printer tdb file corruption

[Jeremy Allison]

On Mon, Apr 24, 2006 at 01:11:22PM +0300, Kristaps Rāts wrote:
> Hello,
> we have a samba 3.0.14a host as a print server and today it went awol -
> an smbd process is eating up 99% cpu time and printing doesn't work.
> After restarting samba everything looks normal, but when someone tries
> to print the same thing happens again. On startup the logs go something
> like this:
> 
> [2006/04/23 15:18:55, 0] tdb/tdbutil.c:tdb_log(725)
>   tdb(/var/cache/samba/printing/HPLJ1200PS.tdb): remove_from_freelist:
> not on list at off=107956
> [2006/04/23 15:18:55, 0] tdb/tdbutil.c:tdb_log(725)
>   tdb(/var/cache/samba/printing/HPLJ1200PS.tdb): tdb_free: right free
> failed at 107956
> [2006/04/23 15:18:55, 0] tdb/tdbutil.c:tdb_log(725)
>   tdb(/var/cache/samba/printing/HPLJ1200PS.tdb): remove_from_freelist:
> not on list at off=106764
> [2006/04/23 15:18:55, 0] tdb/tdbutil.c:tdb_log(725)
>   tdb(/var/cache/samba/printing/HPLJ1200PS.tdb): tdb_free: left free
> failed at 106764
> [2006/04/23 15:18:55, 0] tdb/tdbutil.c:tdb_log(725)
>   tdb(/var/cache/samba/printing/HPLJ1200PS.tdb): rec_read bad magic
> 0xd9fee666 at offset=68072
> [2006/04/23 15:40:56, 0] tdb/tdbutil.c:tdb_log(725)
>   tdb(/var/cache/samba/printing/HPLJ1200PS.tdb): rec_read bad magic
> 0xd9fee666 at offset=68072
> [2006/04/23 15:41:49, 0] tdb/tdbutil.c:tdb_log(725)
>   tdb(/var/cache/samba/printing/HPLJ1200PS.tdb): rec_read bad magic
> 0xd9fee666 at offset=68072
> 
> After removal of the offending tdb everything went back to normal -
> samba created a new tdb (about 5 times smaller than the original) and
> printing works again. Anyone care to look at that pesky tdb?

Did you have any crash messages or aborts in any smbd log file ?
This can be caused by a smbd crash. It's probably not worth looking
at (IMHO) unless it's reproducible in the latest 3.0.23 pre-release.

Jeremy.