The active NIC has been failing back and forth on two of our T2000s. I don't know when the problem originally began, but I can see failovers in the messages.* files. The network guy says the switches are clean (no errors) so I'm left looking at the configuration or any other issue that may be causing the symptom, and I'm at a loss on how to continue troubleshooting.
Here is a sample from messages.
IPMP is configured as follows:
Edit to add:
Hi,
use 127127-11 kernel patch or 128028-03 (single patch, obsoleted by: 127127-11 SunOS 5.10: e1000g patch) for help.
Thank you. I installed the 127127-11 kernel patch.
Are you able to fix this problem by applying patch?
One problem I've run into is a busy gateway. IPMP uses the first 5 destinations in your routing table to determine if the current running connection is valid. While there may not be errors on the router, it can get busy, and ping requests are given the lowest priority. As discussed on the page linked above, you can increase the timeout value in /etc/default/mpathd.
One way around this is to add static routes to your routing table to addresses that you know will not be delayed in a proper working environment. Not an ideal solution, but it works. This is explained in further detail here, if you have a sun online account.