Changing the Queue Block Size

Modifying the queue block size is a major change to the Replication Server configuration and affects all connections to the Replication Server. You must suspend log transfer and quiesce Replication Server.

1. To maintain data integrity, you must stop data flowing into the Replication Server you want to configure before you change the queue block size.:
   1. Suspend log transfer from all Replication Agents to the Replication Server you want to configure.
   2. Suspend all upstream log transfer from Replication Agents.
   3. Quiesce all upstream Replication Servers.
   4. Suspend all incoming routes to the Replication Server you want to configure.
   5. Quiesce the Replication Server you want to configure.
2. Use configure replication server with the set block_size to 'value' clause to set the queue block size on the Replication Server you want to configure. This command:
   1. Verifies that there is no subscription materialization in progress.
   2. Verifies that all log transfer is suspended.
   3. Verifies that all incoming routes are suspended.
   4. Verifies that the Replication Server is quiesced.
   5. Purges queues.
   6. Zeros the values in the rs_locater RSSD system table to allow Replication Agents to resend transactions that may have not been applied to the replicate database when you started the queue block size change procedure.
   7. Sets the queue block size to the value you entered.
   8. (Optional) If you include the with shutdown option, Replication Server shuts down. The queue block size change takes effect when you restart the Replication Server. Shutting down ensures that Replication Server clears all memory.
3. After changing the queue block size, resume data flow:
   1. If you used the with shutdown option, restart the Replication Server.
   2. Resume log transfer from Replication Agents.
   3. Resume all incoming routes.
4. Check for data loss at all downstream Replication Server RSSDs and data servers. Usually, there is data loss from the RSSD of the Replication Server you configured. Ignore the data loss from a replicate RSSD that receives data from the RSSD of the configured Replication Server.
   Follow the procedures to fix data loss at the data servers. If there is data loss at a RSSD, you see a message similar to this in the log of the affected Replication Server:

   E. 2010/02/12 14:12:58. ERROR #6067 SQM(102:0 primaryDS.rssd) - /sqmoqid.c(1071)
   Loss detected for replicateDS.rssd from primaryDS.RSSD

   replicateDS is the replicate data server name and primaryDS is the primary data server name.