[Linux-cluster] ILO fencing works but cluster service isn't relocated

yilmaz ylmzlist at gmail.com
Thu Sep 17 08:32:38 UTC 2009


Hello,
    In my HP blade test system, I have two servers (server1 and server3)
 I have tested fencing with the command:

*/sbin/fence_ilo -a server1-ilo -l admin -p adminpass -o off -v*

and the server1 has been removed from the cluster properly and services are
also relocated. I made another test
and tried removing cluster IP from the first node (server1-cluster)
manually. I saw the messages in server1 that it leaves the cluster and after
that
it shut down itself immediately. However services in server1 aren't
relocated into the server3. When I check the cluster status in server3
I saw the following output.

Service Name         Owner (Last)                   State
  ------- ----         ----- ------                   -----
 SERVER11              none                  started
  WEB11                none                  started


 There was no owner of the service but states were started.
My cluster configuration is attached below. I have redhat AS 4 (update 6)
and ilo firmware version is 1.60 Jul 11 2008
I suppose the services should be relocated into the other node but it didn't
happen.

Thanks for your comments in advance.




<?xml version="1.0" encoding="UTF-8"?>
<cluster config_version="14" name="mmsc">
  <cman expected_votes="1" two_node="1"/>
  <clusternodes>
    <clusternode votes="1" name="server1-cluster">
      <fence>
        <method name="hardware">
          <device hostname="server1-ilo" name="ilo"/>
        </method>
        <method name="last_resort">
          <device ipaddr="server1-cluster" name="last_resort"/>
        </method>
      </fence>
    </clusternode>
    <clusternode votes="1" name="server3-cluster">
      <fence>
        <method name="hardware">
          <device hostname="server3-ilo" name="ilo"/>
        </method>
        <method name="last_resort">
          <device ipaddr="server3-cluster" name="last_resort"/>
        </method>
      </fence>
    </clusternode>
  </clusternodes>
  <fencedevices>
    <fencedevice passwd="adminpass" action="off" login="admin" name="ilo"
agent="fence_ilo"/>
    <fencedevice name="last_resort" agent="fence_manual"/>
  </fencedevices>
  <rm log_facility="uucp" log_level="6">
    <failoverdomains>
      <failoverdomain restricted="1" ordered="1" name="PreferredFirst">
        <failoverdomainnode priority="0" name="server1-cluster"/>
        <failoverdomainnode priority="1" name="server3-cluster"/>
      </failoverdomain>
      <failoverdomain restricted="1" ordered="1" name="PreferredSecond">
        <failoverdomainnode priority="0" name="server3-cluster"/>
        <failoverdomainnode priority="1" name="server1-cluster"/>
      </failoverdomain>
    </failoverdomains>
    <resources/>
    <service domain="PreferredFirst" name="SERVER11" autostart="1"
recovery="relocate">
      <script name="SERVER11" file="/opt/cluster/SERVER11_service.sh"/>
      <ip monitor_link="1" address="172.16.5.227"/>
      <ip monitor_link="1" address="172.16.5.228"/>
      <ip monitor_link="1" address="172.16.5.229"/>
      <fs name="/disk1" force_unmount="1" fstype="ext3" device="/dev/sde1"
mountpoint="/disk1" options="rw,nosuid"/>
    </service>
    <service domain="PreferredFirst" name="WEB11" autostart="1"
recovery="relocate">
      <script name="WEB11" file="/opt/cluster/WEB11_service.sh"/>
      <ip monitor_link="1" address="172.16.5.230"/>
      <ip monitor_link="1" address="172.16.5.231"/>
      <fs name="/web11d" force_umount="1" fstype="ext3" device="/dev/sdf1"
mountpoint="/web11d" options="rw,nosuid"/>
    </service>
  </rm>
  <quorumd votes="2" log_level="6" tko="10" interval="1" label="priquorum"
log_facility="uucp" device="/dev/sda"/>
  <fence_daemon clean_start="1"/>
</cluster>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20090917/2e7e4484/attachment.htm>


More information about the Linux-cluster mailing list