Sending async MDN's fails and stops working when a read fails on an object about django-pyas2 HOT 4 CLOSED

Yes that is true but the large issue seems to be this error which is not expected. Do you have any idea how an MDN with empty payload got generated?

from django-pyas2.

@abhishek-ram : Last year, this happened twice and I could not identify this for sure. Only suspicion I have is, because it is run in a Kubernetes environment, the shut-down grace period of the container was not considered to finalize the job properly: MDN created, container stoped before payload was saved. This grace period was extended and I did not have this case anymore.

from django-pyas2.

@abhishek-ram : Just a follow up on this. The cases have increased and shown up in other areas as well which led me to think that DB connections dropped in mid-process. I therefore switched config settings to: 'ATOMIC_REQUESTS': True which resolved the issue in most cases.

I still have an issue when a partner sends me ASYNC MDN's for which I don't have a message for. Maybe management command messages are not affected by atomic request handling or something else. I'll review at some point, but currently no capacity for that.

from django-pyas2.

Yes, the ATOMIC_REQUESTS is only applicable to Views. We would need to manually put the command inside a transaction

from django-pyas2.