SUMMARY: ADVFS: Is my data lost?

From: Bernt Christandl <beb_at_rosat.mpe-garching.mpg.de>
Date: Tue, 18 Jul 95 09:26:20 +0200

Hello Managers,

i have been too fast with my question, i apologize.
Indeed I did have another idea and that one works...

My question was:

> one of my alphas (3000/400, osf/1 v3.0) yesterday crashes three times
> with an advfs internal inconsistency error (Module = 41, Line = 552)
> within 4 hours...
>
> Since then i am not able to mount that filesystem, which covers nearly
> 2 disks, again.
>
> The good things are that
> - i can "see" both disks after shutdown and power-cycle with "sho dev"
> at the console prompt
> - test scsi results in scsi ok
> - when i boot the alpha, both disks are recognized with hardware-type
> and firmware level (dec-disk dsp5200 each)
> - and at last after the boot, "disklabel -r" "sees" the correct
> disklabels (only partition g, starting at an offset, used as AdvFS)
>
> The bad things are that
> - a mount-try results in an i/o-error
> - showfdmn gives the following output:
>
> showfdmn: unable to get info for domain 'data06b_dmn'
> showfdmn: error = E_VD_DMNATTR_DIFF (-1079)
>
> At the moment i'm out of ideas what to try next.
> Is my data lost? Is there a chance to recover it? How?


My solution has been:

The above error 1079 is explained (man advfs_err) as one or more partitions
or disks is somehow invalid for the domain.

With that hint i deleted first the one and than the other link under
/etc/fdmns/data06b_dmn (was Murphy, first i hit the wrong one ..;)

After this, i could mount the filesystem, but not access it.
A "ls" results in a crash of the machine... (why that???)

Then i tried a chvol on both partitions and got reasonable results,
so i tried a "chvol -A" on the critical one. The "vdf" then "saw" the
second partition, but the sizes were only those of the first one.

After an "umount" and "mount" my filesystem was accessible again.

Puuhhhh!

Now i do restore to another machine and disk and then i will try to
investigate/understand the *real* problem that causes the original crashes
and why advfs went mad the last time...

Sorry for the bandwidth...

Bernt
                                                                       
-------------------------------------------------------------------------
- Bernt Christandl / Max Planck Institut fuer Extraterrestrische Physik -
- D-85740 Garching / Phone: +49/89/3299-3346 / Fax: +49/89/3299-3569 -
- Internet: beb_at_mpe-garching.mpg.de -
-------------------------------------------------------------------------
Received on Tue Jul 18 1995 - 10:30:49 NZST

This archive was generated by hypermail 2.4.0 : Wed Nov 08 2023 - 11:53:45 NZDT