Ask Your Question
0

Issues with linux mco discovery and report backs...

asked 2015-04-03 13:26:29 -0500

SKasai gravatar image

updated 2015-04-06 13:32:08 -0500

Within the last month, our CentOS 6 machines revved to 2.8.1 of MCollective. The Puppet Master is able to MCO ping our windows machines in MCollective and itself, however, the CentOS machines we have are not responding back to MCO ping.

We have a reboot process which involves "mco rpc --agent=reboot --action=reboot -I /^[Ee][Ss]/"

Where it is to reboot the CentOS machines with names starting in 'es'.

However, it is not seen by those machines. If we do the command but put in '-I es-1 -I es-2'... the machines see the request, however, do not return back to the caller that it is doing so, so we get no response and have to check it some other means.

When we send out the command "mco rpc rpcutil get_fact fact=operatingsystem -I /^[Ee][Ss]/" we don't get a response back from those systems, it times out

With debugging, we do see:

warn 2015/04/03 11:24:06: client.rb:251:in `start_receiver' Could not receive all responses. Expected : 1. Received : 0

We are using ActiveMQ connector and earlier in March, we saw a lot of our CentOS machines bomb out of mCollective as it could not find the Activemq.ddl until I made a symbolic link to the connectors to the location was searching.

I am at a loss to the issue we are having now and wondering if anyone else has ideas why mcollective broke for linux. We do see our windows mcollective with no issue however.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

answered 2015-04-09 02:00:40 -0500

SKasai gravatar image

Finally found the issue. Apparently, we use cfengine to do linux configuration and was were putting in a custom config.rb which looks like it was a tweak done by our group to get around an issue we were seeing with previous versions of mcollective-common.

The 2.8.1 release has a slightly improved version of the config.rb or refers to a newer version that did not go over well with our tweak. This tweak also corrupted the pathing which causes mcollective to not find the activemq.ddl in its proper place. One of the other tweaks we did to make it 'work' was to put in a symbolic link to allow mcollective to start, but this turns out to break the rest of the discovery system.

We got it working now

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower

Stats

Asked: 2015-04-03 13:26:29 -0500

Seen: 485 times

Last updated: Apr 09 '15