IBM Mainframe Discussion List <IBM-MAIN@bama.ua.edu> wrote on 04/07/2010
02:37:52 AM:
> Last night (well, way into this morning) we had a severe WTO buffer
shortage
> that nobody saw. Here's why:
>
> *IOS071I 0500,**,CONSOLE, START PENDING 378
> *IOS002A 0500,NO PATHS AVAILABLE
> IOS2002I 0500 NO PATHS AVAILABLE 381
> STATUS FOR PATH(S) ED
> SUBCHANNEL PATH AVAILABLE, BUT DEVICE NOT OPERATIONAL (C0)
> NO FURTHER INFORMATION AVAILABLE OR UNKNOWN CONDITION (FF)
>
> 3.5 hours later:
>
> *IEA405E WTO BUFFER SHORTAGE - 80% FULL
> IEE889I 01.04.41 CONSOLE DISPLAY 111
> MSG: CURR=2401 LIM=3000 RPLY:CURR=18 LIM=99 SYS=yyyy PFK=00
> xxxxxx 46 COND=A AUTH=MASTER NBUF=2390
> 0500 AREA=Z,A MFORM=T,S,J
> xxxx DEL=RD RTME=1/4 RNUM=15 SEG=14 CON=N
> K S,DEL=R,L=xxxxxx
> IEE151I DELETE REQUEST INCONSISTENT-NO DISPLAY ON SCREEN
> COMMAND ISSUED IS K E,D,L=xxxxxx
>
> Severe WTO buffer shortage a short while later. No other IOS messages or
> CNZ messages (no cnz messages before the buffer shortage) until the
console
> was varied online again *hours* later (while still in severe wto
shortage):
> V 500,ONLINE
> IEE302I 0500 ONLINE
> LOGOFF - ISSUED BY IEECVFTG
> IEA406I WTO BUFFER SHORTAGE RELIEVED
>
> At the time of the IOS002A another system in another sysplex but on the
> same box got this:
> *IOS050I CHANNEL DETECTED ERROR ON 0500,**,**,**04
> CNZ4200I CONSOLE zzzzzz HAS FAILED. REASON=IOERR
>
> Now my question (mainly to IBM - will probably be faster than going
through a
> tedious ETR): Why the heck isn't the console address space notified that
no
> paths are available? Why is the console address space still issuing
> messages to
> that console? Why isn't console notified that the console 500 is
notavailabe
> anymore?
>
> Nobody was able to even *see* the shortage messages since nothing got
> displayed on that console anymore!!!!
>
> And if this is broken as designed, will I have to code an mpf exit that
makes
> sure all buffer shortage messages will be sent to all consoles in
> the sysplex?
> Operators will be extremely happy about that (since many don't even know
> where to see which system issued the message- reading a formatted syslog
> message appears to be a dying/lost art).
>
> Regards, Barbara Nitz
I am no IOS expert, but you're not likely to find one awake at this
time of night, so I will make a few (possibly incorrect) comments.
I think an IOS driver can request being posted back on a no paths
condition. That might be by setting
IOSCC3WE EQU X'20' ..Set by a driver to request deferred
condition code 3 posting (post code
CONSOLE is not an IOS driver. It uses EXCP for MCS consoles.
It looks like there is a corresponding bit in the IOB.
IOBCC3WE EQU X'40' - USER REQUESTS THAT IOS POST A X'6D' FOR A
* CONDITION CODE 3 ON ATTEMPTED I/O OPERATIONS
* (OS/VS2) (MDC310) @Z309P9A
CONSOLE probably does not set that bit (I don't see it used
by a console module in the BCP XREF).
So the when there are no paths to the console, the IOSB remains
queued to the UCB until the no paths condition gets resolved.
Now as to how the no paths condition got resolved without
generating some kind of interrupt to drive the queued request,
that might be a question for whatever is
providing the control unit function for the console device.
Apparently when you later issued the VARY ONLINE for the console
device, the condition had been resolved, and the simulated interrupt
from VARY ONLINE did the redrive for the queued request.
Jim Mulder z/OS System Test IBM Corp. Poughkeepsie, NY
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
