I just realized that the logger emails sent to the support address when an error occurs appear as coming from a non existing host (indico-test instead of indico) according to the From: line. I have no clue where it is coming from and this can be changed…
I understand… but there is clearly a problem on our instance and I’d prefer to fix it rather than hide it (by defining WORKER_NAME explicitly)… Do you have any suggestion of a small test script that I could run to see if I can reproduce the problem outside Indico?
I think I found the problem… Indico is hosted on a VM whose system image has indico-test as a hostname and the actual hostname is set during the VM contextualization with httpd restarted afterwards. I guess that we also need to restart indico-celery and uwsgi, is not it?