Graphite datasource timeout - http status 499


We use nginx as a reverse proxy in front of graphite.
All clients (including grafana) access graphite via nginx.
As grafana queries to graphite get longer than ~3 seconds we get a significant number of status 499s logged by niginx. My understanding is that nginx 499 statuses are caused by client hangups.
I’m looking for suggestions on how to troubleshoot this.



Did you get anywhere with this? Are there any javascript/network errors in the Chrome Developer console? Could it be Grafana that is causing the connections to be closed?

Rather late last night I figured out what is happening.
It is not grafana that is dropping out - it is graphite when it is performing remote finds.
Neither grafana nor graphite send a user agent header which would be really useful.

Just discovered that graphite can set an optional user agent header. this will make future troubleshooting easier.

1 Like