About a month ago, our backups went from taking ~8 hours to taking
16-24 hours or more.  During this time, two specific shares on one of
our servers, alva, started giving data timeout errors (e.g., alva
/home lev 4  FAILED [data timeout]).  Other shares still backed up
fine, even if they were the same size or larger and were located on
the same RAID array on the same server.  Furthermore, Amanda would
often report the failures twice:
 alva       /home              lev 0  FAILED [data timeout]
 alva       /home              lev 0  FAILED [data timeout]
If I checked the server's ps list while Amanda was running, I could
see two different attempts to back up a drive running at the same
time.

The problem appeared to start when I upgraded the OS (and hence
Amanda) on another Amanda client, but I don't know how upgrading one
client would affect the speed of backing up another client (unless
I've hit a bug in Amanda).

Any suggestions on how to fix this problem?  The symptoms seem odd
enough that I'm not even sure where to begin.  We're running Amanda
2.5.0p2-4, as provided with CentOS 5.  I know that newer versions of
Amanda are available, but I'd prefer to stick with what came with the
OS unless there's compelling reason to upgrade.

Thank you.

Josh Kelley