This is a first cut talking to a lab MVS system to verify that their TCPNJE is viable. The final configuration has not been established. It will probably include from 3 to 10 MVS systems and 2 VMs.
KEEPALIV is the default for the PARM (CONFIG file) or DEFINE (command) for a TCPNJE link. Obviously z/OS decided to do something different. Regards, Richard Schuh -----Original Message----- From: The IBM z/VM Operating System [mailto:[EMAIL PROTECTED] On Behalf Of Alan Altmark Sent: Monday, March 19, 2007 3:30 PM To: IBMVM@LISTSERV.UARK.EDU Subject: Re: TCPNJE On Monday, 03/19/2007 at 11:50 MST, "Schuh, Richard" <[EMAIL PROTECTED]> wrote: > What happened? Did you forget about a NOT operator somewhere? The link > stays alive with KEEPALIV=NO and drops with KEEPALIV=YES. Reminds me > of the OS2 dialog boxes. "You told me to do x. Reply YES to not do it; > NO to go ahead and do it." :-) So, back in the Dark Ages, we used autodial modems with *real* inactivity timers. If you want to keep the modem going and, hence, your connection, you used Keepalive packets. These days their only purpose is to verify that the path is 'alive' and to tell your application if not. I've got good news and bad news. The good news is that we have keepalive packets. The bad news is that we have keepalive packets. Now you see first hand why they are not generally accepted into all TCP implementations. It also tells me you have only a single route between your VM and MVS system. NG. This is the worst environment in which to use keepalive packets. With two or more paths, a failed keepalive indicates a *major* problem, not just a glitch as someone fiddles with a port definition somewhere. Alan Altmark z/VM Development IBM Endicott
