Time skew shouldn't matter to much because of the way that CSN
generation works - older clocks are always "pulled up" by the faster
clocks as their recieve their CSNs via replication.
Are you specifically seeing an issue because of time skew?
Thanks,
William,
On Mon, 2018-04-09 at 16:11
this is the correct document for the full cleanup.
it may be possible to recover by just deleting the nsState on a master and
doing a reinit of all the replica, but there may be stale csn traces in the
ruv in each replication agreement.
M.
On Mon, Apr 9, 2018 at 5:49 AM, Paul Whitney
Hi guys,
I have been reading up on a fix for this and only found one set of procedures
to repair this issue.
(http://directory.fedoraproject.org/docs/389ds/howto/howto-fix-and-reset-time-skew.html).
Would removing/recreating the replication agreement resolve this issue
instead of the steps