faraday timeout errors for particular project

[Ameya Savale]

Hi,

I'm running into a strange situation. I have sentry setup with redis and celery. I am able to see the majority of events, except for events from two projects both of which are under the same team. When I try to send an event to the project using its dsn I get the following error

ERROR -- : ** [Raven] Unable to record event with remote Sentry server (Faraday::TimeoutError - execution expired)

After I restart the sentry instance I'm able to pass it events, but after a while it stops working. I have also enabled rate limiting. Has anyone run into something like this?

- thanks all

[David Cramer (zeeg)]

Is it possible that the server running the web is the same as the workers and it's getting cpu and memory bound?

Rate limits will help throw things away but it could still be overloaded. Also which broker are you using with celery? And which version of sentry?

_____________________________
From: Ameya Savale <am...@simplyhired.com>
Sent: Thursday, April 2, 2015 7:24 PM
Subject: faraday timeout errors for particular project
To: <gets...@googlegroups.com>

--
You received this message because you are subscribed to the Google Groups "sentry" group.
To unsubscribe from this group and stop receiving emails from it, send an email to getsentry+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ameya Savale]

By workers do you mean celery workers then yes, but the memory and cpu seem fine on the box. For celery I'm using a redis broker and the version of sentry I'm using is 7.0.0

--
You received this message because you are subscribed to a topic in the Google Groups "sentry" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/getsentry/lQpHBUilZHo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to getsentry+...@googlegroups.com.

[David Cramer]

Is the database on the same server as well? One thing to check, especially when it gets slow, is if theres a bunch of active queries, in addition to the overall size of the queue. I don’t recall how to query the size of the queue with redis (it might just be len [queue name]), but database should be fairly straightforward.

[Ameya Savale]

the database is on the same box, but I'm not seeing a large amount of queries taking place. If that was the choke point wouldn't I not be getting events from my other projects as well, and one of the projects generates around 10,000 events per hour, but all those projects are logging events fine, except for two which are both under the same team.

[David Cramer]

Is there anything off about the projects?

i.e. do they have a higher rate of events? are the number of unique events higher?

[Ameya Savale]

Hi

Finally managed to figure it out, the python raven client being used was 5.2 which was not working with sentry server version 7.0.0. I upgraded the sentry server and events started coming in. Thanks for all your help

- Ameya Savale