PreviousNext

Recovering Slave Replicas

Because slave replicas are not backed up, you must recreate a replica to restore a replica that is corrupted. To do so, use the following procedure:

1. Use standard UNIX commands to manually delete the replica's database files and master key file. To do this, delete all the files in the following locations:

· /opt/dcelocal/var/security/rgy_data

· /opt/dcelocal/var/security/.mkey

2. Use the set _s(sec) command to bind to the master and then the dcecp registry delete -force command to delete the replica from the master's replica list. The next two commands show how to bind to the master and then delete the replica

dcecp> set _s(sec) /.../musee.com/subsys/dce/sec/master
dcecp> registry delete /.../musee.com/subsys/dce/sec/art -force
dcecp>

3. Use standard UNIX commands to copy the file /opt/dcelocal/etc/security/pe_site from the machine running the master to the machine that will run the replica.

4. Use /etc/dce_config (or your provider's equivalent) on the replica machine to do the following:

a.. Stop DCE daemons.

b. Start DCE daemons.

c. Configure a security server replica. This configuration creates the replica's database and starts secd.

5. When you configure the replica in the previous step, you assign it a name. If you did not give this replica the same name that it previously had, the old name must be deleted from CDS by performing the following steps:

a. Deleting the replica's server entry name from /.:/subsys/dce/sec

b. Deleting the replica's name from the CDS group /.:/sec