Was uninstalling an LDAP replica on my test system before doing production using the multi-server install guide as a ref. I think there may be a typo there that could cause a subsequent ldap stop/start to bomb.
My test ldap had replication enabled based on what i found in slapd.conf, although there never actually was a replica. I made the edits to slapd.conf per the guide and all seemed ok, the ldap stop/start worked ok. I never checked slapd.conf or I would have seen that my edits got overwritten and the master was still enabled for replication. But a few days later I restarted the test server w/ldap on it and slapd didn't start. Took me bit to figure it out because the slapd error message returned is mostly associates with db corruption or permissions problems in the forums.
If you follow the multi-guide to remove a replica, all appears fine but the master may still config'd for replication because edits should be to slapd.conf.in, not slapd.conf. And the next step in the guide, deleting accesslog dir, creates a situation where the next ldap stop/start will fail because the repilcation-enabled master looks for accesslog.
So basically if your slapd.conf.in is set up for replication and you follow the multi-server guide to uninstall a replica, the second ldap stop/start after uninstalling the replica goes bang! and you get
[zimbra@email1o conf]$ ldap stop;ldap start
slapd not running
Failed to start slapd. Attempting debug start to determine error.
backend_startup_one: bi_db_open failed! (-1)
I looked for a bugzilla on this, is that where doc RFE's are ?