Re: dasd utilization at 99.56%

Pat Barber wrote:

I recall an IBMer giving a speech on this subject and according to
him, the system will stop allowing jobs to start but will continue to run.

It will also stop any sign on attempts except from the console.

I don't believe you can actually stop it dead, but certainly put the system in a position that it can no longer process work.

A very full system will continue to run in my opinion, although you
may not like the method it chooses to run.

Walden H. Leverich wrote:

When it runs out of space. Not when it's "almost out", or "nearly
out", or "damn close to out", but out.

A job will not start due to storage, only if that job requires storage that is not available when requested by work management in order to start the job; e.g. to create a new job message queue. There is no generic work management feature which tests against any particular threshold before starting interactive or batch jobs; although there are tests like that, for some specific jobs like some SNA/DS jobs, however those are done as a feature of the product\component owning the job rather than as part of the general job work management. What generally prevents signon is a ENDSBS *CTRLD, which might be a special-value or coded threshold action, in response to a threshold warning.

The system does not shutdown except where IIRC a special value [e.g. *PWRDWN] is the response to a threshold, or the coded response does so; see system values QSTG* Rob noted, which IIRC QSTGLOWACN system value offers the ability to define some special-value actions [*ENDSBS, *PWRDWN, etc.] or a programmed response [via exit program per *REGFAC] for messages in the CPI0900 range, perhaps CPI0954 or CPI099C.

The system /crashes/ when a critical system task or job fails to complete an action which can not [or is not coded to] tolerate that failure condition; i.e. *FC versus recoverable error. There are two specific MCH messages [mch4601 & ??] which define conditions of insufficient storage for the request, and either might be origin for a failure in a critical operation that is not tolerant, and so results in immediate abnormal shutdown of the system.

Regards, Chuck