I've seen some commentary regarding "instability" of DSO data activation in 7.3, also issues related to master data activation (particularly with deadlocks). I'm specifically speaking of data, not activating an object.
My own experience has included intermittent errors with both. The error messages are not terribly helpful, and almost seem to be false errors. By that I mean, the parent job (for example in Attribute Change Run) reports an error activating an infoobject (presumably related to a child BIACTI* job), but the child job executed and ran successfully. Or in the case of DSO activation, the parent job reports a problem with a batch process confirmation error from a child, but the background BIBCTL job successfully runs to completion. Of course since the parent job reports an error, the process chain reports as red, and follow-on steps of course are never executed. Click repeat or repair and it moves on to the next step without applying any correction to data.
Tracing through active jobs at the time via SM37, it appears that we have enough background processes. I see nothing else running at that time that would cause a lock, we've been careful to separate processes/objects appropriately. This is the same process chain that ran with high reliability in BW v7.0.
I've dialed down the parallelization for both DSO Activation and Attr Change down from 3 processes to 2 (RSBATCH). Given the intermittent nature of the problem, it's too early to judge whether or not that's solved the issue.
I'm currently evaluating notes 1697407 and 1488759 (from SP8, we are on SP7), although there's not much description associated with either.
Wondering if others have had similar experiences, and if so, what may have helped.