, but rather to the little changes we made to the deployment setup last week:
I could reproduce the error with the last deployment setup: it occurs when the backend process doesn't have correct access rights to the persistence directory. The problem disappeared when this was ensured as described.
Could it be that you forgot to set these permissions?
Previously, the described option to ensure proper access of the backend to the persistence directory was to assign the backend service in docker-compose.yml
a suitable system user. However, since we are trying to adapt the deployment process so that none of the versioned deployment files require local modifications, the opted for solution a.t.m is to use chmod -R 777 ...
on the persistence directory.
I hope this solves the issue for you as well, else don't hesitate to follow up (you can also contact me by mail: lucien.clin@uni-tuebingen.de)
Thank you very much! (Re-)applying the permissions did not fix the issue. However, clearing all data present in the node did fix the issue. This is not an issue, as this one is test anyway, for production it would be a problem...
Thank you for the follow-up.
OK, then it might not have been the access permissions, but really parsing errors while loading the present JSON data -- although I was quite sure there weren't any breaking changes in the data objects recently, and thus discarded this potential error source.
I'll add more detailed logging output to enable easier problem diagnosis in such situations.
Thanks again for the feedback.
P.S.
I published a new "backend" container image with the aforementioned additional logging output. Of course, this is not a proper solution to the problem you experienced and reported, with breaking changes in the persisted data structures, but for now this will at least facilitate error diagnosis in such cases.
Eventually, for production, we'll have to plan migrations of persisted data in case of breaking changes. I'll keep this mind.
Recents builds of the backend produce lots of error messages, while the API is also unavailable with the following error:
This is the error in the log:
Affected versions:
ghcr.io/kohlbacherlab/dnpm-dip-backend@sha256:28e51a9b306a01cb586a5197a4a57cc993b74a0a948b9f85c70e2ddedd68b188
and upwardsThis is the last docker image that works:
ghcr.io/kohlbacherlab/dnpm-dip-backend@sha256:d7e703dbea28cd2db76081021a4b1f8141433a0c6d341a51a8ac1c4bf4a8f867
Tested with deployment @ da3f5d6a7a51b0a4fe6a026a0cb4f6a6ce7eb526