One cluster missing from MONITOR CLUSTER uility

[SANJAY MUNDHRA]

I am setting up a four node ES80 cluster. After the cluster configuration all cluster communication is ok, except that when I give Monitor/cluster utility it gives a stack dump. The Monitor/cluster utility works ok is there are three members in the cluster, I have increased the net$acpbuffer_unit, but still get the same problem. All sysman commands are getting executed on all the four nodes, also SHOW CLUSTER dislays all the cluster members.

Anyone faced this problem before.

[Karl Rohwedder]

Can you post the errormessage you get, are all nodes running the same VMS version?

regards Kalle

[John Abbott_2 · Accepted solution]

Common system disk ?

Check your DECnet configuration - $mon/clu uses the decnet vpm service. Startup file sys$startup:net$application_startup.ncl under decnet/osi or online via $mc ncl show session control application vpm all - check details on each node

Also look into sys$sysdevice:[sysn.vpm$server]net$server.log files for any errors.

Not aware of any patches in this area, what version of vms ?

Hope this helps.
Regards
John.

[SANJAY MUNDHRA]

I am using AXP-VMS 8.3. The cluster is using a common system disk on two EVA's and Volume shadow. Infact I was able to do get all the cluster members from one node , from another node only 3 members available while the third memebr was giving a bug-check. Finally fixed the problem, looks like the customer had remove the default gateway setting in TCPIP for all the nodes. Some had their TCPIP services restarted, while some were still running with old information, though all of them in the same subnet. Another change done was I had disabled SCA traffic on the IP networks. After adding back the default gateway backa nd restarting all the nodes, its working fine now.

Going again next week, will check it again.

[The Brit]

Just a comment, you mentioned that you disabled SCS communication over the IP network. I assume that you did a "stop/lan ". Rather than do this, you should consider lowering the priority on the IP networks. This will allow the cluster to survive in the event that your interconnect fails. Use

SCACP> set/lan/priority=

Dave.

[SANJAY MUNDHRA]

Thanks very much, will set a lower priorty for SCA on LLa0.

Regds,
Sanjay

[SANJAY MUNDHRA]

Thanks very much, I will close this thread as of now.