Worst possible scenario for SG,11iv2 with June 2007 patches

[S.N.S]

Hi,

I know I would probably be pulverized.However, I dont mind..

Case:
OS-11.23; MC SG 11.18; TWO 2-Node Clusters,
BUT
swlist -l bundle|grep -i qpk
QPKAPPS B.11.23.0706.064 Applications Quality Pack Bundle for HP-UX 11i v2, June 2007
QPKBASE B.11.23.0706.064 Base Quality Pack Bundle for HP-UX 11i v2, June 2007

Yes, the bundles are of June 2007.

Issue
The Primary Node of Cluster 1:
The Cluster doesnt start on booting - even when AUTOSTART_CMCLD=1

The Primary Node of Cluster 2:
cmruncl -n node1 >> "The cluster is already running"
But ofcourse, cmviewcl gives node1 > down
However, if I run the same command once more, I am able to run the cluster.

As per my understanding, this is erractic behaviour - patching needed 1000% << why not >>> Long story

1. Is my Inference Correct?
2. How much long practically can the server survive unpatched?

Point for ALL those who Help :-)!!

Merci/Danke
SNS

[njia_1]

Hi there

We have clusters running SG 11.18 and December 2007 qpk and it is running fine.

Just have few questions
Have you checked all your network connections ?, cluster lock disk ? anything in the cluster startup log file ?

after you started the cluster, is it stable ?
you said the cluster started when you run the command again, did this happen on both clusters ?

thanks

[Carsten Krege · Accepted solution]

Hi SNS,

regarding your cluster1: During boot the cluster boot scripts will execute "cmrunnode -v" when AUTOSTART_CMCLD=1 is set. This will only start the cluster when all other cluster members execute cmrunnode at the same time (e.g. boot up at the same time or someone issues cmrunnode manually on the other nodes). If this is not the case, you need cmruncl to start the cluster.

Re cluster2:
You should be extremely careful with the comamdn "cmruncl -n "! I hope you read and understood the warning being printed along with this command.
If you can confirm that before your cmruncl command the cmcld process is not running on any cluster member (this is basically the meaning of "The cluster is already running"), then the cluster should start on the node.

In any case I doubt that a big pack of patches will help you. It would be way more reasonable to try the latest SG patch for your version first and see if this helps. However, my feeling is that you don't experience a bug here.
You should really open a call with your HP Support and they should be able to explain what to do, resp. what to do better.

Best regards,

Carsten

[singh sanjeev]

man page of cmeuncl says :
This command should only be run when the cluster is not active on any
of the configured nodes. This command verifies the network
configuration before causing the nodes to start their cluster daemons.
If a cluster is already running on a subset of the nodes, the
cmrunnode command should be used to start the remaining nodes and
force them to join the existing cluster.

============
cmruncl -n node1 : if other node is already running the cluster then you should start via cmrunoder -n node1..

but you said it failed to start at boot time...please rc.log file what happen..why this whether cluser script is got excuted at time of booting or not or it got skeeped and find the reason forso,...if you find any details please let us know.