On Fri, 2013-05-31 at 12:36 +0100, Paul Littlefield wrote: > On 31/05/13 12:25, Paul Littlefield wrote: > > Workaround: put guid names in /etc/hosts if you experience this bug. > > > I don't believe it. That has fixed it. > > No more WERR_BADFILE errors in /usr/local/samba/var/log.samba > > Domain Computers that were not in DC2 now are. > > Many thanks Giedrius <giedrius+sa...@su.lt> > > Replication now works. > > > > Aaahhh, interesting. I will eat my hat if this fixes it. > > > It's a bit chewy, but well worth it. :) That's my lunch sorted, and my > Windows Serverless network. > > Amazing... and so typical of Linux. The solution is there, you just need > someone to point it out. > > Thanks to everyone. > > > PS: ok, so who is going to add it to the official Samba wiki page now?
We really don't want that in the wiki, because you should not be putting fixed things in /etc/hosts, we have to fix it another way. What we do want is to know: - is this still an issue in GIT master? - can we detect the difference between this error and a real non-existent name by looking at return codes (we had a similar issue on FreeBSD, and caught it this way). I'm hoping we might just be missing that fix, or there was a regression in that. We have (multiple) alternative DNS resolvers internally, and we do use them. We try to use nss when we can, but fall back for the cases that don't work. If we can't rely on the error code, we may have to use a pre-check of the name instead. Any assistance tracking this down most valued, as it seems to be biting folks. Thanks, Andrew Bartlett -- Andrew Bartlett http://samba.org/~abartlet/ Authentication Developer, Samba Team http://samba.org -- To unsubscribe from this list go to the following URL and read the instructions: https://lists.samba.org/mailman/options/samba