Vampire a Samba4 domain to Samba3?

[Andrew Bartlett]

On Fri, 2011-03-18 at 15:37 +0100, Christian Huldt wrote:
> Hi Mat
> 
> fre 2011-03-18 klockan 09:47 +0300 skrev Matthieu Patou:
> > On 18/03/2011 05:29, Taylor, Jonn wrote:
> > > On 03/17/2011 05:18 PM, Matthieu Patou wrote:
> > >> Hi Taylor&  Christian
> > >>>> The default, ldb.
> > >>>>
> > >>>> I assume we would need to get the dc replication up as a simple
> > >>>> rsync of
> > >>>> (parts of) samba/private wouldn't suffice as replication?
> > >>>>
> > >>>>
> > >>> This is the same problem that I have been having. I was seeming is stuck
> > >>> ldb transactions or samba4 would completely crash. I even sent you logs
> > >>> and dumps on this problem.
> > >>>
> > >> Well normally the replication for ldb using vampire should just work I
> > >> vampirized a domain one more time with success last days.
> > >>
> > >> Can you resend the logs related to your pb (or point me the message).
> > >> I might have just forget about it !
> > >> Sorry !
> > >>
> > >> Matthieu.
> > >>
> > > Mat, I forwarded you the 2 messages that I sent you.
> > >
> > Yeah got them, at least one didn't ring a bell at all for me so I don't 
> > remember reading it ;-(.
> > 
> > In order to have a more useful backtrace we need a real dump of samba to 
> > see how it was when it crashed.
> > This can be done by adding:
> > panic action    =       /usr/local/src/samba4/selftest/gdb_backtrace %PID%
> > 
> 
> I added that, but the problem has come up three times i six months, so I
> can't really say when we get something out...
> 
> We had yet another incident today (before I added the panic action
> unfortunately), the only thing I found in the log was a few entries like
> this (a few entries per machine that lost connectivity to the server
> that is) and not a huge build up om memory usage or anything, I assume
> it was those two samba processes that didn't go away with a simple
> kill...
> 
> [Fri Mar 18 13:30:06 2011 CET,
> 0 ../../libcli/auth/schannel_state_tdb.c:158:schannel_fetch_session_key_tdb()]
> schannel_fetch_session_key_tdb: Failed to find entry with key
> SECRETS/SCHANNEL/CA048

Hmm.  I'll need to look over the code, but I'm pretty sure that should
not be a level 0, more like 10 - it's quite normal for clients to try
this (using their old key, not knowing the server restarted) and to then
need to establish a key. 

Andrew Bartlett

-- 
Andrew Bartlett                                http://samba.org/~abartlet/
Authentication Developer, Samba Team           http://samba.org
Samba Developer, Cisco Inc.