I have a pretty large set of OLAP databases with quite a few daily
partitions in which more recently (apparently due to the size), when I
incrementally update the shared dimensions, for a short while after any MDX
queries result in #ERR. After waiting a 1/2 an hour or so, the same MDX
query comes back fine.

It appears that this may be due to the lazy aggregations processing -
normally, this isn't supposed to have any effect, but appears that it does
now (as per the reference:
http://msdn.microsoft.com/library/de...sions_55rn.asp,
where once dimensions are incrementally updated, the cubes that include the
dimension are refreshed automatically - both tasks are not supposed to
affect end-user access).

In trying to debug this, I have also updated the registry key for
ProcessAheadSegmentSize to insure that it is >70% of physical memory (along
with the high memory limit and other settings). Saying this, does any one
have any additional ideas how to resolve this?

Closest thing to a support reference I have found so far is:
http://support.microsoft.com/default.aspx?kbid=820358
TIA,
Denny