Re: [monit] bug report / help request: monit+mongrel occasional misbehavior w/ port 8014

[Martin Pala]

Hi,

upgrade to monit-5.0 may solve the problem ... there were changes which fixed situations similar to your description (monit now start the services synchronous and it also allows to customize the start method timeout).

You can get latest monit beta here:

http://www.tildeslash.com/monit/dist/beta/

(note that although it's beta, it is very stable and used in production on thousands of places already).

I it won't help, please run monit in verbose mode (-v option) and provide the logfile for the timeframe where the start attempt failed so we can see the root cause.

Thanks,

Martin

On Sep 22, 2008, at 2:50 AM, Elijah Wright wrote:

Hi all,

We're having some seemingly monit-related misbehavior on one of our machines.

We use monit to control a pool of mongrels, 15 per host.

Our pool runs on ports 8000 through 8014.  Of these mongrels, most of them work normally.

However... occasionally the mongrel on port 8014 of a given host will just act *crazy*.  Either
it won't start up after a monit restart is issued to the pool, or it will fail to restart when it is told to do so.

Here's what our monit config for port 8014 looks like:

check process mongrel_8014 with pidfile /home/rails/application/production/current/log/mongrel.8014.pid
  group mongrel
  start program = "/usr/local/bin/mongrel_wrapper start 8014"
  stop program = "/usr/local/bin/mongrel_wrapper stop 8014"
  if failed url
    http://127.0.0.1:8014/status
    and content == "OK"
    with timeout 120 seconds for 2 cycles
    then restart
  if memory > 120 Mb then restart


This fragment is included by monitrc - in our local environment (Ubuntu Hardy on amazon ec2), I'm tending to drop these rules into files in /etc/monit.d.

This is monit 4.10 - hand-compiled, not installed from the ubuntu .deb.

Has anyone seen an issue like this previously?  We'd be happy to pay someone with the appropriate skills to debug it - drop me a line if you're interested in that work (and qualified, of course).

--elijah

--
To unsubscribe:
http://lists.nongnu.org/mailman/listinfo/monit-general