tickets #94540
closed
mailman - openqa-admin has 11597 non-members - nginx timeout
Added by pjessen over 3 years ago.
Updated 6 months ago.
Description
I was going to tidy up some of those thousands of non-members (reply+some-identifier@reply.github.com), but I keep getting a 504 nginx time-out.
I'm just recording it here, just in case. It looks like the number grows slowly but steadily.
I don't know what this list is for, but those messages are not being held, so I guess they are okay - I'm going to add a regex filter to accept them by default, maybe that will stop the number growing :-)
- Private changed from Yes to No
I have added
^reply\+[^@]+@reply.github.com$
to the non-member acceptances.
Well, that made no difference, they are still coming and being recorded.
Now 13058 non-members listed, so approx 1500 new in a month.
I get the 504 after 60seconds - after some research, I changed ./vhosts.d/lists.opensuse.org.conf:
location / {
include /etc/nginx/uwsgi_params;
uwsgi_pass 0.0.0.0:8000;
uwsgi_read_timeout 600s;
client_body_timeout 300s;
}
After hitting 'Delete' for a non-member, it takes (for instance) 4m52s for the next page "Confirm remove role" to turn up. Hit "Remove" and I estimate it took another 3-4minutes for it to complete, but I forgot to time it. On a 2nd attempt "Remove" timed out after 5m50s.
That is just simply ridiculous - not much point in trying to use the web interface for dealing with those non-members, now about 20000.
- Related to tickets #113111: lists.opensuse.org - webserver backand timeouts - nginx 504 gateway timeout added
- Status changed from New to Rejected
I understand the annoyance, but think significantly increasing timeouts on the webserver to have it wait for such special cases of huge jobs is not reasonable, as it will cover up broken small jobs.
I recommend administrators to use the Python client for large jobs. We could think of making a collection of scripts to ease the barrier.
Also available in: Atom
PDF