[PATCH] Fix BAD_NET_RESP error in DsReplicaSync

[Andrew Bartlett]

On Sun, 2017-07-23 at 00:22 +0200, Stefan Metzmacher wrote:
> Am 22.07.2017 um 12:29 schrieb Andrew Bartlett via samba-technical:
> > This error has been plaguing out tests for a while making our
> > replication tests flap.
> > 
> > I got it locally and with a network trace, which made it clear the
> > issue is a lack of mutual exclusion between outbound GetNCChanges and
> > DsReplicaSync operations.
> 
> Pushed, thanks!
> 
> Can you upload the network trace to the bug report? I'd like to have a look.

Done.  So far just the problematic TCP stream, the rest is 160MB even
xz compressed.  Of course I can upload that as well if you like.

The trick to finding the bad packet was time correlation with the
failure in the subunit logs.

I won't say we have seen the last replication flapping test, but I'm
feeling much better about it than I have for a number of years. :-)

Andrew Bartlett

-- 
Andrew Bartlett                       http://samba.org/~abartlet/
Authentication Developer, Samba Team  http://samba.org
Samba Developer, Catalyst IT          http://catalyst.net.nz/services/samba