Hi,
Wile installing  the fix-packs for SDS servers that are active/passive with peer-to-peer replication configuration,     

• do wee need to stop the replication on passive node?
• is there data loss if the replication is not suspended?
• Best process to apply the fix packs with out any data loss please...

Thanks
Latha

------------------------------
Latha Balne
------------------------------

Hello Latha,
Installing fixes (IFs/FPs) is an offline task with SDS, which means you need to stop ibmslapd and ibmdiradm processes on a given server which is going through fix pack installs.

While you are doing fix pack installs, better practice is to suspend any replication queues on all other servers coming into the server that's going through fix pack install. Even if you don't do that, when you stop ibmslapd on the server that's going through the fix pack install, the replication on all other servers leading to this server will go into a retrying mode (if there are any active updates happening on other servers). After fix pack install when you restart the ibmslapd process, you'd have to resume the replication from all other server (assuming you have suspended them before), OR the replication will automatically connect if its in retrying state and the replication will proceed.

There will be no data loss in either of these cases.

The best practice would be as below (assuming 4 servers in peer topology, namely server1, server2, server3 and server4):

1. Apply the fix pack to one server at a time in a given network topology. Lets say server1 is the candidate for fix pack install.
2. Suspend the replication from all other servers (server2, server3 and server4) to the identified single server (server1) that's going through fix pack install. This is an action that need to be performed either via web admin tool or via scripts (with idsldapexop for suspend replication) on other servers (server2, server3 and server4).
3. Stop ibmslapd and ibmdiradm processes on server1.
4. Install the fix pack on server1.
5. Start ibmslapd and ibmdiradm processes on server1.
6. Resume replication from server2, server3 and server4 by sending in idsldapexop to resume on these three servers.
7. Verify the replication status on server2, server3 and server4 to make sure the replication queue got cleared up and all the changes are sent.

Now you can repeat the process for the next server (server2) and so on.

I hope this clarifies.
Thanks,
Ram.

------------------------------
Ramamohan Reddy
------------------------------