[PATCH] Cache messaging dgm connections

Ralph Böhme slow at samba.org
Thu Sep 15 05:08:17 UTC 2016


On Wed, Sep 14, 2016 at 10:01:56PM -0700, Jeremy Allison wrote:
> On Thu, Sep 15, 2016 at 06:52:48AM +0200, Ralph Böhme wrote:
> > Hi Garming,
> > 
> > On Thu, Sep 15, 2016 at 11:36:55AM +1200, Garming Sam wrote:
> > > Seen some flapping tests in autobuild, and it looks like it's related to
> > > the dgram changes. I originally thought it was another replication
> > > issue, because of an IO timeout, but apparently not.
> > > 
> > > I'm not sure if it's the callers fault here, but I thought it would be
> > > easier if you could take a look and give some advice.
> > 
> > sorry for this.
> > 
> > > https://git.samba.org/autobuild.flakey.sn-devel-144/2016-09-14-1112/samba.stderr
> > 
> > The problem is here:
> > 
> > #29 0x00002af8894bdd54 in dns_notify_dnssrv_done (req=0x2af892d1f170) at ../source4/dsdb/samdb/ldb_modules/dns_notify.c:72
> > 
> > This calls talloc_free() on an imessaging context and the crash is
> > similar to what I was trying to fix in 02bb6b2: at the time a tevent
> > timer destructor is called in the course of the talloc_free() on the
> > imessaging context
> > 
> > #9  0x00002af88224ba1d in tevent_common_timed_destructor (te=0x2af88c8a4b70) at ../lib/tevent/tevent_timed.c:140
> > 
> > some required object is already freed. Still trying to understand
> > which object, I don't think it's the tevent context like in 02bb6b2
> > but it could be the same problem.
> > 
> > Still looking into it...
> 
> Yeah me too. I've been studying for a couple of
> hours, and haven't found it yet :-). Not being
> able to catch it under gdb is the hard part.
> 
> Let's see which of us finds it first :-).

appreciate your help, all this ref and handle stuff is really
voodoo.

Cheerio!
-slow



More information about the samba-technical mailing list