An in-SAN-e scheme? Trying to keep a single copy on the SAN

Suppose I have two identical servers, not clustered. They are exactly identical except for user data (and their Notes names and IDs). Suppose Server A has Notes data on disk D1 which is disk D:, while Server B has Notes data on disk D2 which is D:. Suppose that all access control and security is done through groups, and, Servers A and B are in the exact same groups.

Suppose that disks D1 and D2 are virtual disks created by my SAN. Suppose that my SAN also tricks Windows Server on Server A, telling it that disk D2 is its E: disk, but, that disk is offline.

Suppose I:

– shutdown server B successfully, both Notes and Windows.

– stop Notes Server successfully on Server A

– replace Notes.ini on Server A so that all data files come from E:,

– make all other changes to notes.ini needed to imitate Server B’s notes.ini

– get my SAN to convince Windows on Server A that Disk D2 is now online as E:, and, that disk D: is now offline,

– start Notes Server on Server A

Will it start? Will it serve up the user data from Disk D2? What will go wrong?

If it works at all, it will appear to the users that the data has moved to a different server.

For web access, we can put a router in front of both machines which can switch B’s traffic over to A upon request.

Extra credit: same questions, except:

– we use transactional logging, and,

– server B crashed instead of shutting down.

TIA!

Subject: an in-SAN-e scheme? Trying to keep a single copy on the SAN

Are you expecting Server A to remain named Server A, or do you want it to become Server B?

If so, then as long as you have provided Server A with Server B’s id (and the Notes.ini reflects the change), then it will work great. I have upgraded server hardware like this many times.

If B were to have crashed, then any database that was open at the time of the crash will need a consistency check. The transactional logging will only work if Server A is configured to use Server B’s transactional log (ie has access to it.) If it doesn’t have access to the log then transactional logging on Server B i useless in this scenario and would need to be deleted BEFORE Server B is restarted.

Just don’t ever let the two servers access the same physical file at the same time!!!

Subject: RE: an in-SAN-e scheme? Trying to keep a single copy on the SAN

Thanks for your help! This is useful info.

I was really expecting Server A to stay Server A, with its own ID file.

What we’d love to do is to “plug” (via SAN) one of Server B’s “disk volumes” into Server A, and, have Server A serve both its own files and Server B’s files. The B disk would only have files that have no replicas on Server A. At that time, we would add, in the A Disk, a Directory Link pointing to the B disk.

We’d be sure to remove the Link later and “unplug” the disk once Server B was back. I imagine this would lead to user problems with workspace icons, at that time.

Can there be two sets of transaction logs? Server A would have its own, so we cannot simply use Server B’s logs on Server A. It’s just a point of curiosity, because, it turns out that these servers do not do transaction logging.

Thanx again!