Somewhere in the AdminGuide there is the statement that the BA DB process will
preempt all other processes.  We have in the past presumed this to be true, tho
we've never observed it in reality.

Out current hardware setup is the number of disk storage pools plus the DB is
one greater than the number of tape drives available.  This caused no problems
in V3R7 of earlier releases (in V2 and V3R1, there were two less drives than
disk pools), but since upgrading to V4R1 in January, we have experienced the
system going down with the log full about once a week.  Setting the log trigger
to 40% has not remedied the situation: if all disk storage pools are in
migration, the only other tape process I allow during the backup production
window, the incremental DB dump seems to wait until a drive frees naturally,
and with collocated tapepools, that can take longer that the logfile can hold
out.

The question is, did I miss something when installing the upgrade about setting
priorities for processes?  or was I just lucky that the system I inherited when
we upgraded to V2R1 never was busy enough to trigger the situation?  I should
add that I process about 800-850 clients a night on each of my servers.