Problem with Cluster_config.com VMS V7.3-2

[Ajay Behari]

It seems LAN is not being recognised as a cluster interconnect by cluster_config.com.
While adding a node to an existing 6 nodes cluster, I ran into the error as indicated in the attached document.

What am I doing wrong?
All help is very much appreciated.
Ajay.

[Karl Rohwedder]

Ajya,

try again and specify Y as answer for the 'satellite' question. You specified no shared scsi/fibr and no satellite, so the routine assume some older interconnect (CI/DSSI).

regards Kalle

[Ajay Behari]

Karl,
I didn't specify satellite as I was not adding a satellite but a full node member.
Maybe, I am confused. I will give it a try.

Thanks,
Ajay.

[Volker Halle]

Ajay,

what do you want to do ? What's your current cluster config ?

You are about to configure new node NEPTUN on the current system's system disk. How is that node going to access that disk, if there is no shared SCSI/FC and no CI/DSSI and that new node is not a satellite ?

Are you trying to configure a new node with it's own system disk and only connected via the LAN ? Then you would have to run @CLUSTER_CONFIG on that nodes system disk and just make that node a cluster member.

Volker.

[Ajay Behari]

Volker,

It is a LAN cluster I am told.
I checked to see NISCS_LOAD_PEA0 parameter set to 1.

I want to add another AXP system as another full fledged member of the cluster.

Ajay.

[Ajay Behari]

It's a cluster with a common system disk.
All 6 nodes booting from the same disk.

[Volker Halle]

Ajay,

please have a look into the OpenVMS Cluster manuals first and make yourself familiar with your current configuration.

You can find the following manuals:

- Guidelines for OpenVMS Cluster Configurations
- OpenVMS Cluster Systems

at

http://h71000.www7.hp.com/doc/os83_index.html

Volker.

[Volker Halle]

Ajay,

how do those 6 existing systems access the common system disk ? Which IO subsystem ?

Volker.

[Hoff]

If the new node is joining the cluster and is reaching the (shared) system disk using a LAN connection (and not accessing its disk via CI or DSSI, nor by direct shared SCSI access), then you are adding a satellite node.

There's no such thing as a "full member" of a cluster, because there's no concept of a "partial member". A node is either in the cluster, or it is not in the cluster.

As Volker says, the cluster manuals are invaluable here, and can help you avoid stomping on your data.

One cluster configuration error seen fairly regularly involves incorrect or "creative" settings for VOTES and EXPECTED_VOTES, and I've written up an introduction here: http://64.223.189.234/node/153
Incorrect and "creative" settings don't do what you want, and can open your disk data to corruption.

Stephen Hoffman
HoffmanLabs LLC

[Dean McGorrill]

Hi Ajay,
if I'm reading this right (since you selected N for a satellite), you have a
node with its own system disk that you wish to join in to an existing cluster via LAN?
I've tended to do this by hand. for
various reasons I've put systems together for file xfers etc. On your new node to
add, in sysgen set

NISCS_LOAD_PEA0 = 1
MSCP_LOAD = 1
MSCP_SERVE_ALL = 2
ALLOCLASS = 0
VAXCLUSTER = 2
VOTES = 0

set these below to what the other systems have (if any on quorum)

EXPECTED_VOTES
RECNXINTERVAL
DISK_QUORUM
QDSKVOTES
QDSKINTERVAL

copy from the existing system, sys$system:cluster_authorize.dat to the
new system sys$system.

I think thats all, doing this from memory. reboot and you have a new cluster member.
Remember your running with individual files,
you may wish to share with the existing cluster. theres threads in here to help
you do that, ie sysuaf rightslist etc. hope this helps -Dean