We do not like the wasted time ether.  We have multiple scripts, two for
each phase of our daily run.  Each phase has a do it part and a check when
done part.

The do-it-part checks to see if it should run or skip.  We do this with a
script that gets renamed. For example if this is phase-1 then there is a
phase-1-ok script.  Rename it to phase-1-skip and phase one will be skipped.
Rename it to phase-1-stop and after phase-1 is done then the chain will be
broken.  If we run then execute commands to do phase-1 (such as setting pool
hi and low to zero to migrate).  If run or skip then create the
phase-1-check schedule in 5 min and rename phase-1-skip if necessary.

The phase-1-check script monitors the given processes.  If they still exist
then reschedule itself for later.  If done then check for phase-1-stop.  If
exists then rename and delete its own schedule.  If not stop then delete its
schedule and run phase-2-do-it.

To stop at any phase just rename the phase-x-ok to phase-x-stop.  To start
at any point just run phase-x-do-it.

Our phases are as follows:

backup disk pools to tape         (disk to tape backup)
migrate disk pools to tape        (disk to tape migrate)
backup tape primary to tape copy  (as a catch all  tape to tape)
backup db, plan, volhistory, etc
run expire
run reclaims

--
Phillip Ford
Phillip Ford
Senior Software Specialist
Corporate Computer Center
Schering-Plough Corp.
(901) 320-4462
(901) 320-4856 FAX
phillip.ford AT spcorp DOT com