Cluster Removal in VMS7.2-1

[Learn_1]

I have a system Alpha 4100A running vms7.2-1 in cluster mode.the system and quorum disk are on the shared storage (RA3000).one of the member is no more running and only one system is running.now i want to move the system disk on the local storage and want to remove cluster configuration so that the system keep functiong from local disk (as i will be moving system disk to local storage)without the shared storage as i will be using this shared storage with other system.
i will have to remove the mount opoints that are configured to mount data disks of RA3000 at startup.
how can i accomplish all this smoothly.
Thanks.

[Martin P.J. Zinser]

Well, the easiest solution is to just lower the quorum for the cluster to 1 giving your node 1 vote. Booting form the local disk is just achieved by setting the boot device in the console mode (after you did prepare a local disk with OpenVMS on it ;-)

[Mike Naime]

What Martin replied with will indeed work, you wind up with one cluster member.

If you really want to uncluster thsi node
When you move the system disk from the shared storage to the local disk. You will need to Set VAXCLUSTER = 0 in your modparams.dat file. Run AUTOGEN to implement this change and then re-boot your system. You should now be unclustered on this node.

Note #1: you may want to uncluster the other node as well.

Note#2: Clustering the nodes back together will require aditional re-boots.

[Åge Rønning]

In addition to setting VAXCLUSTER = 0, it's also recomended to remove (if they exist) NISCS_LOAD_PEA0, NISCS_CONV_BOOT, MSCP_LOAD, TMSCP_LOAD from MODPARAMS.DAT.
DISK_QUORUM could also be removed.

[Martin P.J. Zinser]

Yup, my approach is minimalistic.

Advantages:
*Easy ;-)
*You can re-establish the cluster at any time

Disadvantages:
*There is a short delay during boot while the system tries to figure out if there is already an exisiting cluster it could join.
*You continue to need the cluster license on this node.

[Veli Körkkö]

There are additional gotchas too. Some software, e.q. ADVANCED SERVER forces cluster alias setup and also TCPIP stack may have some of services configured in the "cluster common".

See e.q. $TCPIP SHOW CONF ENABLE SERVICE
and $TCPIP SHOW CONF ENABLE SERVICE /COMMON

And on a cluster some licenses have been bound to actual node name, on a non-clustered node this is not needed.

So, going from multinode cluster to a single system cluster is sort of sure approach. The approach of really unclustering the system offers some benefits like maybe 1 minute quicker boot but you may have some additional unexpected work ahead.

_veli

[Willem Grooters]

Martin,
just a small question:

you said:

*You continue to need the cluster license on this node.

what would happen if you removed the license? My idea would be that all custer-related components would fail to start but you would be able to run the rest of VMS - standalone of course. But am I wrong there?

[Lokesh_2]

in additions can use sys$manager:cluster_config.com ( cluster_config_lan.com ) to remove the root of another node.

For check the startup files to remove the mount points of data disks

Best regards,
Lokesh

[Learn_1]

Thanks Guys,
According to your suggestions i ahve modified the the modparams.dat.infactthe original system disk was on the shared storage and the image backup of the system disk was on the local storage.i have disconnected the shared storage and after booting from OS cd and mounting the local system disk modified the MODparams.dat file.the values that i have modified are as under.
1.remarked NISCS_LOAD_PEA0=1
2.VAXCLUSTER=0
3.remarked disk_quorum="$1$dka301" (this was the system disk on the shared storage).
4.remarked QDSKVOTES
5.REMARKED INTERCONNECT="N1"
6.Remarked MSCP_LOAd=1
7.Remarked MSCP_SERVE_ALL=2
8.remarked MC_SERVICES_P2=1
9.Remarked interconnect ="NIMC"

after doing this when i do
RUN SYS$SYSTEM:SYSMAN
SYSMAN>DO @SYS$UPDATE:AUTOGEN GETDATA SETPARAMS
The system just keep prompting but no result.
similarly when i boot from the local disk the system stops at
"AITOMATIC DISK SERVING"

What can i do to configure this node so that this node (this is the only node in the cluster configuration as other node is already down)boot from local system disk.
how can cluster_configure.com can help to resolve this issue.

Your urgent reply innthis regard will help to resolve this issue asap.

[Learn_1]

A bit clarification.
if i boot the system from local system disk it just stops at
"MSCPLOAD-I-CONFIGSCAN,Enabled automatic disk serving.
The system just stops at this point and does not go any further.