Koen Verwimp wrote: > Hi! > > > > I have defined a resources called rg_alfresco_ip . This resource consists of > a OCF script (AlfrescoIP). This is script is a copy of IPAddr but with a > customized status/monitoring procedure. > > > > <group id=" rg_alfresco_ip "> > > <primitive class="ocf" type="AlfrescoIP" provider="heartbeat" id="ip_AS"> > > <meta_attributes> > > <attributes> > > <nvpair id="Alfresco_ip_meta2" name="resource_failure_stickiness" > value="-INFINITY"/> > > </attributes> > > </meta_attributes> > > <instance_attributes id="ip_alfresco_attributes"> > > <attributes> > > <nvpair id="IPaddr2_AS_pair1" name="ip" value="192.168.103.52"/> > > <nvpair id="IPaddr2_AS_pair2" name="nic" value="eth0"/> > > <nvpair id="IPaddr2_AS_pair3" name="iflabel" value="VIP_AS"/> > > <nvpair id="IPaddr2_AS_pair4" name="broadcast" > value="192.68.103.255"/> > > <nvpair id="IPaddr2_AS_pair5" name="cird_netmask" > value="255.255.255.0"/> > > </attributes> > > </instance_attributes> > > <operations> > > <op id="alfresco_ip_op1" name="start" timeout="60s" prereq="fencing" > on_fail="restart"/> > > <op id="alfresco_ip_op2" name="stop" timeout="60s" on_fail="fence"/> > > <op id="alfresco_ip_op3" start_delay="30s" name="monitor" > interval="30s" timeout="10s" on_fail="restart"/> > > </operations> > > </primitive> > > </group> > > > > > > Because I have configured a constraint (see below), rg_alfresco_ip will be > started on his default node (DocuCluster03). > > > > <rule id="prefered_location_alfresco" score="100"> > > <expression attribute="#uname" id="prefered_location_alfresco_expr" > operation="eq" value="DocuCluster03" /> > > </rule> > > > > You can see the monitor operation defined the resource group above. If the > monitor status returns a fail code, the on_fail procedure is set on > “restart”. Because the resource_failure_stickiness is equal to –INFINITY, the > resource will be restarted/fail over on my second server (DocuCluster04). > > > > Problem: > > If rg_alfresco_ip also fails on the second node (DocuCluster04), it don’t try > to fail_over on node1 (DocuCluster03) again. > > > > Anyone an idea why Heartbeat doesn’t try to fail over back on node1 > (DocuCluster03)?
The score for both nodes is negative, ie the resource cannot run there. End of story. > Do I have to reset some error counter on node1? Yes. Try crm_failcount --help And - since you're speaking of resource failure stickiness - please consider upgrading to pacemaker. Regards Dominik _______________________________________________ Linux-HA mailing list [email protected] http://lists.linux-ha.org/mailman/listinfo/linux-ha See also: http://linux-ha.org/ReportingProblems
