Puma fails when it restarts itself

169 views
Skip to first unread message

Rob Daniels

unread,
Aug 29, 2013, 5:51:24 PM8/29/13
to rubyonra...@googlegroups.com
I'm having an issue with Puma, it fails when it tries to restart itself. When this happens it kills all the Puma processes, resulting in nginx delivering a 502 Bad Gateway to our users. I'm able to start, stop & restart it with my /etc/init.d/puma script (it runs /usr/local/bin/run-puma),  it just has issues when it restarts itself. The puma.log file doesn't show anything just useful for troubleshooting, just

Puma starting in single mode...
* Version 2.5.1, codename: Astronaut Shoelaces
* Min threads: 16, max threads: 32
* Environment: production
* Listening on unix:///home/deploy/app/shared/sockets/puma.sock

and my workers. Any help would be greatly appreciated.

Also - i have whats probably a pretty dumb question. Is there a way for me to allocate more resources to Puma so it doesn't need to restart itself so frequently? I'm currently only using around 25% of the server's memory.

Thanks for the help

Hassan Schroeder

unread,
Aug 29, 2013, 6:42:47 PM8/29/13
to rubyonrails-talk
On Thu, Aug 29, 2013 at 2:51 PM, Rob Daniels <r.da...@gmail.com> wrote:
> I'm having an issue with Puma, it fails when it tries to restart itself.
> When this happens it kills all the Puma processes, resulting in nginx
> delivering a 502 Bad Gateway to our users. I'm able to start, stop & restart
> it with my /etc/init.d/puma script (it runs /usr/local/bin/run-puma), it
> just has issues when it restarts itself.

So, "restarts itself"? Do you have any idea why? I've never seen any
of my own Puma instances spontaneously restart.

Have you looked in the system logs for anything relevant? Is there
a mismatch between you-the-user-manually-starting-it and the user
that the process is normally running as/restarting as?

What Ruby are you running?

--
Hassan Schroeder ------------------------ hassan.s...@gmail.com
http://about.me/hassanschroeder
twitter: @hassan

Rob Daniels

unread,
Aug 30, 2013, 8:46:49 AM8/30/13
to rubyonra...@googlegroups.com
FIXED
I had the restart_command in puma.rb trying to run 'puma.sock restart', I pointed it to my init.d file and it no longer has an issue when it boots workers

restart_command '/etc/init.d/puma restart'

I still occasionally get a 504 timeout under very heavy load, but i have the god gem monitoring too and it recovers from that quickly
Reply all
Reply to author
Forward
0 new messages