Ok.. We hit this problem late last year on an IBM z9BC. I supppose the PMR is 
still out there on the database if anyone's got access to it.

I just dug out our internal problem log for the morning's events and looked 
through the timeline. Our IBM support engineer first took the device offline on 
all LPARs, and that did not resolve the problem. We then IPL'd each LPAR; 
again, that did not resolve the problem. We then did a POR of the box, and the 
devices were then available.

This is the email they (IBM) sent me:

"Here is the information about the problem. 

This a known problem and there is a microcode fix available for your machine.   

                                                                
The intent of this TDR is to provide information to determine if 
Segmentation Offload is enabled, and how to disable it.         
                                                                
                        
                                                                
Symptoms:  OSA Card will lock up. 

                                                                
                                                              
Check for Segmentation Offload being active:                 
                                                              
>From the z/OS console on each Image, enter                   
 D TCPIP,procname,NETSTAT,DEV.                               
 If Segmentation offload is functioning, you will see         
 SEGMENTATIONOFFLOAD:  YES for the OSA.                       
 If this specific line is not displayed, segmentation offload 
 is not available. at this level or previously disabled       
 There is not a NO option for SEGMENTATIONOFFLOAD.           
 Its either YES, or not displayed.                           
                                                              
There are two ways to disable segmentation offload.           
The maintenance level on the system determines which         
one must be used 
If your customer has applied PK21685 or PK26905,             
it must be done via the configuration file.                 
                                                            
1) If the PTFs are not applied, use the following procedure: 
Update the TCPIP PROFILE to comment out the START           
statements for the affected OSA devices.                     
                                                            
Issue the following commands every time TCPIP is restarted: 
MODIFY tcpipprocname,NOSEGOFFLOAD                           
 (then for each OSA)                                         
VARY TCPIP,tcpipjobname,START,osa_device_name               
(tcpipjobname is optional on this command in a single stack 
environment)                                                 
                                                            
If the MODIFY command is rejected with a                     
EZZ3250I THE MODIFY COMMAND IS NOT SUPPORTED message, 
Then either:                                                   
The system does not support Segmentation Offload.               
Only z/OS 1.7 and above or z/OS 1.6 with the PK02490 PTFs       
applied have this support, earlier systems do not have any     
concern with this problem.                                     
The PTFs for PK21685 have been applied,                         
thus the following method needs to be used instead.             
                                                                
2) If the PTFs are applied,                                     
add the NOSEGMENTATIONOFFLOAD keyword to the GLOBALCONFIG       
statement in the TCPIP PROFILE.                                 
For further help with these commands, contact software support."

-----Original Message-----
From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED]
Behalf Of Scott Rowe
Sent: 09 April 2008 20:40
To: IBM-MAIN@BAMA.UA.EDU
Subject: Re: z/OS 1.4->1.7 gotchas

???
 
AFIAK there is no fix, microcode or otherwise, for the segmentation offload 
problem.  I think the default has since been changed been changed to off.  
Also, it did not affect "old" OSA cards, it affects ALL OSA Express cards.  
There was also no need to POR, or even IPL, to recover the card, all that is 
needed is to configure it offline to ALL LPARs, and then back on. 


-----------------------------------------
Email sent from www.virginmedia.com/email
Virus-checked using McAfee(R) Software and scanned for spam

----------------------------------------------------------------------
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

Reply via email to