Hello,
at the Institute for Geophysics and Meteorology, we experienced a strange
problem with one of our AFS-fileservers which lead to data loss in the end.
The actual OS version is SunOS 5.6 and the AFS version 3.4 5.28. The chain
of events started with a power outage during normal operation. It seems that
the /vicepa-partition on the fileserver was in a special state (although we
don't know why), because after power was turned back on, Transarc's fsck
wanted to repair some damaged nodes; no "repaired" inodes were placed in
lost+found; subsequently, the salvager process reported that there was no
longer any association between Vblabla.vol-files and actual data of the
volumes.
We do not know why this happened. We are using AFS since 1993 and it was
the first time we experienced this problem. On the other hand, we also
want that this was the last time.
We would like to know from other users of AFS if they experienced this type of
data corruption, too. Especially we would like to know if this problem
occurs only under very special circumstances, or if it is frequent. Are there
possibilities to avoid these effects of fsck/salvager?
Any comments would be very appreciated. Thank you for any help.
Daniel Tiggemann
Dept. of Meteorology,
University of Cologne
