How to replace a shadow system disk

[Bojan Nemec]

Hi,

I have a 3 node cluster with sistem disk in shadow set (VMS 7.3-2). The shadow set is made from disks attached to a HSG80 and an EVA. All nodes boots from the HSG80.

Now the HSG80 will be replaced by another EVA and two of nodes will change place in the building. Before moving the systems the new EVA disks will be added in the shadow sets as the third member. After that the systems will be moved and finaly the HSG80 shadow set member will be removed.

The environment is 24x7.

The question is:

Can I shutdown one node, move it and boot it from the EVAs shadow member. Then repeat the operation with the second node and at end reboot the last node from the EVA.

My concern is in the statement from the documentation:

"When multiple nodes boot from a common system disk shadow set, ensure that all nodes specify a physical disk that is a source member of the system disk shadow set."

and for a short time (maybe few hours) the systems will not be booted from the same phisical disk.

Bojan

[Karl Rohwedder · Accepted solution]

I think, when the shadwoset has reached a steady state, it should be o.k.

regards Kalle

[Bojan Nemec]

Karl,

My intention was to wait untill all disks are merged. Probably this will be done some days before the move operation.

I will be glad to hear somebody who has done a similar operation.

Bojan

[Phillip Thayer]

Bojan,

I agree with Kalle. Once the shadow set is in a steady state it should not be a problem. From my past experience I have done something similar to this with system disk shadow sets and had no problems with it.

Phil

[Bret Graham]

Bojan,

What you propose will work. The mention of booting from the same member of the shadow set is meant to ensure that you always boot from a disk that is a source member of the shadoow set. If one node has booted from this given shadow member, the shadow set will exist with at least that boot member.

Different nodes booting from different shadow members works fine as long as the member they boot from is a full, source member of the shadow set at the time of the boot. Otherwise you will see a SHADBOOTFAIL bugcheck.

Bret

[Bojan Nemec]

The operation works fine.

The third member was added from the new EVA two days before the moving operation so it have all the time to become a valid source member.
Then both systems were shut down and moved to the new place. We booted them from the new EVA and they become members of the cluster. After that the thisrd node was shut down and booted from the same physical (virtual) disk as the other two.
At the end the old HSG disks were removed from the shadow set.

Bojan

[Bojan Nemec]

Thread closed. See previous post.