do_syncrepl: ... rc 68 retrying

[Peter Mogensen <apm@mutex.dk>]

I have a mirror which has stopped replicating. (slapd 2.4.21)

Somehow it has ended up in a situation where it looks like it is trying 
to replicate a modrdn operation, but the new RDN already exists on the 
consumer. Strangely, it doesn't exists on the producer.

Like this:

On Server1 (producer) I have:
contextCSN: 20100202142811.287095Z#000000#001#000000

cn=A,dc=example,dc=com
entryCSN: 20100127080339.261941Z#000000#001#000000

On Server2 (consumer) I have:
contextCSN: 20100127075402.694373Z#000000#001#000000

cn=A,dc=example,dc=com
entryCSN: 20100113084604.888851Z#000000#001#000000

cn=B,dc=example,dc=com
entryCSN: 20100127075357.910468Z#000000#001#000000


I notice that the contextCSN of server2 is only a few seconds after the 
entryCSN for cn=B,dc=example,dc=com

The log says:

syncrepl_entry: rid=003 LDAP_RES_SEARCH_ENTRY(LDAP_SYNC_ADD)
syncrepl_entry: rid=003 be_search (0)
syncrepl_entry: rid=003 cn=A,dc=example,dc=com
syncrepl_entry: rid=003 be_modrdn cn=B,dc=example,dc=com (68)
do_syncrepl: rid=003 rc 68 retrying


Any suggestion for how this could happen?

regards,
Peter