[Linux-HA] how to restart a failed resource

[Andrew Beekhof]

On 9/28/07, dxj_600 <dxj_600 at 126.com> wrote:
> Thanks to "Andrew Beekhof" and "Joseph Lamoree", i can trigger heartbeat to start the resource now
> but I met new problem as followings:
>  1. I configured a clones resource, which should be run at two nodes
>  node host_192.168.2.220 host_192.168.2.221
>  2. first my resource run at only one node
>  # crm_resource -W -r rsc_lcm_0:0
>  resource rsc_lcm_0:0 is running on: host_192.168.2.220
>   # crm_resource -W -r rsc_lcm_0:1
>  resource rsc_lcm_0:1 is NOT running

why is this?

>  3. then i trigger the heartbeat to start rsc_lcm_0:1
>  # crm_resource -C -r rsc_lcm_0:1
>  4. then heartbeat do something like this
>  on node host_192.168.2.221

please attach the actual logs, they convey far more information than
you have reproduced here.

>  [2007-09-28 18:20]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra monitor
>  probe
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra start
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra notify
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra stop <---- because my ra timeout
>  on node host_192.168.2.220
>  [2007-09-28 18:20]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra notify
>  probe
>  [2007-09-28 18:20]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra monitor
>  [2007-09-28 18:20]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra notify
>  [2007-09-28 18:20]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra notify
>  [2007-09-28 18:20]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra stop <---- why stop me????
>  [2007-09-28 18:20]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra monitor
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra notify
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra stop -;___ in fact it's a restart
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra start -'
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra notify
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra notify
>  [2007-09-28 18:21]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra monitor
>  [2007-09-28 18:22]: /usr/lib/ocf/resource.d//heartbeat/owlcm-ra monitor
>  5. finally result is
>  # crm_resource -W -r rsc_lcm_0:0
>  resource rsc_lcm_0:0 is running on: host_192.168.2.220
>   # crm_resource -W -r rsc_lcm_0:1
>  resource rsc_lcm_0:1 is NOT running
>  My question is:
>  since the result 5 is same as result 2, rsc_lcm_0:0 is still running at node host_192.168.2.220, then WHY heartbeat stop the resource and restart it again, and WHY not heartbeat keep rsc_lcm_0:0 running untouched, how can i avoid this scenary.
>  my resource configuration is:
>  <clone id="rsc_lcm" ordered="false" interleave="false" notify="true">
>  <instance_attributes id="ow_lcm_instance_attributes">
>  <attributes>
>  <nvpair id="ow_lcm_clone_max" name="clone_max" value="2"/>
>  <nvpair id="ow_lcm_clone_node_max" name="clone_node_max" value="1"/>
>  </attributes>
>  </instance_attributes>
>  <primitive class="ocf" type="owlcm-ra" provider="heartbeat" id="rsc_lcm_0" is_managed="true">
>  <operations>
>  <op id="rsc_lcm_op0" name="start" timeout="50s" disabled="false"/>
>  <op id="rsc_lcm_op2" name="monitor" interval="5s" timeout="4s" disabled="false" role="Started"/>
>  </operations>
>  <instance_attributes id="ow_lcm_ra_instance_attributes">
>  <attributes>
>  <nvpair id="rsc_lcm_vip" name="vip" value="192.168.2.231 192.168.2.232"/>
>  </attributes>
>  </instance_attributes>
>  </primitive>
>  </clone>
>  thanks a lot
>
> $B:_(B2007-09-26$B!$(B"Joseph Lamoree" <jlamoree at gmail.com> $B<LF;!'(B
> Take a look at the crm_resource page:
> http://www.linux-ha.org/v2/AdminTools/crm_resource
> There is an example of stopping/starting resources there. Also, most
> of the executables in heartbeat support a --help option to display a
> lot of very useful information.
> --
> Joseph Lamoree
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>