TSM Bare Metal Restore on EVA 3000

[Keith Reed]

Has anyone else done anything like this: We have a server that boots windows from the EVA 3k. We want to test Tivoli's Bare Metal Restore on it by presenting a new 9GB LUN from the EVA as the target for the BMR boot image restore. When the machine boots, the HBA identifies the LUN, but after the ASR floppy boots, the restore process says that it cannot find the disk to restore to. We are using the HBA drivers that the BMR process pulled from the original OS image.
Does anyone have any experience with this or any thoughts?

[Ivan Ferreira]

I don't know if you can "simulate" the restore process without actually doing it. I mean, if you can stop the restore process before destroying your current data.

What I'm thinking is that you should try to run the restore using the current VDISK, but of course the real data should not be erased. What I want to know is if the LUN is detected, considering that is the original LUN.

The new disk created in the EVA, has the same size? I'm pretty sure that windows does not know about units ID, but you could try setting the same OS unit ID.

[Keith Reed]

What I did was take the original, working, boot LUN and unpresent it. I then created a new Vdisk of the same size, and presented that new Vdisk to the server with the same unit ID that the original Vdisk had. I did this precisely so that I could leave the data on the original disk undisturbed. So, the new Vdisk has no formatting or signature, but is the same size and unit ID as the original. When the server boots, the HBA sees this LUN, and tries to boot from it, which naturally fails and proceeds to the TSM BMR process of booting from a floppy. When the TSM process tries to format the drive, it reports a general, vague failure message and aborts. The IBM consultant blames this on the EVA.

[Uwe Zessin]

Windows does not use the "OS Unit ID".

[Gert Albertsson]

I'm not sure but if you are trying ASR I guess you must have a diskette with drivers
for the disks, you must press F6 when you start the ASR process and then insert the diskette with drivers for the disks when asked for.

Not sure that this is what you are trying though.

[Steven Clementi · Accepted solution]

Keith:

Are you presenting only 1 path to the vdisk during the process? If TSM BMR is not multipath aware (highly doubtful, but am unsure due to lack of experience with the product), I can see you having a problem formatting and copying the image over to the new vdisk while there are multiple paths.

Try zoning out 3 of the paths (or 1 if your only using 1 HBA) and see what happens.


Steven

[Keith Reed]

That's a good idea, Steve, I'll try that.
Gert, I've been "assured" by the engineer that the drivers are on the disk, but I'll confirm. We have a testing window open next Tuesday, and I'll post the results for posterity.
Uwe, by Unit ID, I was referring to the LUN assigned to the VDISK at the EVA.

I'll let the forum know what progress I make.
Thanks, everyone.
--Keith

[Uwe Zessin]

Keith, my response was to confirm Ivan's guess about the 'OS unit ID'.

I wonder if the software hits a standby path by accident - that would explain a vague error message, if the software does not know how to cause a failover to bring the LUN into an online state.

If that happens, can you try to change the preferred path setting to the other controller?