topic Problem creating the cluster in Operating System - Linux

Problem creating the cluster

Benoît — Thu, 11 May 2006 07:35:36 GMT

Hi,

I get this message when I create the cluster :
cmapplyconf -C $SGCONF/elites.config -v

Checking cluster file: /usr/local/cmcluster/conf/elites.config
Checking nodes ... Done
Checking existing configuration ... Done
Warning: Unable to get configuration for cluster elites.
cmapplyconf : Command can not be executed remotely.
Addition of node ELITE-00 to cluster elites is possible only when
command is issued from a node which is configured to be a part of the cluster.
Please login to one of the cluster member nodes of cluster elites
and reissue the command.

This appens on the 2 nodes.

Config file :
CLUSTER_NAME elites

QS_HOST elite-99
QS_POLLING_INTERVAL 120000000
QS_TIMEOUT_EXTENSION 2000000

NODE_NAME elite-00
NETWORK_INTERFACE bond0
HEARTBEAT_IP 10.10.165.83
NETWORK_INTERFACE bond1
HEARTBEAT_IP 192.168.0.1

NODE_NAME elite-01
NETWORK_INTERFACE bond0
HEARTBEAT_IP 10.10.165.84
NETWORK_INTERFACE bond1
HEARTBEAT_IP 192.168.0.2

HEARTBEAT_INTERVAL 1000000
NODE_TIMEOUT 5000000
AUTO_START_TIMEOUT 600000000
NETWORK_POLLING_INTERVAL 2000000
MAX_CONFIGURED_PACKAGES 5

cmclnodelist:
ELITE-00 root
elite-00 root
10.10.165.83 root
192.168.0.1 root
ELITE-01 root
elite-01 root
10.10.165.84 root
192.168.0.2 root
ELITE-99 root
elite-99 root
10.10.165.86 root
192.168.0.3 root

Thanks.

Ben

Re: Problem creating the cluster

John Bigg — Thu, 11 May 2006 11:08:48 GMT

I think the problem is likely to be that you do not have the IP addresses for all interfaces configured as required by SG. You need to ensure that every IP address on every interface on every node in the cluster resolves to the hostname of that node or have this as an alias.

What I suspect is happening is that during the cmapplyconf when we switch from the bootstrap cmclnodelist file to the access control policies we suddenly do not recognise the interface the message is coming from as being within the cluster.

i.e. in your case you need to have something similar to this in /etc/hosts on all nodes:

10.10.165.83 elite-00
192.168.0.1 elite-00
10.10.165.84 elite-01
192.168.0.2 elite-01

You could have this setup as an alias if you do not want the primary lookup to be the same, i.e. you could also have the following instead:

10.10.165.83 elite-00
192.168.0.1 alt-00 elite-00
10.10.165.84 elite-01
192.168.0.2 alt-01 elite-01

if you have any lan cards with IP addresses which you have not configured in elites.config these still need to follow the same rules since these could be used for command access even if they are not configured as heartbeat lans.

I would expect that in the syslog (messages) file you will have messages indicating access is being attempted from unauthorised nodes listing either the IP addresses or the names.

If this does not resolve the problem then please paste any errors reported by cmclconfd processes from the syslog file here. However, I think this should resolve your issue.

Re: Problem creating the cluster

Benoît — Thu, 11 May 2006 13:14:05 GMT

I've found that identd was not running.
After correcting that, no problem.

Thanks for your help.