@marclaporte_gitlab: I did a small amount of benchmarking to compare OSS sentry vs GlitchTip. My generic conclusion, is that we are close-ish to OSS Sentry. I don't know much about current, proprietary sentry.
On a small server, OSS Sentry gets 110 to 200ms response times on accepting events (not counting first request). GlitchTip gets pretty close.
Now, we don't have to compare everything to Sentry if we don't want to. I believe sentry does some synchronous checks before responding with the accepted event ID. I chose to make GlitchTip do that too. But it's probably an option to instead check only authentication and then drop it into celery. That would surely increase response time. The cost might be general overhead and the server may respond that everything is fine, when really only authentication is fine and the event errored out while saving.
Currently my priority is fixing the environments filter to be more performant.
/adminshould not be necessary. Are you saying you couldn't get invite working without /admin? @mikkeschiren
SELECT strip(jsonb_to_tsvector('english', jsonb_agg(data), '["string"]')) FROM (SELECT events_event.data from events_event where issue_id = x limit 1000) as data;
ERROR: total size of jsonb array elements exceeds the maximum of 268435455 bytesand
string is too long for tsvector (1101146 bytes, max 1048575 bytes)and is potentially slow.
I haven't tested discord, do you think you could open an issue on gitlab and ideally a merge request with a failing unit test?
Webhooks use the slack-like message syntax. So if an app doesn't support it, then it won't work.