Re: Problem with N-Way multimaster replication after an node fail and restore

[Quanah Gibson-Mount <quanah@symas.com>]

--On Thursday, May 09, 2019 11:21 AM +0200 David Tello 
<david.tello.wbsgo@gmail.com> wrote:

> I have a problem with the N-Way replication multimaster. My enviroiment
> is debian Stretch and slapd (2.4.47+dfsg-3~bpo9+1) from Stretch
> backports. 

Hi David,

I notice your slapo-syncprov overlays are missing a sessionlog setting.  It 
is critical that this value be large (approximately the same as the number 
of entries in your database).  I'd also note that the order in which 
overlays (particularly syncprov) are instantiated on a database can matter. 
It's generally advised that syncprov be the first instantiated overlay (In 
your configuration, it's the 7th overlay on the mdb database with index 
{6}).  I.e., it should be the overlay at index {0}.

These changes may or may not address your issue, but I note them as they 
are items that should be addressed in your configuration.

If you are able to reproduce the issue after fixing the above config items, 
I would ask that you file an ITS at <http://www.openldap.org/its>.  Even 
better is if you're able to script the reproduction case (it seems rather 
straight forward) so we can reproduce it locally to work on a fix.

Regards,
Quanah

--

Quanah Gibson-Mount
Product Architect
Symas Corporation
Packaged, certified, and supported LDAP solutions powered by OpenLDAP:
<http://www.symas.com>