Re: strange behavior with clumanager?

This is my /etc/cluster.xml file. Also, the shared storage is a FC RAID5 with 2 controllers. Both nodes see that storage the same way.


<?xml version="1.0" ?>
<cluconfig version="3.0">
  <clumembd broadcast="no" interval="330000" loglevel="4" multicast="yes" multicast_ipaddress="225.0.0.11" thread="yes" tko_count="15"/>
  <cluquorumd loglevel="4" pinginterval="" tiebreaker_ip=""/>
  <clurmtabd loglevel="4" pollinterval="4"/>
  <clusvcmgrd loglevel="4"/>
  <clulockd loglevel="4"/>
  <cluster config_viewnumber="17" key="d3fc0d804565d54138cff39a156d6387" name="xfiler"/>
  <sharedstate driver="libsharedraw.so" rawprimary="/dev/raw/raw1" rawshadow="/dev/raw/raw2" type="raw"/>
  <members>
    <member id="0" name="rhel1" watchdog="yes">
    </member>
    <member id="1" name="rhel2" watchdog="yes"/>
  </members>
  <services>
    <service checkinterval="0" failoverdomain="None" id="0" name="nfs01" userscript="None">
      <service_ipaddresses>
        <service_ipaddress broadcast="None" id="0" ipaddress="192.168.0.200" netmask="None"/>
      </service_ipaddresses>
      <device id="0" name="/dev/sdc5" sharename="">
        <mount forceunmount="yes" fstype="ext3" mountpoint="/nfs01" options=""/>
        <nfsexport id="0" name="/nfs01">
          <client id="0" name="*" options="rw,async"/>
        </nfsexport>
      </device>
    </service>
  </services>
  <failoverdomains>
    <failoverdomain id="0" name="dom1" ordered="yes" restricted="yes">
      <failoverdomainnode id="0" name="rhel1"/>
      <failoverdomainnode id="1" name="rhel2"/>
    </failoverdomain>
    <failoverdomain id="1" name="dom2" ordered="no" restricted="no">
      <failoverdomainnode id="0" name="rhel2"/>
    </failoverdomain>
  </failoverdomains>
</cluconfig>

----- Original Message ----- 
From: "Sean Oh" <oh-/[email protected]>
To: <[email protected]>
Sent: Wednesday, September 10, 2003 11:20 PM
Subject: strange behavior with clumanager?


> The clumanager-1.1.82-1 RPM missed several utilities, such as clusvcadm, and etc. 
> 
> The new version of clumanager, compared to AS 2.1, where is tha favorite cluadm CLI util?
> 
> Also, in the new version, how do I make the preferred member?
> 
> I have been playing around with 1.1.82 and noticed very strange behavior.
> 
> ------
> members:
>   rhel1
>   rhel2
> 
> Failed domain:
> 
> Services:
> nfs01
>    192.168.0.200 <service ip address>
>    /dev/sdc5
>       /nfs01 <nfsexport>
>          * <client>
> ------
> 
> In this simple configuration, I like to have rhel1 as the prefered member and if the rhel1 is down, rhel2 takes over the service. Also, after rhel1 re-joins the cluster, rhel1 takes over the service, since rhel1 is prefered.
> 
> But, what is happending is that when rhel1 is down( I did /etc/init.d/clumanager stop in rhel1), rhel2 is auto-rebooted and service is not available at all.
> 
> I tried with a few things with failed domain, but it did not help..
> 
> Also, how do I configure additional heartbeat channel in the new clumanager?
> 
> BTW, the cluster adminiration doc need to updated. it is out-dated and does not help much.
> 
> Appreciated any help.
> 
> 
> 
> 
> --
> Taroon-beta-list mailing list
> [email protected]
> http://www.redhat.com/mailman/listinfo/taroon-beta-list
> 
> 


--
Taroon-beta-list mailing list
[email protected]
http://www.redhat.com/mailman/listinfo/taroon-beta-list