My query:
>Managers,
>
>I've run into the curious problem with the Solstice (Online) DiskSuite 4.1.
>
>Hardware:
>Sparc Ultra Enterprise 3000
>512 Mb of RAM
>2 x CPU 250 MHz
>serial console
>2 x internal 2.1 GB disk
>2 x Sparc Storage Array 100, each with 18 x 2.1 GB disks
>Solaris 2.5.1, 4/97
>
>It's a brand new installation, where I was to create 18 mirrors of the parallel
>disks in two SSA's (no stripes). I decided to use Disksuite as I'm familiar
>with it. SSA took over the controller 0, so the internal disks are on the
>controller 1, with SSA 1 on controller 0 and SSA 2 on controller 2.
>I've adopted the following naming sequence for the metadevices:
>mirrors: d40 to d57
>submirrors on SSA 1: d0 to d17 (corresponding to disk nos in format)
>submirrors on SSA 2: d20 to d37 (corresponding to disk nos in format)
>
>Because of the serial console I was forced to use command line.
>I've created 6 copies of the database.
>To create mirrors I used ex:
> metainit d0 1 1 c0t0d0s0
> metainit d20 1 1 c2t0d0s0
> metainit d40 -m d0
> metattach d40 d20
> .........
>NOW comes the curious part:
>FOR ALL EVENLY NUMBERED MIRRORS IT WORKED OK.
>FOR ALL ODDLY NUMBERED MIRRORS the metattach command resulted in the message:
>Can't attach labeled submirror to an unlabeled mirror
>I tried to:
> compare the output from metastat for all metadevices: exactly the same for odd
> and even
> repeatedly cleared out metadevices and started from scratch: no joy
> used md.tab entries instead of command line: no joy
> shuffled the database copies to and fro among disks: no joy
> in desperation set up a network connection to another Sun with graphics
> console and used metatool to redo all: no joy
> had a friend Sun engineer to check: no joy
>
>I AM STUMPED !
>
>I'll probably do the job with the Volume Manager, but for the sake of my
>quiet sleep will someone please tell me what gives and what's the meaning
>of that error message ???
>
>Will of course summarise and thanks in advance
>
>Chris Wozniak
>FBA chris.wozniak@fba.com.au
resulted in only 2 answers.
Many thanks to:
Oleg Olovyannikov
Tawanda Queen t_queen@acs.org
Oleg's answer is correct. I put md database replicas on the odd numbered disks,
that became first submirrors without creating separate partitions for them.
That lead to the situation described below by Oleg. I wish it were mentioned in
the DiskSuitedocumentation though.
I went Veritas anyway and found it relatively easy to use for simple tasks
and much further removed from nitti-gritties than Online (Now for some that
may a a minus :)) )
Here's Oleg's answer:
>Chris,
>
>My recollection of this is a bit fuzzy, but I'm fairly certain that the
>error pertains to trying to attach a submirror which begins at track 0,
>cylinder 0 of a disk (say, c0t0d0s0) to a metamirror initialized with a
>disk beginning at another track/cylinder (say, c1t0d0s1). This is because the
>VTOC label is stored at that location.
>
>So, if you have (with the conditions above, that c0t0d0s0 was partitioned
>starting at cylinder 0, track 0):
>
>d10 -m d11
>d11 1 1 c1t0d0s1
>d12 1 1 c0t0d0s0
>
>metattach d10 d12 will give you this error message, since c0t0d0s0 contains the
>VTOC label (thus, it's labeled) and c1t0d0s1 doesn't. SDS doesn't like this.
>
>I've used two solutions: Reverse the submirrors or partition c0t0d0s0 starting
>at cylinder 1.
>
>Cheers,
>
>Oleg
Chris Wozniak
FBA chris.wozniak@fba.com.au
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:11:59 CDT