Didn't see an answer to this problem we had, so for documentation
purposes, posting it here, in case someone else runs across it.
Configuration server was Centos 5.11, multi-master with 3 more replicas
and a Windows Sync agreement. (We're working on moving to RHDS, so this
was "frozen" in production while trying to get that project moving.)
From the yum.log when the upgrade was committed:
Apr 21 12:48:16 Updated: centos-ds-base-8.2.8-2.el5.centos.x86_64
Apr 21 12:48:18 Updated: centos-ds-admin-8.2.1-1.el5.centos.x86_64
Apr 21 12:48:18 Updated: centos-ds-console-8.2.0-4.el5.centos.noarch
Apr 21 12:48:19 Updated: centos-ds-8.2.0-2.el5.centos.x86_64
Apr 21 12:48:44 Updated: centos-ds-base-8.2.8-2.el5.centos.x86_64
Just prior to upgrade:
[21/Apr/2016:12:44:39 -0400] NSMMReplicationPlugin - Found replication
agreement named "cn=ds0 to ds1, cn=replica, cn="dc=cise, dc=ufl,
dc=edu", cn=mapping tree, cn=config".
[21/Apr/2016:12:44:39 -0400] NSMMReplicationPlugin - agmt="cn=ds0 to
ds1" (ds1:389): No linger to cancel on the connection
[21/Apr/2016:12:44:39 -0400] NSMMReplicationPlugin - agmt="cn=ds0 to
ds1" (ds1:389): Disconnected from the consumer
[21/Apr/2016:12:44:39 -0400] NSMMReplicationPlugin - agmt="cn=ds0 to
ds1" (ds1:389): State: start -> ready_to_acquire_replica
[21/Apr/2016:12:44:39 -0400] NSMMReplicationPlugin - agmt="cn=ds0 to
ds1" (ds1:389): State: ready_to_acquire_replica -> wait_for_changes
[21/Apr/2016:12:44:40 -0400] NSMMReplicationPlugin - supplier: {replica
...
And as the upgrade finished:
ds1" (ds1:389): repl5_inc_stop: protocol stopped after 1 seconds
[21/Apr/2016:12:48:44 -0400] NSMMReplicationPlugin - Found replication
agreement named "cn=ds0 to ds1,cn=replica,cn=dc\3Dcise\2C dc\3Dufl\2C dc
\3Dedu,cn=mapping tree,cn=config".
[21/Apr/2016:12:48:44 -0400] NSMMReplicationPlugin - The replication
agreement named "cn=ds0 to ds1,cn=replica,cn=dc\3Dcise\2C dc\3Dufl\2C dc
\3Dedu,cn=mapping tree,cn=config" could not be correctly parsed. No
replication will occur with this replica.
[21/Apr/2016:12:49:20 -0400] NSMMReplicationPlugin - Found replication
agreement named "cn=ds0 to ds1,cn=replica,cn=dc\3Dcise\2C dc\3Dufl\2C dc
\3Dedu,cn=mapping tree,cn=config".
[21/Apr/2016:12:49:20 -0400] NSMMReplicationPlugin - The replication
agreement named "cn=ds0 to ds1,cn=replica,cn=dc\3Dcise\2C dc\3Dufl\2C dc
\3Dedu,cn=mapping tree,cn=config" could not be correctly parsed. No
replication will occur with this replica.
[21/Apr/2016:12:50:09 -0400] NSMMReplicationPlugin - Replication
agreement for agmt="cn=ds0 to ds1" (ds1:389) could not be updated. For
replication to take place, please enable the suffix and restart the
server
[21/Apr/2016:12:50:09 -0400] NSMMReplicationPlugin - Replication
agreement for agmt="cn=ds0 to ds1" (ds1:389) could not be updated. For
replication to take place, please enable the suffix and restart the
server
What I saw in the logs that made me think this might be the problem:
[09/May/2016:16:52:10 -0400] NSMMReplicationPlugin -
replica_config_modify: replica does not exist for dc=cise,dc=ufl,dc=edu
The solution that worked was to remove the replica from all the servers,
and disable changelog on the partition.
Then renabled changelog, and re-set with multi-master (ds0/configuration
server first), I also changed the replica numbers ... just in case.
Re-initialized the replicas, and re-set up the Windows synchronization.
After that, the logs are looking normally, and replication is again
working.
Addison