Jill, That looks right, so I guess that's not the problem. I'll have to think about this some more.
Mark Post -----Original Message----- From: Jill Grine [mailto:[EMAIL PROTECTED]] Sent: Friday, April 26, 2002 1:28 PM To: [EMAIL PROTECTED] Subject: Re: TCPIP - VM Guest GATEWAY for VSE versus Linux Hey Mark, Thanks for including command syntax. Kernel IP routing table Destination Gateway Genmask Flags Metric Ref Use Iface 10.1.2.120 0.0.0.0 255.255.255.255 UH 0 0 0 ctc0 0.0.0.0 10.1.2.120 0.0.0.0 UG 0 0 0 ctc0 -----Original Message----- From: Post, Mark K [mailto:[EMAIL PROTECTED]] Sent: Friday, April 26, 2002 12:51 PM To: [EMAIL PROTECTED] Subject: Re: TCPIP - VM Guest GATEWAY for VSE versus Linux Jill, What does your routing table on the Linux machine look like? route -n Mark Post -----Original Message----- From: Jill Grine [mailto:[EMAIL PROTECTED]] Sent: Friday, April 26, 2002 11:51 AM To: [EMAIL PROTECTED] Subject: Re: TCPIP - VM Guest GATEWAY for VSE versus Linux Hi Mark, On our system, all packets for 10.1.5.x are directed to 10.1.2.120, the VM machine. (Sorry, I left that piece out.) So I guess my question is there any reason I can't route to the Linux system the same way I do VSE? In other words, right now incoming packets for 10.1.5 get sent to 10.1.2.120 by the LAN. The VM TCP/IP stack redirects them to the appropriate system, either one of several VSEs or Linux. Is there any reason Linux should be treated differently than VSE? Alan, if you're out there, any chance this presentation Mark mentions is available on the net? Jill -----Original Message----- From: Post, Mark K [mailto:[EMAIL PROTECTED]] Sent: Friday, April 26, 2002 11:29 AM To: [EMAIL PROTECTED] Subject: Re: TCPIP - VM Guest GATEWAY for VSE versus Linux Jill, It isn't that TCP/IP is "different in Linux." The requirement is that the routers that connect to your VM system have to know where to route packets. If the router isn't told that packets for 10.1.5.150 have to go through your VM TCP/IP stack, it will effectively drop them. How you accomplish this can be done a number of ways: static routes in the router, dynamic routing protocols to advertise the routes, etc. Alan Altmark gives a very good presentation on this topic at SHARE. Unfortunately I haven't been given a copy to post on the linuxvm.org site, or I would point you to it. Mark Post -----Original Message----- From: Jill Grine [mailto:[EMAIL PROTECTED]] Sent: Friday, April 26, 2002 11:03 AM To: [EMAIL PROTECTED] Subject: TCPIP - VM Guest GATEWAY for VSE versus Linux Good morning all, Back again, but with a quick direct question. When we run VSE as a guest under VM, we code the TCP/IP GATEWAY statement for VM as follows: GATEWAY 10 = TR1 1500 0 10.1.5.130 10.1.1.120 CTCVSE 1500 HOST (10.1.1.120 is VM first hop) When I use the same syntax for Linux, it can't PING 10.1.1.120. 10.1.5.150 10.1.1.120 LNX1 1492 HOST I can ping the DNS server (10.1.3.140) and the linux machine 10.1.5.150. An earlier post said that we had to use a separate subnet for Linux, but a response said not so. The redbook Mark posted shows their Linux setup: HOME 9.12.14.155 WTSCVMT 9.12.9.178 LINUX5V GATEWAY 9.12.9.184 = LINUX5V 1500 HOST Is this the part we OS/390 folks were warned about with regards to TCPIP being different in Linux? The explanation for the VSE setup was that incoming packets would route to the VM TCP/IP stack (first hop), then get sent forward to VSE. Can someone clarify for SuSE running under VM as guest? Thanks as always, Jill Grine
