For both of the solutions described it is critical that the latest versions of DS.NLM and DSREPAIR.NLM are used on all servers. We have seen unexpected results of these operations with older versions of these modules.
Please realize that the advanced options switch for DSREPAIR (-a) should ONLY be used when instructed directly by Novell Technical Support or when specifically asked for a purpose described in a TID.
In a case of synthetic time on one or more partitions the following procedure addresses the problem :
- Load DSREPAIR.NLM with the -a switch on the server holding the master replica of the partition in question
- Select Advanced options menu, Replica and partition operations
- Select the partition and then "Repair time stamps and declare a new epoch"
- Log in (remember to use full distinguished name like .admin.containerx.org)
This operation will set the modification timestamps of all objects in the partition (and only in that partition) to the current time and set all replicas except the master to new state. A result of this will be that all objects will be sent to the readable (RW or RO) replicas and may - depending on the partition size - cause a considerable amount of network traffic during the partition operation.
If it is future timestamps on the schema, follow this procedure.
- Load DSREPAIR.NLM with the -a switch on the server holding the master replica of the [Root] partition
- Select Advanced options menu, Global schema operations
- Log in and select Declare a new epoch
This will set the modification timestamps of the complete schema (classes, attributes, schema partition) to the current time and synchronize the schema to all servers in the tree.