I'm posting this as a bug here since I didn't get any replies on the mailinglist.
Even if it doesn't counts as a bug, I think that there at least should be some comments on this matter.
Please read the mail at the URL provided.
Do you have an idea how windows behaves in this situation?
(In reply to comment #1)
> Do you have an idea how windows behaves in this situation?
Hmm, well no not really. However my spontaneous guess is that nobody in our Windows administration is aware whether a "Windows-computer" fails to update the DNS or not. I guess Windows asks the DNS for it's NS record (just as samba does) and thus it's possible to get one of the bind servers first in that list and thus failing the update. (This is just pure guessing though, I have no idea how Windows handles this).
Imho that's just a broken DNS setup. Either all your authorative mail servers allow updates or they don't. Unless the Windows machines keep retrying until they get lucky, I don't think we should.
Hmm, yes. As I stated in my mail I'm aware of the "broken" DNS setup. However it would be interesting to me if anybody knew how a Windows client acted in this scenario, but maybe that is rather hard to find out.
Just checked in a modified version of your patch to master and to the upcoming 3.6 branch, do you need that fix for earlier versions as well ?
Ah ok, thanks a lot! No it's not needed for earlier versions, 3.6 is good enough. Thanks again!
Thanks for checking, closing as fixed.
Hmm, I was a bit quick when I posted the answer on that this fix is not needed in earlier versions of Samba, we use Rhel 6 and it would be preferable if this patch could go into the version that's shipped with Rhel 6.1.
Rhel 6 is using Samba-3.5.4, I don't know what Rhel 6.1 will be using, but i doubt it will be 3.6.
Günther, this seems to be more like a RHEL product question now.
I've created a bug-report in their bugzilla, lets see if they pick it up.