[Pacemaker] configuring pingd and hbaping with CentOS 5's heartbeat rpms.

Dejan Muhamedagic dejanmm at fastmail.fm
Fri Jun 19 13:17:51 UTC 2009


Hi,

On Thu, Jun 18, 2009 at 04:45:29PM -0600, Lundgren, Andrew wrote:
> > 
> > Are you sure that the attribute is called "pingd"? You can check
> > in the CIB status section what is being set.
> > 
> Thanks for your reply
> 
> Based on your input I looked at the output in the cibstatus.
> To make sure I was looking in the right place, I downed the
> interface, did a query and then brought it back up and did
> another query.  The following is the output:
> 
> -- down if--
> #cibadmin -Q > 6
> -- up if
> #cibadmin -Q > 7
> # diff 6 7
> --snip--
> <              <nvpair id="status-3c4db0b2-54e9-4dc5-9b9e-493b90866836-pingd-gateway" name="pingd-gateway" value="0"/>
> ---
> >              <nvpair id="status-3c4db0b2-54e9-4dc5-9b9e-493b90866836-pingd-gateway" name="pingd-gateway" value="1000"/>
> 
> So based on that I think the attribute is called pingd-gateway.  No?

Right.

> I went back in and changed my XML to look like the following:
> 
> ...
> 
>        <rsc_location rsc="Filesystem_mount" id="MDS-no-connectivity">
>          <rule score="-INFINITY" id="pingd-exclude-rule">
>            <expression operation="not_defined" id="pingd-exclude" attribute="pingd-gateway"/>
>          </rule>
>        </rsc_location>
>        <rsc_location id="rsc_location_Filesystem_mount" rsc="Filesystem_mount">
>          <rule id="prefered_location_Filesystem_mount" score="100">
>            <expression attribute="#uname" id="prefered_location_Filesystem_mount_expr" operation="eq" value="hostA"/>
>          </rule>
>        </rsc_location>

I'm not 100% sure, but I think you should put the two rules
together, in a single rsc_location. Then they should be correctly
evaluated and scores assigned.

> The following remain unchanged:
> 
>        <primitive class="ocf" provider="heartbeat" type="Filesystem" id="Filesystem_mount">
>          <operations>
>            <op name="monitor" timeout="60s" start_delay="0" disabled="false" role="Started" id="Filesystem_mount_mon" interval="120s" on_fail="restart"/>
>          </operations>
>          <instance_attributes id="Filesystem_mount_inst_attr">
>            <attributes>
>              <nvpair id="Filesystem_mount_attr_0" name="device" value="/dev/ExHwRaid10VolGroup/mdt"/>
>              <nvpair id="Filesystem_mount_attr_1" name="directory" value="/mdt"/>
>              <nvpair id="Filesystem_mount_attr_2" name="fstype" value="ext3"/>
>            </attributes>
>          </instance_attribues>
>          <meta_attributes id="Filesystem_mount_meta_attrs">
>            <attributes>
>              <nvpair id="Filesystem_mount_metaattr_resource_stickiness" name="resource_stickiness" value="INFINITY"/>
>              <nvpair id="Filesystem_mount_metaattr_description" name="description" value="MDT/MGS"/>
>              <nvpair id="Filesystem_mount_metaattr_target_role" name="target_role" value="started"/>
>            </attributes>
>          </meta_attributes>
>        </primitive>
> 
> ...
> 
>        <clone id="pingd-clone">
>          <primitive id="pingd" provider="heartbeat" class="ocf" type="pingd">
>            <instance_attributes id="pingd-attrs">
>              <attributes>
>                <nvpair id="pingd-name" name="name" value="pingd-gateway"/>
>                <nvpair id="pingd-dampen" name="dampen" value="5s"/>
>                <nvpair id="pingd-multiplier" name="multiplier" value="1000"/>
>                <nvpair id="pingd-hosts" name="host_list" value="192.168.1.1"/>
>              </attributes>
>            </instance_attributes>
>          </primitive>
>        </clone>
> ...
> 
> When I start and stop the interface, the status change is
> detected and recorded, but my resource still doesn't move.

It should.

> I feel like I am missing the connection between the ping and
> the resource, but I don't know what is missing.  The constraint
> looks correct to me.

So, try to combine them, i.e. just list rules in one constraint.
If that doesn't help, please open a bugzilla.

Thanks,

Dejan

> 
> Thank you.
> 
> --
> Andrew
> 
> 
> 
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker




More information about the Pacemaker mailing list