[Linux-cluster] active-active httpd issue
Panigrahi, Santosh Kumar
Santosh.Panigrahi at in.unisys.com
Wed Jan 21 12:13:02 UTC 2009
Hello,
I have a 2 node HA cluster and I am configuring the httpd service as
active-active. I want to run httpd service in both the cluster nodes
simultaneously. I have configured 2 failover domains FD1 and FD2.
FD1 = p6pv1(priority 1) & p7pv1(priority 2)
FD2 = p6pv1(priority 2) & p7pv1(priority 1)
nodes(both installed RHEL5.2):
--------------------------
p6pv1: 10.1.40.204
p7pv1: 10.1.40.206
and here's the "/etc/cluster/cluster.conf":
-----------------------------------------
[root at p6pv1 ~]# cat /etc/cluster/cluster.conf
<?xml version="1.0"?>
<cluster alias="HA" config_version="17" name="HA">
<fence_daemon clean_start="0" post_fail_delay="0"
post_join_delay="3"/>
<clusternodes>
<clusternode name="p6pv1.tr.unisys.com" nodeid="1"
votes="1">
<fence>
<method name="1">
<device name="man1"/>
</method>
</fence>
</clusternode>
<clusternode name="p7pv1.tr.unisys.com" nodeid="2"
votes="1">
<fence>
<method name="1">
<device name="man2"/>
</method>
</fence>
</clusternode>
</clusternodes>
<cman expected_votes="1" two_node="1"/>
<fencedevices>
<fencedevice agent="fence_xen" domain="p6pv1" host="p6"
name="man1"/>
<fencedevice agent="fence_xen" domain="p7pv1" host="p7"
name="man2"/>
</fencedevices>
<rm>
<failoverdomains>
<failoverdomain name="FD1" nofailback="0"
ordered="1" restricted="1">
<failoverdomainnode
name="p6pv1.tr.unisys.com" priority="1"/>
<failoverdomainnode
name="p7pv1.tr.unisys.com" priority="2"/>
</failoverdomain>
<failoverdomain name="FD2" nofailback="0"
ordered="1" restricted="1">
<failoverdomainnode
name="p6pv1.tr.unisys.com" priority="2"/>
<failoverdomainnode
name="p7pv1.tr.unisys.com" priority="1"/>
</failoverdomain>
</failoverdomains>
<resources>
<ip address="10.1.40.200" monitor_link="1"/>
<ip address="10.1.40.201" monitor_link="1"/>
</resources>
<service autostart="1" domain="FD2" exclusive="0"
name="httpd2" recovery="relocate">
<ip ref="10.1.40.201">
<script file="/etc/rc.d/init.d/httpd"
name="httpd"/>
</ip>
</service>
<service autostart="1" domain="FD1" exclusive="0"
name="httpd1" recovery="relocate">
<ip ref="10.1.40.200">
<script file="/etc/rc.d/init.d/httpd"
name="httpd"/>
</ip>
</service>
</rm>
</cluster>
[root at p6pv1 ~]#
-----------------------------------------
When these two nodes are both running, I am seeing configured services
httpd1 and httpd2 are running on p6pv1 and p7pv1 respectively.
--------------------------------------------
[root at p6pv1 ~]# clustat
Cluster Status for HA @ Tue Dec 23 06:31:47 2008
Member Status: Quorate
Member Name ID
Status
------ ---- ----
------
p6pv1.tr.unisys.com 1
Online, Local, rgmanager
p7pv1.tr.unisys.com 2
Online, rgmanager
Service Name Owner
(Last) State
------- ---- -----
------ -----
service:httpd1
p6pv1.tr.unisys.com started
service:httpd2
p7pv1.tr.unisys.com started
---------------------------------------------------
But in real the httpd service is running only at one node at any
particular instant of time. Here expected behavior is httpd should run
on both the nodes. I am getting below log messages.
[root at p6pv1 ~]# !ta
tail -f /var/log/messages
Dec 23 06:06:37 p6pv1 init: Id "x" respawning too fast: disabled for 5
minutes
Dec 23 06:09:10 p6pv1 ccsd[11169]: Update of cluster.conf complete
(version 16 -> 17).
Dec 23 06:09:22 p6pv1 clurgmgrd[11282]: <notice> Reconfiguring
Dec 23 06:09:23 p6pv1 clurgmgrd[11282]: <err> Primary attribute
collision. type=script attr=name value=httpd
Dec 23 06:11:38 p6pv1 init: Id "x" respawning too fast: disabled for 5
minutes
Dec 23 06:16:39 p6pv1 init: Id "x" respawning too fast: disabled for 5
minutes
Dec 23 06:21:40 p6pv1 init: Id "x" respawning too fast: disabled for 5
minutes
Dec 23 06:26:41 p6pv1 init: Id "x" respawning too fast: disabled for 5
minutes
Dec 23 06:31:42 p6pv1 init: Id "x" respawning too fast: disabled for 5
minutes
Dec 23 06:36:43 p6pv1 init: Id "x" respawning too fast: disabled for 5
minutes
Can some one help me to fix this issue? Any suggestions appreciated.
Thanks and Regards,
Santosh
More information about the Linux-cluster
mailing list