On Tue, Apr 21, 2015 at 10:20 AM, Robert A. Rosenberg <[email protected]>
wrote:

> At 06:50 -0500 on 04/21/2015, John McKown wrote about Re: ENQ for the life
> of the job:
>
>  ŠThe initiator (in general terms) is what reads the Šparsed JCL and
>> creates
>> the SWA control blocks which represent the job. This code then knows the
>> DSNs in the job and issues a single ENQ for _all_ of them before starting
>> the first step. As each step ends, the initiator does a DEQ on the DSNs
>> which are not going to be used in subsequent steps. At the end of the job,
>> it DEQs whatever DSNs are left.
>>
>
>
> While this description of ENQ and DEQ handling is correct it leaves out
> the fact that due to a IMO major design flaw in the ENQ process the
> Initiator can be forced to hold an ENQ for subsequent steps where it is no
> longer needed. The case I am talking about is there is no way to convert an
> EXC ENQ into a SHR one. The Initiator is forced to keep an EXC ENQ active
> for steps where only SHR is what is enough/desired. This design flaw in ENQ
> mains that if a DSN is DISP=OLD/MOD in step 1 and DISP=SHR in all the
> following steps the DISP=OLD/MOD ENQ is unnecessarily held until the last
> DISP=SHR step (or the JOB) is over. This prevents other jobs that should be
> allowed to run from being allowed to run due to the no longer needed EXC
> ENQ being held during DISP=SHR steps.
>
>
> In theory allowing the EXC->SHR ENQ change would be simple since all that
> is needed is to alter the ENQ entry in the queue from EXC to SHR and then
> drive the part of the DEQ routine that runs the queue to release pending
> SHR entries that are waiting for the DEQ of the EXC ENQ at the head of the
> chain.


​The ability to atomically downgrade an ENQ from EXC to SHR is now/soon to
be available in z/OS. I don't remember if it is a PTF or release 2.1.1 . ​

-- 
If you sent twitter messages while exploring, are you on a textpedition?

He's about as useful as a wax frying pan.

10 to the 12th power microphones = 1 Megaphone

Maranatha! <><
John McKown

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to