librelist archives

« back to archive

Redis socket timeout

Redis socket timeout

From:
Eric Tang
Date:
2012-01-27 @ 19:43
Hi,

I'm running resque on an EC2 cluster.  I have about 10 workers running on 2
large instances, and another large instance running redis itself, but every
3 minutes the resque workers get killed because the socket connection to
redis times out.

Has anyone ran into this issue before?

Eric Tang

Re: [resque] Redis socket timeout

From:
Aaron Quint
Date:
2012-01-28 @ 04:13
Yes, we were running into this for a while. There are a couple of
possible contributing factors. 1) You should use the hiredis gem as it
generally deals better with connections. 2) You might need to move
redis BGSAVEs to a slave:

http://dev.paperlesspost.com/blog/2011/12/21/redis-tips-persistence-only-slaves-and-the-slowlog/
3) As a last resort, you can increase the timeouts of the redis
client.

--AQ

On Fri, Jan 27, 2012 at 11:43 AM, Eric Tang <eric.x.tang@gmail.com> wrote:
> Hi,
>
> I'm running resque on an EC2 cluster.  I have about 10 workers running on 2
> large instances, and another large instance running redis itself, but every
> 3 minutes the resque workers get killed because the socket connection to
> redis times out.
>
> Has anyone ran into this issue before?
>
> Eric Tang