Samba4 BDC with Samba4 PDC

[Mike Howard]

On 08/06/2012 08:52, Kai Blin wrote:
> On 2012-06-08 09:09, Mike Howard wrote:
>
> Hi Mike,
>
>>> And you reliably manage to get Samba stuck in a way where you need to
>>> reboot to be able to get DNS working again? As I said, I'd expect a
>>> restart of the "samba" process to be sufficient (you might need to kill
>>> all the running samba processes if they don't exit cleanly).
>> And _I_ reliably manage ..... ? A Samba4 component fails repeatedly, on
>> different hardware, also experienced and reported by other users and
>> it's my fault? Very cute.
> No, I didn't want to imply that at all. Sorry if that came out wrong.
> I'd put the stress on _reliably_.. I can't get Samba to fail that way.
> If you can, and we can figure out how your set-up is different, that'll
> help me to debug this. I'm just very surprised that whatever is
> happening only goes away after the reboot, so samba seems to mess up
> something at the kernel level. That's not supposed to be possible from
> the user-space. The fact that this happens on multiple different
> machines makes this even more curious.
>
>>>> With regards to specific patterns, actually, there was one scenario when
>>>> it always seemed to fail and that was when I was using 'apt-mirror'
>>>> manually (as opposed to from cron) to mirror debian security updates
>>>> (plus others). No idea why that would be.
>>> Ok, that does seem like asking the external forwarder might be
>>> triggering that. Volker recently made that code much more resilient. You
>>> might want to just give this another try.
>> Ah, so it's not my fault then?
> No, the forwarder code was a pretty quick hack to make development of
> some other DNS code easier while running the internal DNS on my dev
> machine. I'm not expecting that to be bug-free at all. :) That said, the
> current code in master should behave better.
>
> Cheers,
> Kai
>
Hi Kai,

Thanks for the clarification.

I'm just running through another iteration hoping to get a BDC 
replicating (DNS wise) and if I can't succeed I'll go back to the 
internal server (using the now beta) and see if I can replicate things. 
If I can, reliably, I'll then set up a trace using tcpdump. Is there 
anything specific you would like that might help, as in log level, 
tcpdump options etc?

Cheers,
Mike.
--