Hi People,

I'm having a bit of a problem with websphere MQ peer-peer replication on a pair of DB2 servers I have. Replication works fine when the capture and apply queues are running on both servers and you add a row to the table, but if a row (or several million!) is added when one server is down, your restart replication and that data is not sent across to the other server. Any new data is. Does any one have any idea? I guess there's somekind of sync'ing command, but I can't find it!

Many thanks,
Ian

if capture is stopped normally it keeps the last handled log entry, and when restarted in warm mode, the capture continues reading the log from last stop point. apply does not need to run at all time, as captured records are held in mq queues waiting to be handled ...
full refresh can be forced, but will copy the complete table again....(not recommended)

Thanks Paul.

The situation I have is that one of my servers died overnight, so capture and apply died. I have rebooted this server and restarted the channel,capture,apply,etc, but there are 100000 rows missing from the database that was down overnight. I would have expected this server to catch up with all the old stuff when I brought it back up, but it does not. Any idea why?

Cheers,
Ian

should have done so.
if capture died, it would redo some updates already done
if apply died, it would have todo a redo as rows have not been flagged as executed
any errors in applytrail or other logs ?
seems not correct anyhow and needs investigation to avoid data loss
the best would be to force now a full refresh, to have target in synch with source.

I thought it was probably supposed to! I've had a look at the error messages, and it is showing 'The REINIT_SUB command is no valid for Q subscription "LOADTEST0002". The Q was not reinistialized.'

Last question (I hope!!), Do you know why this would be?

Thanks for all your help,
Ian

ARGHHH! Just found out that reinitialisation is only for unidirectional queues! Completely stuck again now!

I am experiencing a similar problem where somehow some changes where not applied and now my databases are no longer identical. since i noticed the difference, this has not happened again and all updates where captured and applied properly. but, i need them to be identical.

przytula mentioned that I can do a full-refresh. I have never before, How is it done?

I've no idea, but I fixed my problem by setting up replication again. The details we were sent for setting up replication queue maps were slightly incorrect, but working great now. Just got to speed it up a bit now