mcollective upstart script starts too many processes

[xian]

It looks like the upstart script for mcollective does not work

Installed mcollective 1.2.1-0ubuntu2.1

No mcollectived processes running.
# ps -ef | grep mcollectived
#

Start mcollective
# service mcollective start
mcollective start/running, process 13451
# ps -ef | grep mcollectived | grep -v grep
root 13454 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13459 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13464 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13469 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13475 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13480 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13485 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13494 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13499 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13504 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13509 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
#

Obviously results in multiple mcollective processes started

Attempting to stop mcollective does not work either...
# service mcollective stop
stop: Unknown instance:
# ps -ef | grep mcollectived | grep -v grep
root 13454 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13459 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13464 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13469 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13475 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13480 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13485 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13494 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13499 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13504 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
root 13509 1 0 13:01 ? 00:00:00 ruby /usr/sbin/mcollectived --pid=/var/run/mcollectived.pid --config=/etc/mcollective/server.cfg
#

System info:
# lsb_release -rd
Description: Ubuntu 11.10
Release: 11.10

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Karel Vervaeke (karel-vervaeke)]

Adding an 'expect fork' to /etc/init/mcollective.conf fixes the problem

[Karel Vervaeke (karel-vervaeke)]

This patch adds 'expect fork' to the upstart file. There used to be an 'expect daemon' in natty, but it got replaced by 'respawn' in this commit: http://bazaar.launchpad.net/~ubuntu-branches/ubuntu/precise/mcollective/precise/revision/9/debian/mcollective.upstart

[Marc Cluet (lynxman)]

This is due to having daemon=1 on the mcollectived configuration file which is recommended against in upstart.

The right fix is to set daemon=0 on mcollective config

[Marc Cluet (lynxman)]

Marked as invalid since the proposed solution by Karel would confuse upstart's tracking of the process, that's why it was changed in the first place

Thank you anyway Karel :) much appreciated!

[Karel Vervaeke (karel-vervaeke)]

Thanks for the info! Btw, my mcollective configuration was managed by
this puppet module:
https://github.com/puppetlabs/puppetlabs-mcollective

I'll see if I can get it fixed there...
On Tue, Jan 24, 2012 at 2:25 PM, Marc Cluet <email address hidden> wrote:
> Marked as invalid since the proposed solution by Karel would confuse
> upstart's tracking of the process, that's why it was changed in the
> first place
>
> Thank you anyway Karel :) much appreciated!
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/918348
>
> Title:
>  mcollective upstart script starts too many processes
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ubuntu/+source/mcollective/+bug/918348/+subscriptions

[xian (mchugh19)]

Thanks! Daemon=1 is needed on debian, but not for ubuntu.

I have a custom puppet module to for mcollective which uses ssl certs instead of a shared key. I added logic for debian/ubuntu to push out the proper server.conf and it works great.