#mcollective

chitopunk: sorry a bit late, but for master/slave to work you need the uri to look like this: uri="masterslave:(tcp://activemq1.example.net:61616,tcp://activemq2.example.net:61616)"

the master is listed first, the slave second in BOTH configs.

without that you don't have master/slave, you only have basic peering.

Hi all. I have only set up MCollective for the first time this week. It was working perfectly until I came in today. It's installed on 24 servers but any queries return 843 responses. I've discovered that this is because Server28 is registered 821 times but I can't figure out how this happened or how to undo it. Any pointers?

stetho: Server28 will be running 821 mcollective processes, at a guess

and 'how it happened' -> I guess your init script is broken, and puppet or chef or whatever started it 821 times (as the init script's status action is broken?)

'how to undo it' -> kill the wrong prcoesses, fix the init script..

t0m: The how it happened bit makes sense except for the fact that it would have happened on the other servers too, wouldn't it? But your diagnosis is correct - I was looking at rabbitmq being the culprit, not the actual server.

depends :)

someone may have echo '' > /var/run/mcollective.pid ; chattr +i /var/run/mcollective.pid on one box (or similar)

and given a broken init script, that'd cause that behavior..

I mean, yeah, I dunno why one and not another, but it's totally possible :)

t0m: I can't see anything immediately obvious. At least I know not to blame RabbitMQ if it happens again. I'll keep digging.

[mcollective-puppet-agent] jantman opened pull request #31: MCOP-67 make runall work with compound filters (master...MCOP-67) http://git.io/Gi5TqQ

Hi, I want to know if the following scenario is feasible, I have 2 locations, i have setup an activemq cluster for location 1, I have setup an activemq single server for location 2, then all mcollective server configuration in location 1 have the following order.. plugin.activemq.1.host = location1-server1, plugin.activemq.2.host = location1-server2, plugin.activemq.3.host = location2-server1, same way for the location2 servers plugin.

then if I bring activemq down for location2, all servers connects to the location1 one, good, but when the location2 activemq retunrs, the servers are still connected to the location1 ones..

is there a way to tell the servers,.. check if your activemq.1.host is up, if so connect to it..?

like a timeout stuff

otherwise I have all my servers in location2 connected to location1 activemq