When starting data groups that share an object send process, the first data group to start will start the shared job at that data group’s starting point in the system journal (QAUDJRN). As additional data groups start, each recognizes that the shared object send job is active. The object send job determines whether the starting point for that data group is earlier or later than the sequence number being read. If the data group’s starting point is later, replication will begin when the shared job reaches the data group's starting point. If the data group’s starting point is earlier, the shared job completes its current block of entries, then returns to the earliest point for any of the data groups being started. The shared job reads the earlier entries and routes the transactions to the data group being started. When the shared job reaches the last entry it read at the time of the STRDG request, it resumes routing transactions to all active data groups using the shared job.
If the starting data group has a significant object send backlog, the other data groups sharing the job will not receive transactions to replicate while the backlog for the starting data group is being addressed. Therefore, when a significant backlog exists, it is recommended that you change the data group configuration to use a dedicated job (*DGDFN for object send prefix), start the data group, and allow it to catch up to the current location of the shared job. Then end the data group, change its configuration to use the desired shared job, and restart the data group.