cancel
Showing results for 
Search instead for 
Did you mean: 

VCS 2.0 for Windows 2000


Hi
new here to VCS forums
and in urgent need of a VCS expert to provide some much needed guidance on the following

VCS 2.0 Cluster for Windows on Win2k SP4 system with VVR enabled across 2 sites for replication of application DB.

I am after the correct procedure to shut down a VCS Cluster in a graceful manner. What is the impact of just invoking the shutdown/restart in windows
and how does it affect the cluster?

do i have to freeze/stop the cluster before rebooting both nodes?
There's not much help within the VCS manuals that i have.

your input will be greatly appreciated

Regards,

deemy.
2 Replies

Hello Deemy, I always prefer

Hello Deemy,

I always prefer following:

a) Offline /switch the service groups if you want to shutdown one/some of nodes of cluster

Above step gives you much granular control over the situation. When you mention that you have VVR, is VVR configured in VCS as Global Cluster ?

Also, is it VCS or VCS One as 2.0 seems to be pretty old for VCS...

In case you directly shutdown/reboot the boxes, impacts might be different, VCS would take it as failure & will evacuate service groups to other nodes depending on the policies you have set.... In case of reboot, it might be harmful if you are using IOFencing since not all the modules will be shutdown gracefully which might leave some stale keys on the disks...

I am not a windows expert, more in unix, but above considerations are very generic to cluster concept..

Gaurav

Hi Deemy, VCS 2.0 for Windows

Hi Deemy,

VCS 2.0 for Windows 2000 is very old.

You can shutdown/reboot the servers normally via the windows shutdown/reboot process.  If you are not shutting/rebooting all nodes in the cluster at the same time then VCS should move any online service groups to a surviving node.

You can freeze the service groups if you don't won't them to failover or you can offline them first as Gaurav suggested.  Either way will provide the same end result.

You mentioned that you are running VVR to replicate to another site but you did not mention if you are using GCM (now GCO in the 4.x and newer products.)  With GCM the site failover process is manual so you should not have to worry about this either when rebooting 1 or more nodes in the active site.

The only thing to note is that if the cluster configure is not closed before you reboot all nodes then you will come up in a stale state.  You can check this by running the followin command at a command prompt after rebooting the nodes.

   hasys -state

If the cluster is in a stale state then you can force it into a running state by running the following command.

   hasys -force -sys <node name>

Replace <node name> with the name of a node in the cluster to build from.  Remember that the node names will will need to be entered in all upper case.

Then just check to ensure that your service groups are online and everything is running as expected.

Thanks,
Wally