topic Re: cmviewcl problem in Operating System - HP-UX

cmviewcl problem

harsr — Wed, 02 Aug 2006 06:59:43 GMT

The cmviewcl -v command shows the primary cluster node status as down and failed. I have rlogin remote shell from the other node all are working. Is this any access or name resolution issue

Can anyone please suggest on this

Re: cmviewcl problem

RAC_1 — Wed, 02 Aug 2006 07:05:08 GMT

Are all packages running? What does following say? and do you see any message in package logs files and in syslog.log?

cmviewcl -v -l node

Re: cmviewcl problem

freddy_21 — Wed, 02 Aug 2006 07:18:32 GMT

check your server uptime
# uptime

check file /etc/rc.config.d/cmcluster
AUTOSTART_CMCLD= ----> 0 or 1

thanks
freddy

Re: cmviewcl problem

huruilin — Wed, 02 Aug 2006 09:55:23 GMT

hi harsr:
i think you'd better ues then cmgetconf -c command to get the current ascii file of MCSG..then check it

Re: cmviewcl problem

Mel Burslan — Wed, 02 Aug 2006 09:59:53 GMT

cluster node down/failed status in the cmviewcl output does not necessarily mean that the node is actually down. It means the cluster daemon is not running on this node.

to make it run, you can run the command

cmrunnode

but prior to doing this, you need to investigate the syslog to see why the node had failed. If you had recently reboot the server and cluster services does not automatically set to start, this might be a reason why but I suggest investigating it otherwise.

Re: cmviewcl problem

Stephen Doud — Wed, 02 Aug 2006 10:40:04 GMT

I agree with Mel,
the status refers to the cluster daemons, not the server run-time state.

Inspect /var/adm/syslog/syslog.log or OLDsyslog.log to see what they show for cluster daemon activity

Re: cmviewcl problem

harsr — Wed, 02 Aug 2006 23:13:43 GMT

Thanks for the suggestions actually cmcld daemon was not started and configured it for auto start

When packet switching i also got the following error

Request from root on node to start the cluster on this node failed: not authorised

And no cmcluster commands are not working and also the database got hung and DBA was not able to connect to the listener. But I was halt the package from the secondary node.I have also checked the hosts and resolv. conf for hostname resolution problems but eveything seems ok can anyone please suggest on this.

Re: cmviewcl problem

Thomas J. Harrold — Thu, 03 Aug 2006 07:56:33 GMT

Make sure you have a valid /etc/cmcluster/cmclnodelist, OR use the built-in cluster security. (can be added at package level, or at a cluster level)

The comments in the cluster ascii file do a great job of documenting how to implement.

If you want some examples, let us know.

-tjh

Re: cmviewcl problem

Steven E. Protter — Thu, 03 Aug 2006 08:19:02 GMT

Shalom,

The cluster daemon needs to be running on both nodes for a node to issue commands to the cluster.

The most likely cause of the original problem is a network configuration or security issue.

Its best with 11.15 SG and above to migrate to the cmnodelist security configuration.

The inetd.conf configuration needs these lines to function correctly:
hacl-cfg dgram udp wait root /usr/lbin/cmclconfd cmclconfd -p
hacl-cfg stream tcp nowait root /usr/lbin/cmclconfd cmclconfd -c -i
hacl-probe stream tcp nowait root /opt/cmom/lbin/cmomd /opt/cmom/lbin/cmo
md -f /var/opt/cmom/cmomd.log -r /var/opt/cmom

Make sure there is no binary configuration file sitting in the /etc/cmcluster directory of the system.

Also see that your configuration files are consisten on both nodes.

SEP

Re: cmviewcl problem

Stephen Doud — Thu, 03 Aug 2006 08:31:17 GMT

As of October 2004, patches to Serviceguard and versions of Serviceguard introduced after that date incorporate greater security measures. Therefore, configure your cluster nodes according to this document:

Editing Security Files for Serviceguard, March 2005 --> http://docs.hp.com/en/6283/SGsecurityfiles.pdf

Check the 'identd' line in /etc/inetd.conf
It should look like this...
UX11i and earlier:
ident stream tcp wait bin /usr/lbin/identd identd

UX11iV2:
auth stream tcp6 wait bin /usr/lbin/identd identd

Also, check the syslog.log - there may be a line indicating that the cluster binary file (/etc/cmcluster/cmclconfig) is corrupt:

cmclconfd[14753]: Configuration file
$SGCONF/cmclconfig from local node is corrupt.

The cksum of the file should be the same on each node. If it isn't, copy the file from a node that operates Serviceguard properly.

List ALL nodes in /etc/cmclcluster/cmclnodelist - on each node.
SG versions A.11.16 and later use the file only for initial configuration of the cluster.
Earlier versions of SG use it with every SG command.