[rescue] Faulted Solistice Disksuite array...

[Carl R. Friend]

    Good morning,

On 4/6/19 3:00 AM, Mike Spooner wrote:

> IIRC there is an option to the metareplace command which says to
> just set the status and *not* to reinitialise the disk. I remember
> having to do exactly that kind of thing back in 2000 on a
> production database system (junior tech had pulled the wrong disk!).

    Was that with LVM on Solaris 9?  I'm running Disksuite on Solaris 8.
What I want very much to do is avoid forcing a rebuild/"resilver" of the
drive that's in "Maintenance" and instead have the software RAID driver
flip the "last erred" drive into "Maintenance" and let me see if I can
recover stuff (the "last erred" drive is, I suspect dead).

On 2019-04-05 21:32 EDT Patrick Giagnocavo wrote:

 > https://www.thegeekdiary.com/solaris-volume-manager-svm-command-line
reference-cheat-sheet/
 >
 > what does metadb -i show?

    I don't recall a "metadb -i" syntax, but I'll fire the machine up
later today to see what it says (other than the whine of all the old
disks).  Recall that this is the older version of Disksuite not LVM as
the article alludes to.

 > A silly idea that occurred to me, would be to use dd or ddrescue to
 > create a disk image of each disk. Then, create a VM using QEMU/sparc,
 > install Solaris on it, and attach the disk images. This would allow
 > you to try out different methods of fixing the issue without risking
 > harm to the original disks.  Probably a dumb idea - but it would be
 > neat if it worked!

    That's actually an extremely good idea and one that I'll look into.
Getting individual images of the good disks would allow for later
experiments as well as having a snapshot of the old iron.  The question
now becomes, "Where do I have the space to stash the images?"

    For the curious, here's the history of the machine, which I've
worked with pretty much since it was delivered in the 1990s.  The
thing was originally configured with a single 2GB boot drive, with
Solaris 2.5.1 on it, an early version of Disksuite, and a web-interface
to the RAID bits.  The array was 11 2GB disks and a single 2GB hot-
spare.  It remained in that configuration until 2001 or so when I
acquired it, and immediately upped the array to use 4GB disks instead
of the original 2s.  Over the years, several updates to the device
increased the size of the array, saw the memory grow and the OS updated
to Solaris 8 running in 64-bit mode.  The current configuration is a
72GB disk in the internal position behind the "front panel", a mirror
72GB disk for the boot drive in slot 15 of the array bay, and 10 36GB
disks in a RAID-5 and a hot spare.  One slot in the array backplane
seemed wonky at some point so I de-populated it.

    If I can solve my current quandry, my intent is to return it to its
original configuration.

    Cheers!

-- 
+------------------------------------------------+---------------------+
| Carl Richard Friend (UNIX Sysadmin)            | Boylston            |
| Minicomputer Collector / Enthusiast            | Massachusetts, USA  |
| mailto:crfriend at rcn.com                        +---------------------+
| http://users.rcn.com/crfriend/museum           | ICBM: 42:20N 71:43W |
+------------------------------------------------+---------------------+