Forum hitting some sort of limits and responding slow or not at all


#1

Hey everyone.
I just recently dug into discourse and docker and installed it via some howtos/the docker installation guide on an AWS host running ubuntu.

The forum does have some elevated attention at times and this leads to it becoming unresponsive, despite the fact the machine does not seem to be fully utilizing it’s resources (load low, mem usage low).

Is there some sort of application based limitation that would be open for tweaks? (nginx, or unicorn)

Thanks in advance!
dym

EDIT: Most common error at these times is 504.


(Régis Hanol) #2

Do you see anything that might be related in /logs?


#3

Well, theres just a a lot of

45#45: *342611 upstream timed out (110: Connection timed out) while connecting to upstream,

going on in the nginx config.

Is there somewhere else to look for debugging purposes?


(Régis Hanol) #4

Sorry, I meant in http://your.discourse/logs :wink:


#5

No worries :slight_smile:
But no, unfortunately it shows nothing unusual at all.


(Régis Hanol) #6

I would tweak the number of :unicorn: workers


#7

Ah, okay. That’s what i was looking for.
The value was still on default.

Is there anything else tweakable?
Also: Does this always require a rebuild?


(Régis Hanol) #8

Well, it highly depends on what your bottlenecks are.

Yes. Every time you change something in the container definition, you have to do a rebuild.