Although Quanah hasn't been able to trigger this yet, the regression test in master he wrote has been consistently able to trigger for me on my machine, so I've started to investigate.
For posterity and in case anyone is interested, I have uploaded the testrun/ directory from a failing run (I suspect the fact of this laptop having a slow 2-core CPU helps) with a slightly patched slapd that records the thread ID as well since, in part, this seems like a race of some sort.
The tgz is available at ftp://ftp.openldap.org/incoming/its8444-regression-testrun-sync,stat.tgz
So far it looks like replica #3's threads 7f50fb7fe700 and 7f51017bb700 are both trying to apply the modification with CSN 20170605125334.856475Z#000000#001#000000 which sends it into a full refresh.=20
--=20 Ond=C5=99ej Kuzn=C3=ADk Senior Software Engineer Symas Corporation http://www.symas.com Packaged, certified, and supported LDAP solutions powered by OpenLDAP