tdb_lock failed

[Jeremy Allison]

On Tue, Oct 28, 2003 at 03:43:27PM -0500, Brandon Craig Rhodes wrote:
> "Eric Boehm" <boehm at nortelnetworks.com> writes:
> 
> >     Brandon> This is happening in two different labs under both
> >     Brandon> Solaris 2.7 and 2.8 and renders samba-3 essentially
> >     Brandon> unusable.
> 
> >     Jeremy> Ok, do you have the Sun patch for the "thundering herd"
> >     Jeremy> problem with fcntl locks?
> 
> > We're are seeing this problem with Samba 2.2.8a on Solaris 8 *WITH*
> > the patch to fix fcntl (108528-19 or later). We've also seen it with
> > 2.2.8a and Solaris 8 with a later patch (108528-22). Perhaps the
> > Solaris patch doesn't fix the problem? Is there a way to verify this?
> 
> Oh, man, our labs were working with samba-3 pretty well today - I
> performed the upgrade late last night, when not many students were
> around - but a few minutes ago the whole thing has collapsed.  The
> cluster file server has a low load average, but keeps getting very
> large (like, three times the number of machines) numbers of smbd
> threads all apparently having problems with the fcntl(2) lock on
> secrets.tdb.
> 
> Meanwhile the password server seems completely loaded down, though we
> are investigating this.
> 
> Volker, I will try your test script later this evening or tonight to
> see if I can artifically produce the problem.
> 
> As the world burns around us,

Hopefully not :-(. Did you try Andrew Bartletts suggestion of 
running winbindd to cause it to linearise the queries to the DC ?

That was a good idea.

Jeremy.