Locking db getting corrupt?

[Michael Steffens]

Ray Frush wrote:
> We've been seeing a re-curring problem on one of our Samba servers:
> Samba 2.2.7 running on HP-UX 11.0.
> 
> After a week of running relatively quietly, we'll get a rash of errors 
> in smbd.log associated with a call from a user that their connection is 
> getting dropped. 
> " I_ve been dropped off twice in
> the past couple of days. So far I_ve been able to
> get back on -- once took a reboot of my pc, this
> time it only took trying to reconnect several
> times."
> 
[...]
> [2003/02/19 12:45:24, 0] locking/brlock.c:(235)
>  Failed to open byte range locking database
> [2003/02/19 12:45:24, 0] locking/locking.c:(328)
>  ERROR: Failed to initialise locking database
> [2003/02/19 12:45:24, 0] locking/brlock.c:(235)
>  Failed to open byte range locking database
> [2003/02/19 12:45:24, 0] locking/locking.c:(328)
>  ERROR: Failed to initialise locking database

We haven't had these since running 2.2.7a for about three weeks
now on HP-UX 11.11. But messages do remind those smbstatus does
report when the locking database just doesn't exist, yet.
So it might be temporarily unaccessible?

Does "dmesg" report anything suspicous, like it has
run out of nfiles or nflocks? In that case these kernel
parameters would need to be increased.

Michael