On Friday, 01/23/2004 at 12:17 CST, "Lucius, Leland" <[EMAIL PROTECTED]> wrote: > I've been running into a problem every so often that I can't track down and > thought I'd just throw it out here to see if I get any bites. > > I have 2 Gbit OSAs that I share among my guests and use VIPA to provide > failover in case one goes down. So, every guest has an eth0, eth1, and an > lo:1 interface. All access to the guests use the IP address assigned to > lo:1. > > The problem is that sometimes a guest is unreachable using the lo:1 IP addr. > You can still access the guest through eth0 or eth1. I've found 2 ways to > get out of the situation. The first, is to drop all the interfaces and > bring them back up. The other is to use arping like so:
You need to figure out if the cards are responding to ARPs or not. I realize that you used qetharp, but that tells you what OSA has loaded into its own ARP cache, not whether it is responding to ARPs. Use OSA/SF or the HMC to look at IP addresses loaded into the cards (which is not the same as looking at the ARP cache). Of course, you can another host to figure out if OSA is responding to ARPs by looking in the other host's ARP cache. Alan Altmark Sr. Software Engineer IBM z/VM Development
