I thought I would report back here how this was resolved, in case someone follows this thread in the future.
We used the Hardware Management Console to take the CHPID offline from -all- systems, then placed it back online. At that point I used TCP/IP on VM to gain access to the OSA adapter using a strange portname. TCP/IP came up nicely. Next I confirmed that Linux was able to use the OSA card if I supplied the same portname. Furthermore, if I left Linux up and took TCP/IP down on VM and changed the portname, I found that TCP/IP on VM couldn't use the device. That pretty well settles it in my mind that someone out there on one of our many test systems was setting a portname for the card... and we didn't know about it. The good news is that now that I have set a strange portname, whoever that was is likely to come crying. This time we'll document who's using what and life will be easier in the future. My thanks to all who provided helpful ideas for resolving this! -----Original Message----- From: Scully, William Sent: Tuesday, January 29, 2002 1:43 PM To: [EMAIL PROTECTED] Subject: Re: OSA Hmm. I confess my changes to /etc/modules.conf included a portname in lower-case. I can retry this with upper-case. I'll also try to give one path to the OSA adapter to VM and one to Linux, both using the same pathname. Thanks for the idea. I'll report back here with my results! -----Original Message----- From: Post, Mark K [mailto:[EMAIL PROTECTED]] Sent: Tuesday, January 29, 2002 1:19 PM To: [EMAIL PROTECTED] Subject: Re: OSA William, Well, the "obvious" thing to try is have VM (or some other system) _specify_ a portname, so that can match that value (using all uppercase) on your Linux/390 system. :) Mark Post -----Original Message----- From: Scully, William [mailto:[EMAIL PROTECTED]] Sent: Tuesday, January 29, 2002 1:06 PM To: [EMAIL PROTECTED] Subject: Re: OSA The Linux systems are running under VM in an LPAR. Other nodes using the OSA adapters are in separate LPARs. I've asked and I'm told that no other system explicitly specifies a portname. I wanted to check this assertion. So we took one of the "failing" OSA adapters and gave it to TCP/IP on VM, which did initialize and successfully used the OSA adapter. Had OS/390 on another node specified a portname I would have expected that TCP/IP on VM would -not- be able to use the card. I'm not certain that this result is definitive but given what I read in the TCP/IP documentation for VM (which says the portname -MUST- match) I expect so. Thanks for the suggestion. Any other ideas would be appreciated! -----Original Message----- From: John P Taylor [mailto:[EMAIL PROTECTED]] Sent: Tuesday, January 29, 2002 12:44 PM To: [EMAIL PROTECTED] Subject: Re: OSA Is the VM system in an LPAR? If so and another system uses the GB ethernet adapter, which is IPLed before yours, then I believe that system would determine what the portname is. Just a thought. >> ...we're suddenly having trouble using our OSA adapters. The problem first surfaced after VM was recently restarted >> after the upgrade of the processor microcode... -- John P Taylor Linux & VM Systems Support Hursley IT Email: [EMAIL PROTECTED]
