SYSKEC@GSUVM1.BITNET (02/08/90)
Here is the situation:
MVS/XA 2.2.0 & JES 2.1.5
1 3420 online
7 offline
The online 3420 is allocated to Job A
Job B starts and needs 1 3420
IEF238D ... REPLY DEVICE NAME,'WAIT', OR 'CANCEL
Reply WAIT
IEF433d ... REPLY 'HOLD' or 'NOHOLD'
Reply HOLD
Vary another 3420 online
Job C starts - A 1 step job that writes to a 3420
The mount message never comes up
RMFWDM shows that JOB C is waiting exclusively for resource
SYSIEFSD.Q4 and that Job B owns it shared
When Job A completes Job B and Job C mount messages appear
I understand Job B is waiting on Job A's 3420, but I don't understand/like
Job C waiting. This is repeatable and does not happen if the replyto
IEF433D is NOHOLD. I do have a sneaking suspicion that the answer to
all this is 'works as designed', but I have'nt found any documentation
admitting it.
Keith Campbell
SYSKEC@GSUVM1.BITNET
Ga. State University
Systems ProgrammingLDW@USCMVSA.BITNET (Leonard D Woren) (02/08/90)
It's probably "working as designed." Replying 'HOLD' to that message is so bad that I use TSSO (an automated console operations tool) to always reply NOHOLD. Doing so causes the WTOR for "reply ... WAIT or CANCEL" to be reissued more often, but it's better than tying up everything waiting for a drive to be freed up. Leonard D. Woren Senior MVS Systems Programmer <LDW@USCMVSA.BITNET> <LDW@MVSA.USC.EDU> University of Southern California
SYSBILL@UKCC.BITNET (Bill Sallee) (02/08/90)
For the last few days I've been getting two copies of everything on this list--one from AKRONVM and another resent from TREARN. Bill Sallee University of Kentucky
PAPP@BCVMCMS.BITNET (John F. Papp) (02/08/90)
You're right. It is working as designed. The answer is in what you've told the system to do. When you reply "HOLD", you are saying you want the job to hold on to all current allocations and wait until this particular allocation can be resolved via the available devices at the time. It doesn't change when you bring on a new device and locks up allocation for that class device until one is freed. Generally, you want "NOHOLD" as a default. This will release all currently allocated resources and allow you to bring online similarly classed devices. Also, I would normally bring the device online via the request not a vary command. -John-
SYSKEC@GSUVM1.BITNET (02/09/90)
Yep, Its official according Level 2 and every reply I received it works as
designed. The messages manual's entry for 'IEF433D .. REPLY 'HOLD' OR 'NOHOLD'
could be a lot more informative!
A job waiting on a unit to come available is actually waiting for an eligible
unit to go through unallocation. The only place the waiting job is posted is
unallocation. Level 2 said they had closed an APAR and accepted a suggestion
to have a vary online of an eligible unit to post the waiting task.
Thanks for all the help
(and I promise to never
never answer HOLD)
Keith Campbell
SYSKEC.GSUVM|.BITNET
Ga. State University
Systems Support