samba - May 2007 - Countless "rec_read bad magic" lines in log.smbd

One of the servers we admin has countless such errors in log.smbd:

[2007/05/07 14:17:08, 0] tdb/tdbutil.c:tdb_log(783)
   tdb(/var/cache/samba/printing/HPLJ3600.tdb): rec_read bad magic 0x443810ec at
offset=28412

The only thread I saw regarding "rec_read" stated:

<><><><><>
If this really happens often to you, then you have some
basic problems with your setup. Either your hardware is
flaky or your smbd processes tend to crash. If you happen to
use reiserfs for /var/cache/samba you might try to change
that fs to ext3.

Volker
<><><><><>

On this server, all partitions are XFS.

This is the only printer at this particular site. It is attached to the LAN via
an integrated JetDirect interface. Samba talks to CUPS, and from CUPS to the
printer.

So, should I suspect the LAN patch cord?

-- 
Michael Lueck
Lueck Data Systems
http://www.lueckdatasystems.com/

On Mon, May 07, 2007 at 02:33:33PM -0400, Michael Lueck
wrote:
> One of the servers we admin has countless such errors in log.smbd:
> 
> [2007/05/07 14:17:08, 0] tdb/tdbutil.c:tdb_log(783)
>   tdb(/var/cache/samba/printing/HPLJ3600.tdb): rec_read bad magic 
>   0x443810ec at offset=28412
> 
> The only thread I saw regarding "rec_read" stated:
> 
> <><><><><>
> If this really happens often to you, then you have some
> basic problems with your setup. Either your hardware is
> flaky or your smbd processes tend to crash. If you happen to
> use reiserfs for /var/cache/samba you might try to change
> that fs to ext3.
> 
> Volker
> <><><><><>
> 
> On this server, all partitions are XFS.
> 
> This is the only printer at this particular site. It is attached to the LAN
> via an integrated JetDirect interface. Samba talks to CUPS, and from CUPS 
> to the printer.
> 
> So, should I suspect the LAN patch cord?
No, I'd run memcheck on that server :-).

Jeremy.

John Drescher wrote:
> If you get a single
> failure you have some type of hardware problem
It is always possible...
> which may include a
> problem with the electricity (1/2 second power outage...).
The systems at this client are powered via a UPS, so I am not suspecting trouble
there.
> I have had
> a system pass a couple day test only to fail 72 hours into testing
> (let the system run memtest over a weekend).
Unfortunately our client leaves for "Summer mode" this weekend, so an
evening of testing is as much as I am hoping for.

-- 
Michael Lueck
Lueck Data Systems
http://www.lueckdatasystems.com/

On Tuesday 08 May 2007, Michael Lueck wrote:
> Michael Lueck wrote:
> > Meanwhile...
> >
> > The client happened to notice a tray icon for the printer. Clicking
that
> > brings up a dialog showing countless jobs.
> >
> > Print jobs get through correctly.
> >
> > Samba communicates to the printer via CUPS and an integrated JetDirect
> > card. CUPS shows no jobs in the printer queue.
> >
> > So, could these "stuck jobs" have anything to do with the
log entries?
>
> Never mind. I stopped samba, deleted
> /var/cache/samba/printing/HPLJ3600.tdb, restarted samba and the jobs are
> history.
I had and have the same problem on some of our servers, and deleting the file 
always solves the problem for me.

I do however not think I have a hardware problem or filesystem issue... 
Everything on the servers works fine, even with long uptimes.

Hope I will solve this mistery one day .-)
> I will monitor the log file today, and this evening try the floppy image of
> Memtest86.
>
> Thanks!
>
> --
> Michael Lueck
> Lueck Data Systems
> http://www.lueckdatasystems.com/