[OpenAFS] volume corruption: directory references disappear!?!

[J. Maynard Gelinas]

  Derrick thanks for your reply,

  Here's what I get in the FileLog on the host which was serving up those
volumes:


Sun Jun 30 04:00:22 2002 File Server started Sun Jun 30 04:00:22 2002
Sun Jun 30 08:17:58 2002 ReallyRead(): read failed device 0 inode 80C2640
errno 5
Sun Jun 30 08:17:58 2002 ReallyRead(): read failed device 0 inode 80C2640
errno 5
[...]

This continues on. This filesystem lives on a SCSI RAID array which shows
no disk or RAID set failure. I am able to touch empty files into the
affected filesystem. This system has only /vicepa currently for reasons
internal to our lab, though it did have six 50GB partitions previously.

  I was able to successfully vos dump each of the volumes from the
original's backup clone and then restore to new volume names on a
different fileserver. There appears to be no data loss from the time of
the clone. This seems strange though... if a clone is a set of pointers to
the original data, and the original volume became corrupted, how did the
cloned data survive?

Cheers,
--Maynard


On Sun, 30 Jun 2002, Derrick J Brashear wrote:

> On Sun, 30 Jun 2002, J. Maynard Gelinas wrote:
>
> >
> >   Here's a strange one:
> >
> >   A few users' volume home directory contents have simply disappeared over
> > the weekend. And it's not like they delted everything with an rm -rf *
> > because even . and .. are gone:
>
> Look at your FileLogs and your SalvageLogs, and tell us what you see
>
>