RE: Disabling Qinter... (Follow up question)

I would say this describes the way the devices Actually are allocated much
better than I did, Tom.  It would be an asynchronous task, I'm sure, and
depending on how the internal tables are stored, and such you've said, it
Would HAFta be slightly unpredictable.  (Except the first SBS to allocate a
DEV should always get the lock, afaik...  Dunno how multiple cores handle
task dispatching, of course.)

Afaik.. Thaz a'least one Mystery down t'day done been Solved, Tom!...;-D


| -----Original Message-----
| From: midrange-l-bounces@xxxxxxxxxxxx
| [mailto:midrange-l-bounces@xxxxxxxxxxxx]On Behalf Of Tom Liotta
| Sent: Friday, March 05, 2004 8:23 PM
| To: midrange-l@xxxxxxxxxxxx
| Subject: RE: Disabling Qinter... (Follow up question)
|
|
| midrange-l-request@xxxxxxxxxxxx wrote:
|
| >   9. RE: Disabling Qinter... (Follow up question) (jt)
| >
| >So, the order the subsystems are started is the key, afaik.
| Again, that may
| >NOT be how it used to work but that's my recollection, and it
| Certainly may
| >not even apply any more as the OS sure Has Changed over the decades.
|
| I've had a slightly different impression over the years that
| somewhat explains the unpredictability of which subsystem gets
| the device. It isn't exactly which subsystem starts first,
| although that certainly can affect the outcome. I've thought it's
| more accurate that it's which subsystem attempts to allocate the
| device first, regardless of which starts first.
|
| Think of two subsystems started in a CL program one after the other:
|
|  ===> strsbs SBS1
|  ===> strsbs SBS2
|
| The first STRSBS command will complete almost immediately, and
| the second command is executed. We then have two subsystems
| starting up at the (more or less) same time. If you watch them
| perhaps via WRKACTJOB, you'll see them both running through
| devices trying to allocate them.
|
| Now, imagine SBS1 has all workstations listed for allocation
| while SBS2 is only defined to allocate device ZZZZZZZZ, which was
| created only recently after a whole bunch of other devices
| already existed.
|
| Since SBS2 has only the single device in its table, it almost
| certainly will be able to get it allocated before SBS1 has the
| time to get around to it. So even though SBS2 "started" last, it wins.
|
| I also have thought that device names, or identifiers, aren't
| necessarily stored with the subsystem description in alphabetical
| order, but instead in perhaps a FIFO order tied to the date/time
| it was first defined to the subsystem description. So, if
| ZZZZZZZZ had been the very first device created and assigned,
| SBS1 might be much more likely to get it before SBS2 can.
|
| Beyond the possible orderings of device identifiers, the handling
| of errors on various devices during allocation can slow the
| process significantly for one or the other subsystem. This would
| also decrease predictability.
|
| If somebody really knows how it works, it might be instructive to
| the rest of us.
|
| Tom Liotta
|
| --
| Tom Liotta
| The PowerTech Group, Inc.
| 19426 68th Avenue South
| Kent, WA 98032
| Phone  253-872-7788 x313
| Fax    253-872-7904
| http://www.powertech.com