Apparent problem with resource ordering using Puppetlabs Corosync module (does not bring up HA resources in correct order when configured to do so)

[SirDiesALot]

I'm setting up a simple application management with corosync via the puppetlabs module. 

What SHOULD happen is: 

1. Ip gets brought up

2. jboss gets brought up

I have defined the resources like this: 

    cs_primitive { 'applayer_ip':

      primitive_class     => 'ocf',

      primitive_type      => 'IPaddr2',

      provided_by         => 'heartbeat',

      parameters          => {

        'ip' => "$servicelayer::jbossip",

        'cidr_netmask' => "$servicelayer::jbossmask",

      },

      operations          => { 'monitor' => { 'interval' => '10s' } },

      ensure              => 'present',

    }

    cs_primitive { 'applayer_jboss':

      primitive_class     => 'lsb',

      primitive_type      => 'jboss',

      provided_by         => 'heartbeat',

      operations          => {

        'monitor'         => { 'interval' => '1min', 'timeout' => '30s' },

        'start'           => { 'interval' => '30s', 'timeout' => '30s', 'on-fail' => 'restart' },

        'stop'            => { 'interval' => '30s', 'timeout' => '30s', 'on-fail' => 'restart' },

      },

      require             => Cs_primitive['applayer_ip'],

      metadata            => {'target-role' => "Started"},

      ensure              => present,

    }

    cs_colocation { 'jboss_with_ip':

      primitives          => [ 'applayer_ip', 'applayer_jboss' ],

    }

    cs_order { 'jboss_ip_before_jboss':

    first   => 'applayer_ip',

    second  => 'applayer_jboss',

    require => Cs_colocation['jboss_with_ip'],

    }

The puppet run succeeds and appears to apply everything without any errors. 

However, when I check the status of the cluster, jboss is up and running while the IP is not: 

============

Last updated: Fri Nov 30 21:46:51 2012

Last change: Fri Nov 30 21:28:41 2012 via cibadmin on test1.prod1

Stack: openais

Current DC: test1.prod1 - partition WITHOUT quorum

Version: 1.1.6-3.el6-a02c0f19a00c1eb2527ad38f146ebc0834814558

1 Nodes configured, 2 expected votes

1 Resources configured.

============

Online: [ test1.prod1 ]

applayer_jboss  (lsb:jboss):    Started test1.prod1 (unmanaged) FAILED

Failed actions:

    applayer_jboss_stop_0 (node=test1.prod1, call=4, rc=1, status=complete): unknown error

Unfortunately I am not fantastically experienced with either puppet or corosync, so it's entirely possible that I'm doing something silly - but if I manually add the IP on the command line, corosync brings it up and manages it normally (with the same arguments exactly). So it seems like the module is doing something funny here. 

Help?

[Nick Cammorato]

Which puppetlabs-corosync are you using?  There are 19 forks of it last I checked(including one by me for my own needs surrounding DRBD which is now horribly out of date).

I would change this:

 cs_order { 'jboss_ip_before_jboss':

    first   => 'applayer_ip',

    second  => 'applayer_jboss',

    require => Cs_colocation['jboss_with_ip'],

    }

to

 cs_order { 'jboss_ip_before_jboss':

    first   => 'applayer_ip',

    second  => 'applayer_jboss',

    score => 'INFINITY',

    require => Cs_primitive['applayer_ip','applayer_jboss'],

    }

and 

cs_colocation { 'jboss_with_ip':

      primitives          => [ 'applayer_ip', 'applayer_jboss' ],

      require => Cs_primitive['applayer_ip','applayer_jboss'],

    }

as autorequires were very, very dicey when I mucked with the module.  Ignore this if they work now.

Also - make sure you're not setting up the IP or the service anywhere else as corosync should be managing it.

Outside of puppet, are you sure IPaddr2 is the right primitive_class?  Can we get the output of a crm configure show, and if it's missing the primitive if it takes:

crm configure primitive applayer_ip ocf:heartbeat:IPaddr2 params ip="whatever" cidr_netmask="whatever"

because I suspect that's the problem.

[SirDiesALot]

Hi, Nick - 

Thanks for the response!

Those changes do appear to have made the manifest apply and work correctly.  Now I just have to figure out why corosync can't get jboss to stop once it has been started (I can't either, 'service jboss stop' seems to hang indefinitely). 

On to the next challenge...