mdaitc@gmail.com writes:

> The system seems to recognise 2 volumes of the same name:

....one containing all but one disk, and the other one containing
the missing disk from the first one.

> Does anyone have any ideas on how to recover from this situation, and
> have a working xlv0 volume again?

I guess you need to figure out why it is trying to attach the one
drive to a separate volume, oddly enough with a name identical to
the original one, and to sort that out. Can't help you with the
details, sorry.

--
Atro Tossavainen (Mr.) / The Institute of Biotechnology at
Systems Analyst, Techno-Amish & / the University of Helsinki, Finland,
+358-9-19158939 UNIX Dinosaur / employs me, but my opinions are my own.
< URL : http : / / www . helsinki . fi / %7E atossava / > NO FILE ATTACHMENTS

Atro Tossavainen wrote:
> mdaitc@gmail.com writes:
>
> > The system seems to recognise 2 volumes of the same name:
>
> ...one containing all but one disk, and the other one containing
> the missing disk from the first one.
>
> > Does anyone have any ideas on how to recover from this situation, and
> > have a working xlv0 volume again?
>
> I guess you need to figure out why it is trying to attach the one
> drive to a separate volume, oddly enough with a name identical to
> the original one, and to sort that out. Can't help you with the
> details, sorry.

My thought was that because the drive controller has changed, there's
effectively a new hardware ID of the drive, but the content of the
drive hasn't changed, and this confuses the system because it doesn't
understand how the hardware id has changed, and as such won't put it
back into the volume, even though the drive content has stayed the
same.

well that's my thoughts!

Matthew

mdaitc@gmail.com wrote:
> Hi,
>
> i've recently taken over administration of an SGI Irix system (IRIX64
> voyager 6.5 04101930 IP27) and have a problem with an xlv that's on the
> system.
>
> The volume is striped over 6 disks, connected to an Origin Vault, and
> is called xlv0.
>
> A while back, one of the drives failed (/dev/dsk/dks5d1s7). Actually,
> the disk controller card was at fault, not the drive itself. We swapped
> the controller card with an (almost) identical drive, in the hope that
> the volume would work. But have encountered a problem.
>
> The system seems to recognise 2 volumes of the same name:
>
> voyager 5# xlv_assemble
> xlv_assemble: Checking for Disk Plexing Option ... done
>
> VOL xlv0 flags=0x1, [missing unique piece]
> (node=voyager)
> DATA flags=0x0() open_flag=0x0() device=(192, 4)
> PLEX 0 flags=0x0
> VE 0 [incomplete]
> start=0, end=106664447, (stripe)grp_size=6,
> stripe_unit_size=128
> NULL
> /dev/dsk/dks5d2s7 (17777424 blks)
> /dev/dsk/dks5d3s7 (17777424 blks)
> /dev/dsk/dks5d4s7 (17777424 blks)
> /dev/dsk/dks5d5s7 (17777424 blks)
> /dev/dsk/dks5d6s7 (17777424 blks)
>
> VOL xlv0 flags=0x1, [missing unique piece]
> (node=voyager)
> DATA flags=0x0() open_flag=0x0() device=(192, 6)
> PLEX 0 flags=0x0
> VE 0 [incomplete]
> start=0, end=106664447, (stripe)grp_size=6, stripe_unit_size=64
> NULL
> NULL
> /dev/dsk/dks5d1s7 (17777424 blks)
> NULL
> NULL
> NULL
> Removing XLV Volume "xlv0" [missing unique piece]
> Removing XLV Volume "xlv0" [missing unique piece]
> xlv_assemble: Setting kernel configuration ... done
>
> Does anyone have any ideas on how to recover from this situation, and
> have a working xlv0 volume again? I'm a little baffled by the
> stripe_unit_size being different on the two "volumes"
>
> Thanks,
>
> Matt

Matt, part of the problem is that the xlv0 at the top has a
stripe_unit_size of 128, while the xlv0 a the bottom has a
stripe_unit_size of 64. The xlv at the bottom also thinks that
dks5d1s7 is the 3rd ve in the stripe, where is the top xlv seems to
indicate that the missing stripe is the first one.

What you *might* be able to do is simply delete the xlvs and recreate
your stripe using the same disks. As long as you don't use fx on any
of the disks, get the order of the stripes right and get the
stripe_unit_size right (128 is, I believe, the default) you should get
your filesystem back. If that doesn't work, well, how good is your
backup?

I don't have an IRIX box handy to run through the commands, but they
should be pretty simple. Just use xlv_mgr to 'delete object xlv0'
(check my syntax on that) then xlv_make to re-create it.

Jeff