Weird replica corruption

[Carlos Eduardo Pedroza Santiviago <carlos@prognus.com.br>]

Hi,

I have an openldap master configured with 8 replicas. Unfortunately,
there's one that randomly begins to lose replicated informations. There
is some unknown erros in the replication log:

ERROR: Critical extension is unavailable: 2043378@^P
replica: replicaX.domain:389
time: 1134562962.0
dn: cn=Domain Users,ou=Groups,dc=DOMAIN
changetype: modify
add: memberUid
memberUid: XXXX
-
replace: entryCSN
entryCSN: 20051214122240Z#000002#00#000000
-
replace: modifiersName
modifiersName: cn=Manager,dc=DOMAIN
-
replace: modifyTimestamp
modifyTimestamp: 20051214122240Z
-

ERROR: Critical extension is unavailable: 2043378@`305C@340fC@
[...]
ERROR: Critical extension is unavailable: 2043378@^P

And the erros go on, until i restart the replica's slapd daemon.

Where does this numbers come from? Has anyone ever experienced this kind
of error??

Another question is: Those entries weren't replicated to that server,
and so it is now unsynched. Is there another way, besides erasing the
slave's data and populating all over again, to replicate only the lost
entries?

As you may note, i'm using slurpd, and the server is SLES9's OpenLDAP,
openldap2-2.2.24-4.5 which may be the vanilla 2.2.24 plus some patches.

thanks.
--
Carlos Eduardo Pedroza Santiviago - <carlos@prognus.com.br>