"Re-mirroring" MSA1000 unit to use both buses

[Galen Tackett]

Our remote site has a cluster of two rx1620 servers which boot from a mirrored unit on an MSA1000. They have VMS version 8.3 and MSA1000 firmware V7.0.

The mirrored unit was initially set up with both drives on the same MSA1000 bus. Now they want to redo it so that there's a drive on each bus--the way it should have been done at first, though they didn't know that yet.

The ACU documentation talks about splitting a mirrored set, but they're using the CLI. Browsing the CLI manual here on hp.com, I haven't turned up a way to do this.

(I don't think they know how to use ACU, and anyway, I think telling someone halfway across the country "type this command and read me the result" works easier than saying "click on these buttons and tell me what happens." :-)

I don't have an MSA1000 on a VMS system here so I can't experiment with one. But one of you has probably dealt with this issue before and can offer some help?

Thanks,

Galen

[EdgarZamora]

It's been a while since I used the MSA1000 and I don't know if newer SW versions have improved its capabilities but I don't think there's a way to "split" a mirrored unit. There was a "migrate" command to switch from one raid level to another (given constraints) and you might want to look at that command, but my suggestion would be to backup your data, delete the unit, recreate it the way you want, then restore the data. Sorry couldn't be of more help.

[Robert Gezelter]

Galen,

I do not have the time to research it at the moment, but an observation about the implications of some of the tricks that I have used in the past.

Check if you can declare a third member of the mirror set (on the separate bus). Then drop one of the old members.

- Bob Gezelter, http://www.rlgsc.com

[Galen Tackett]

Bob,

Thanks for the reminder about adding a new member and yanking the old one. I'd already been thinking along that line but haven't read up on it yet.

I expect we can attain the desired configuration by adding a spare on the other bus to a mirror set, then yanking one of the original members (maybe waiting for the copy to complete). I'd like to feel more certain that this is the best or only approach, and then make sure we also "do the right things" at the MSA1000 CLI prompt.

I wonder how ACU does it??

[EdgarZamora]

The MSA1000 firmware is primitive and I don't think you can accomplish what Bob suggested. If you have spare disks I would recommend creating a new mirrored unit (with the desired bus separation) on the MSA, present this new unit to VMS, copy over the data, then get rid of the old unit. Or better yet, if you have HBVS you can shadow copy the data over.

[Rob Leadbeater]

Hi Galen,

If I remember correctly ACU can only be used from a Windows box, so I doubt they'd be able to get into it from VMS...

Cheers,

Rob

[Galen Tackett]

The MSA1000 firmware does have pretty narrow limits, but it can at least add a spare and fail over to that spare if a member fails--including, I assume, me yanking the drive?

(What really kills me about the MSA1000 CLI is that you can't abbreviate ANYTHING! :-)

[EdgarZamora]

Ok, I see what you're thinking of doing. Yanking one member of the mirrorset and hoping the spare kicks in might work, I guess. Good luck.

And I know what you mean about not being able to abbreviate. It's so annoying.

[Bill Hall]

Galen,

If down time is not an issue... Check the MSA's documentation that covers migrating disks from a Smart Array back-plane array controller on a Proliant to the MSA. I think there is enough meta-data on each disk such that you can power down the MSA1000 and then you can physically move the disks to the specific bus location that you want. Power up the MSA and away you go.

I haven't done this, but the MSA was designed and marketed with this type of disk/data migration as a major feature.

Bill

[Uwe Zessin]

> The MSA1000 firmware is primitive

No, it is not. It is too advanced and is too clever for your idea.



Bill Hall has the only workable solution:
-- drive roaming --

Unlike the HSG, which most of you are apparently familar with ;-)
a SmartArray controller can track if disk drives have been moved while the power was off. And the spare disk feature works different, too, as the controller software will 'fail back' if you replace the defective disk.

So:
- shutdown the hosts
- power down the MSA
- move the disk drives, but make sure they ar properly seated
- power up again

[Jan van den Ende]

Uwe,

I know you are a real storage expert, but

>>>
No, it is not. It is too advanced and is too clever for your idea.
<<<
and
>>>
So:
- shutdown the hosts
<<<

THAT does NOT sound "clever" on my frequency!

Surely any decent modern device can combine with VMS to do migrations of all kinds without downtime?
It need not be trivial, and the shutdown road might be easier, but I find it REAL hard to accept as the only way ...

Disclaimer: I do not (yet?) have any experience with MSAs.

Proost.

Have one on me.

jpe

[Galen Tackett]

Unfortunately I let this thread slip and never got around to saying how they fixed it. Memory dimly suggests that it might have been by merely moving the drive (perhaps with power down?), as several responders suggested.