<div dir="ltr"><div><div><div><div>Hi All,<br><br></div>This might be a basic question but I&#39;m not sure whats taking time for failover switching.<br></div><div>Hope anyone can figure it out.<br><br></div>Scenario - <br>
Pacemaker running DRBD(Dual primary mode)+OCFS2+XEN for Virtual windows machine<br><br></div>Pacemaker startup starts - <br>DRBD -&gt; OCFS2 -&gt; XEN<br></div>Lets consider under Server1  - DRBD, OCFS2(clone) and XEN are started<br>
<br>Server2 - DRBD, OCFS2(clone) are started<br><br><div>Now if Server1 power is OFF<br><br></div><div>The XEN resource which was running under Server1 should be failed over to Server2.<br></div><div><br>In my case, its taking almost 90 to 110 seconds to do this.<br>
<br></div><div>Can anyone suggest me ways to reduce it to within 30 to 40 seconds<br><br></div><div>My pacemaker configuration is -<br>crm configure
<br>property no-quorum-policy=ignore
<br>property stonith-enabled=false
<br>property default-resource-stickiness=1000
<br><br>primitive resDRBDr1 ocf:linbit:drbd \
<br>params drbd_resource=&quot;r0&quot; \
<br>op start interval=&quot;0&quot; timeout=&quot;240s&quot; \
<br>op stop interval=&quot;0&quot; timeout=&quot;100s&quot; \
<br>op monitor interval=&quot;20s&quot; role=&quot;Master&quot; timeout=&quot;240s&quot; \
<br>op monitor interval=&quot;30s&quot; role=&quot;Slave&quot; timeout=&quot;240s&quot; \
<br>meta migration-threshold=&quot;3&quot; failure-timeout=&quot;60s&quot;
<br>primitive resOCFS2r1 ocf:heartbeat:Filesystem \
<br>params device=&quot;/dev/drbd/by-res/r0&quot; directory=&quot;/cluster&quot; fstype=&quot;ocfs2&quot; \
<br>op monitor interval=&quot;10s&quot; timeout=&quot;60s&quot; \
<br>op start interval=&quot;0&quot; timeout=&quot;90s&quot; \
<br>op stop interval=&quot;0&quot; timeout=&quot;60s&quot; \
<br>meta migration-threshold=&quot;3&quot; failure-timeout=&quot;60s&quot;
<br>primitive resXen1 ocf:heartbeat:Xen \
<br>params xmfile=&quot;/home/cluster/xen/win7.cfg&quot; name=&quot;xenwin7&quot; \
<br>op monitor interval=&quot;20s&quot; timeout=&quot;60s&quot; \
<br>op start interval=&quot;0&quot; timeout=&quot;90s&quot; \
<br>op stop interval=&quot;0&quot; timeout=&quot;60s&quot; \
<br>op migrate_from interval=&quot;0&quot; timeout=&quot;120s&quot; \
<br>op migrate_to interval=&quot;0&quot; timeout=&quot;120s&quot; \
<br>meta allow-migrate=&quot;true&quot; target-role=&quot;started&quot;
<br> <br>ms msDRBDr1 resDRBDr1 \
<br>meta notify=&quot;true&quot; master-max=&quot;2&quot; interleave=&quot;true&quot; target-role=&quot;Started&quot;
<br>clone cloOCFS2r1 resOCFS2r1 \
<br>meta interleave=&quot;true&quot; ordered=&quot;true&quot; target-role=&quot;Started&quot;
<br> <br>colocation colOCFS12-with-DRBDrMaster inf: cloOCFS2r1 msDRBDr1:Master
<br>colocation colXen-with-OCFSr1 inf: resXen1 cloOCFS2r1
<br>order ordDRBD-before-OCFSr1 inf: msDRBDr1:promote cloOCFS2r1:start
<br>order ordOCFS2r1-before-Xen1 inf: cloOCFS2r1:start resXen1:start
<br> <br>commit
<br>bye<br></div><div><div><div><div><div>
<div><br>-- <br>Regards,<br>Kamal Kishore B V<br></div><div><br>
</div></div></div></div></div></div></div>