Hi,
I have a problem with RHCS 4 in two node configuration (wayne and
eastwood).
Service ucpgw is running on wayne and httpd on eastwood.
Every node is a Sun V40z server, fencing is done by IPMI.
During cluster tests I unplug both power cables from one server (wayne),
thus
simulating unexpected poweroff (IPMI interface is also unavailable while
server
is out of power). Unfortunately cluster (with only one node active) is
not able to
start the missing service. Node wayne is able to start both services
only after a
reboot.
Clustat info before tests:
Member Status: Quorate
Member Name Status
------ ---- ------
wayne Online, rgmanager
eastwood Online, Local, rgmanager
Service Name Owner (Last) State
------- ---- ----- ------ -----
ucpgw wayne started
httpd eastwood started
Clustat info after tests (clustat waits about 10 seconds on a timeout
before
displaying this message):
Timed out waiting for a response from Resource Group Manager
Member Status: Quorate
Member Name Status
------ ---- ------
wayne Offline
eastwood Online, Local, rgmanager
Here are the logs from running node:
May 29 08:08:24 eastwood clurgmgrd: [3252]: <info> Executing
/etc/init.d/httpd status
May 29 08:08:37 eastwood kernel: CMAN: removing node wayne from the
cluster : Missed too many heartbeats
May 29 08:09:01 eastwood crond(pam_unix)[19600]: session opened for user
root by (uid=0)
May 29 08:09:01 eastwood crond(pam_unix)[19600]: session closed for user
root
May 29 08:10:01 eastwood crond(pam_unix)[19610]: session opened for user
root by (uid=0)
May 29 08:10:01 eastwood crond(pam_unix)[19610]: session closed for user
root
May 29 08:10:07 eastwood fenced[2854]: wayne not a cluster member after
90 sec post_fail_delay
May 29 08:10:07 eastwood fenced[2854]: fencing node "wayne"
May 29 08:11:01 eastwood crond(pam_unix)[19623]: session opened for user
root by (uid=0)
May 29 08:11:01 eastwood crond(pam_unix)[19623]: session closed for user
root
May 29 08:11:51 eastwood ntpd[2895]: can't open /var/ntp/ntp.drift.TEMP:
No such file or directory
May 29 08:12:01 eastwood crond(pam_unix)[19640]: session opened for user
root by (uid=0)
May 29 08:12:01 eastwood crond(pam_unix)[19640]: session closed for user
root
May 29 08:12:27 eastwood fenced[2854]: agent "fence_ipmilan" reports:
Rebooting machine @ IPMI:10.12.83.177...ipmilan: Failed to connect after
30 seconds Failed
May 29 08:12:27 eastwood fenced[2854]: fence "wayne" failed
May 29 08:12:32 eastwood fenced[2854]: fencing node "wayne"
May 29 08:12:32 eastwood ccsd[2751]: process_get: Invalid connection
descriptor received.
May 29 08:12:32 eastwood ccsd[2751]: Error while processing get: Invalid
request descriptor
May 29 08:12:32 eastwood fenced[2854]: fence "wayne" failed
May 29 08:12:37 eastwood fenced[2854]: fencing node "wayne"
May 29 08:12:37 eastwood ccsd[2751]: process_get: Invalid connection
descriptor received.
May 29 08:12:37 eastwood ccsd[2751]: Error while processing get: Invalid
request descriptor
May 29 08:12:37 eastwood fenced[2854]: fence "wayne" failed
May 29 08:12:42 eastwood fenced[2854]: fencing node "wayne"
May 29 08:12:42 eastwood ccsd[2751]: process_get: Invalid connection
descriptor received.
May 29 08:12:42 eastwood ccsd[2751]: Error while processing get: Invalid
request descriptor
May 29 08:12:42 eastwood fenced[2854]: fence "wayne" failed
May 29 08:12:47 eastwood fenced[2854]: fencing node "wayne"
May 29 08:12:47 eastwood ccsd[2751]: process_get: Invalid connection
descriptor received.
May 29 08:12:47 eastwood ccsd[2751]: Error while processing get: Invalid
request descriptor
May 29 08:12:47 eastwood fenced[2854]: fence "wayne" failed
May 29 08:12:52 eastwood fenced[2854]: fencing node "wayne"
May 29 08:12:52 eastwood ccsd[2751]: process_get: Invalid connection
descriptor received.
May 29 08:12:52 eastwood ccsd[2751]: Error while processing get: Invalid
request descriptor
May 29 08:12:52 eastwood fenced[2854]: fence "wayne" failed
May 29 08:12:57 eastwood fenced[2854]: fencing node "wayne"
May 29 08:12:57 eastwood ccsd[2751]: process_get: Invalid connection
descriptor received.
May 29 08:12:57 eastwood ccsd[2751]: Error while processing get: Invalid
request descriptor
Here's my cluster config:
<?xml version="1.0"?>
<cluster config_version="75" name="ucp_cluster">
<fence_daemon post_fail_delay="90" post_join_delay="30"/>
<clusternodes>
<clusternode name="wayne" votes="1">
<multicast addr="224.0.0.1" interface="bond0"/>
<fence>
<method name="1">
<device name="ipmi-wayne"/>
</method>
</fence>
</clusternode>
<clusternode name="eastwood" votes="1">
<multicast addr="224.0.0.1" interface="bond0"/>
<fence>
<method name="1">
<device name="ipmi-eastwood"/>
</method>
</fence>
</clusternode>
</clusternodes>
<cman expected_votes="1" two_node="1">
<multicast addr="224.0.0.1"/>
</cman>
<fencedevices>
<fencedevice agent="fence_ipmilan" auth="none"
ipaddr="10.12.83.176" login="""" name="ipmi-eastwood"
passwd="**********"/>
<fencedevice agent="fence_ipmilan" auth="none"
ipaddr="10.12.83.177" login="""" name="ipmi-wayne"
passwd="**********"/>
</fencedevices>
<rm>
<failoverdomains>
<failoverdomain name="ucp" ordered="1"
restricted="1">
<failoverdomainnode name="wayne"
priority="1"/>
<failoverdomainnode name="eastwood"
priority="1"/>
</failoverdomain>
</failoverdomains>
<resources>
<ip address="10.12.17.135" interface="bond0"
monitor_link="0"/>
<ip address="10.12.17.136" monitor_link="1"/>
<fs device="/dev/ucpsmslogvg/ucpsmslog"
force_fsck="0" force_unmount="1" fsid="38769" fstype="ext3"
mountpoint="/ucpsmslog" name="ucpsmslog" options="" self_fence="1"/>
<fs device="/dev/ucpgwlogvg/ucpgwlog"
force_fsck="0" force_unmount="1" fsid="39307" fstype="ext3"
mountpoint="/ucpgwlog" name="ucpgwlog" options="" self_fence="1"/>
<script file="/home/ucpgw/bin/ucpgw"
name="ucpgw"/>
<script file="/etc/init.d/httpd" name="httpd"/>
</resources>
<service autostart="1" domain="ucp" name="ucpgw"
recovery="relocate">
<script ref="ucpgw"/>
<fs ref="ucpgwlog"/>
<ip ref="10.12.17.136"/>
</service>
<service autostart="1" domain="ucp" name="httpd"
recovery="relocate">
<ip ref="10.12.17.135"/>
<script ref="httpd"/>
<fs ref="ucpsmslog"/>
</service>
</rm>
</cluster>
Is this cluster misconfigured or is it a bug in fenced/ccsd subsystem?
How can I solve this problem?
Regards,
Tomasz Koczorowski
--
Linux-cluster@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/linux-cluster
