[Samba] samba getting stuck, highwatermark replication issue?

[Andrew Bartlett]

On Sat, 2017-10-14 at 12:52 +0200, mj wrote:
> Hi Andrew,
> 
> Thanks for chiming in!
> 
> On 10/14/2017 12:16 PM, Andrew Bartlett via samba wrote:
> > We are carefully following the reports here, but we do expect
> > replication should be much more stable with Samba 4.7.
> 
> OK, that's interesting, because I actually wanted to upgrade ASAP, but 
> (the few) 4.7-upgrade experiences that have been posted, are mostly 
> about replication issues after having upgraded:
> 
> See https://lists.samba.org/archive/samba/2017-October/thread.html
> 
> Have people here generally upgraded to 4.7 already? Without major 
> issues? (does that explain the lack of discussion on 4.7?)
> 
> Or are people mostly waiting until a version 4.7.1 or .2 has been released?

I remain ever thankful to those who are willing to test new Samba
versions, particularly Samba release candidates.

In regards replication issues:
 - Newer Samba versions are becoming much more strict.  While this may
feel like a problem, it is about failing fast and in an obvious way,
rather than allowing latent issues to remain un-detected.

 - dbcheck all the things

 - join a new DC rather than upgrading in-place is a safer option if
you fear replication issues
  
 - You can test if Samba can join an existing domain by running:

samba-tool drs replicate clone-dc-database 

I do realise that some of the reported issues happen after the upgrade.
 This is a worry, but I think represents latent DB issues.

Finally, while taking care for user confidentiality (staff/student
names, unicodePwd and supplimentalCredential values etc) we do need to
see logs, typically less than the last 100 lines, for failures at a
higher log level (keep turning it up to until meaningful info appears).
 We can't really guess very much from errors like this from another
user in another thread:

   [2017/09/29 10:26:15.502219,  0]
../source4/dsdb/repl/drepl_out_helpers.c:959(dreplsrv_op_pull_source_ap
ply_changes_trigger)
       Failed to commit objects:
WERR_GEN_FAILURE/NT_STATUS_INVALID_NETWORK_RESPONSE

Sorry,

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