--On May 5, 2014 at 7:40:31 PM -0700 "Paul B. Henson" <henson@acm.org> wrote:
From: Quanah Gibson-Mount Sent: Friday, May 02, 2014 7:22 PM I would suggest (if you haven't) enabling sync replication logging (loglevel sync) in addition to whatever other loglevels you have. I've found it is possible to send MMR into an endless loop in some cases recently. I'm still working on the reproduction case for it, but it happens 100% of the time for a client of ours eventually.Ouch. By any chance, does your client use the password policy module? I've never had a problem like this until I started testing with it in my development instance.
No, we don't use ppolicy. But the issue for us is arising in a functionality that is similar to ppolicy that's Zimbra specific.
I've enabled the additional logging level and reset my testing to the beginning, and will see if it does it again. When it happens to your client in production, how do you resolve it?
You can gdb slapd, and manually fix the serverID in the syncinfo structure, or you can restart all your slapd servers.
--Quanah -- Quanah Gibson-Mount Server Architect Zimbra, Inc -------------------- Zimbra :: the leader in open source messaging and collaboration