Comments on: Monit makes Mongrel play nice!

By: D. Rothlisberger

D. Rothlisberger — Mon, 07 Jul 2008 14:48:28 +0000

If you are trying to configure monit to start your mongrels and it simply won’t (with no complaints from monit) check your mongrel logs.

I had to set my monit config’s start line to something like this:
start = "/usr/bin/env PATH=$PATH:/usr/local/bin HOME=/home/user /usr/local/bin/mongrel_rails cluster::start -C /home/user/path/to/app/config/mongrel_cluster.yml --only 8000"

By: Francois

Francois — Tue, 25 Mar 2008 19:04:26 +0000

Thanks for the great info. This setup wouldn’t work for me on FreeBSD 6.x unless I add this line to /usr/local/bin/mongrel_rails

ENV['PATH'] = “#{ENV['PATH']}:/usr/local/bin”;

By: Alex Egg

Alex Egg — Wed, 14 Nov 2007 22:57:49 +0000

You should really check out god — much nicer and tailored for rails.

http://god.rubyforge.org/

By: Ilya Grigorik

Ilya Grigorik — Tue, 23 Oct 2007 04:13:51 +0000

Jesse, thanks for the clarification. Unfortunately the link appears to be broken - hopefully it’ll come back!

By: Jesse Andrews

Jesse Andrews — Wed, 17 Oct 2007 00:30:10 +0000

Billy/Ilya - to restart mongrels via capistrano without monit thinking something bad is happening and it relaunching them you need to restart them via monit:

monit -g (group name here) restart all

an example via cap 2

By: Ilya Grigorik

Ilya Grigorik — Sun, 14 Oct 2007 04:01:13 +0000

Billy, that’s really odd behavior. I have a similar setup and I’ve never experienced this problem. Are you using latest distribution of monit? Another solution would be to execute a stop all and then ‘monit quit’. Once you’re done with deploying new code, just bring up monit once again.

By: Billy

Billy — Fri, 12 Oct 2007 02:52:42 +0000

We are having problems when we use capistrano to deploy. We execute a monit stop all before we update the code through SVN, but even when doing monit stop all, monit tries to restart the processes automatically. We could tell it to unmonitor all before we deploy but then we wouldn’t get notices if for some reason the deployment failed and our mongrels didn’t come back up. Have you encountered this problem yourself and if so, how did you get around it ? Great article, btw :)

By: Tom

Tom — Tue, 25 Sep 2007 20:52:39 +0000

Chetan,

Did you abandon using the cluster? I was getting the same results as you:

$ sudo monit start mongrel_8050 'mongrel_8050' start: /usr/bin/mongrel_rails 'mongrel_8050' failed to start starting port 8050

When using the following start program = “/usr/bin/mongrel_rails cluster::start -C /etc/mongrel_cluster/my.conf –clean –only 8050″

By: chetan

chetan — Mon, 24 Sep 2007 05:23:52 +0000

thank lots team we got it !!!!!!!!!!!!!!

By: Ilya Grigorik

Ilya Grigorik — Sat, 22 Sep 2007 21:21:55 +0000

Chetan, mongrel_cluster does not write any PID files since it controls the entire cluster. Hence, all subsequent checks will fail. Instead of using mongrel_cluster, start each mongrel instance individually (use the line provided in my previous reply).