[Pacemaker] ping - resource failover

Josh josh2780 at yahoo.com
Fri Aug 10 14:30:13 UTC 2012


Hello,

I've followed various docs referring to setting up a ping primitive to monitor connectivity to, for example, a gateway.  If that ping fails, resources that rely on that ping (or rely on connectivity to the gateway) should failover to another node.  Below is my configuration.  I have a www1-vip that starts up on one of two nodes.  I can move it around the nodes just fine by either shutting down a node or manually moving it via a 'crm resource migrate'.  When I introduce a location rule, nothing happens when the ping to the gateway stops (i.e. I pull the plug on that interface).

Both nodes have 2 NICs:   eth0 is connected to the 10.0.100.0 network and eth1 is connected to another net specifically setup as a backup cluster connection (I can pull the plug on eth0 but the 2 nodes can still talk over eth1.  The 10.0.100.254 gateway, however, is not reachable from eth1).


node node1
node node2
primitive ping-gateway ocf:pacemaker:ping \
        params host_list="10.0.100.254" multiplier="100"
primitive www1-vip ocf:heartbeat:IPaddr2 \
        params ip="10.0.100.163" nic="eth0" cidr_netmask="24" iflabel="eth0" \
        op start interval="0" timeout="20" \
        op stop interval="0" timeout="20" \
        op monitor interval="10" timeout="20" start-delay="0" \
        meta resource-stickiness="50"
clone clone_ping-gateway ping-gateway \
        meta interleave="true"
location location_www1-vip www1-vip \
        rule $id="location_www1-vip-rule" pingd: defined pingd
property $id="cib-bootstrap-options" \
        dc-version="1.1.7-6.el6-148fccfd5985c5590cc601123c6c16e966b85d14" \
        cluster-infrastructure="openais" \
        expected-quorum-votes="2" \
        no-quorum-policy="ignore" \
        stonith-enabled="false"
rsc_defaults $id="rsc-options" \
        resource-stickiness="90"

If www1-vip is running on node1, if I disconnect eth0 on node1, www1-vip does not move over to node2.  I have to do a manual 'crm resource migrate www1-vip node2' to move it.

Any suggestions?  Thanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.clusterlabs.org/pipermail/pacemaker/attachments/20120810/44ed7cfe/attachment-0003.html>


More information about the Pacemaker mailing list