> -----Original Message-----
> From: IBM Mainframe Discussion List On Behalf Of Gibney, Dave
> [ moved below ]
> -----Original Message-----
> From: IBM Mainframe Discussion List On Behalf Of Chase, John
>
> > -----Original Message-----
> > From: IBM Mainframe Discussion List On Behalf Of Brian Peterson
> >
> > This morning, I checked the Exception Letters for our z9 and z10
> > processors.
> > Both Exception Letters are dated May 28, 2008, and both
> still say that
> > Segmentation Offload needs to be disabled.
> >
> > I am not saying that IBM hasn't fixed the problem - I hope
> they have.
> > But, the Exception Letters still say the problem is NOT fixed.
>
> Interesting.... They're dated three days LATER than the date
> we installed the latest ucode (that was supposed to fix the problem).
>
> We enabled SEGMENTATIONOFFLOAD on the sandbox (z/OS 1.9)
> yesterday, but there's very little TCPIP load on the sandbox
> (still doing IVP of 1.9, and post-installation "cleanup").
> Maybe we can cobble up something that should "stress" it to
> the breaking point (maybe FTP-ing some SMF offload datasets
> or sysdumps?).
>
> Meanwhile, we'll leave it disabled on the other LPARs until
> we hear something more reassuring (or determine that we can't
> "break" it on the sandbox).
>
> -----Original Message-----
> From: IBM Mainframe Discussion List On Behalf Of Gibney, Dave
>
> We ran for some years before it bit us in the ass. And I
> couldn't really find any load related cause at that time. It
> seemed a random strike out of the blue.
>
> So, I wouldn't count on reproducing the error in any
> controlled fashion :(
PK64756 (open APAR, last updated 5/30/08) refers to a previous APAR
PK56723 (PTFs available Feb. 08 for z/OS 1.8 and 1.9) that fixes one
problem with segmentation offload.
PK64756 says essentially that the delays still occur if segmentation
offload is enabled *AND* packet tracing is enabled. I didn't see that
"and" in the Exception Letters.
Anyway, as I interpret the APARs, the original problem wasn't because of
"load" but rather of "packet size", and confined to situations when
segmentation was offloaded when it shouldn't have been.
"The plit thockens."
-jc-
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
