Attention Light on new Power8 set up

We recently logged a Hardware PMR with IBM due to an attention light which kept coming on on the new Power8 system we have just set up. The light would come on at the same time every day and messages would be sent to the QSYSOPR message queue for all of the configured LPAR’s. None of the LPAR’s showed the the attention light in the vHMC, just the server. When we looked at the problem we could see a SRC B1A38B24 had been logged.

IBM looked at the issue with us and suggested that it was a Software issue not a Hardware issue. The service rep identified the error as coming from the vHMC due to a network configuration error, but had no experience with the vHMC so was not sure why. We did a manual reset of the attention light and agreed to wait for it to reappear.

I researched the problem via Google etc and found that the problem was due to each of the network interfaces for the HMC being on the same LAN segment yet when I looked at the configurations via the front panel (option 30) they appeared to be in totally different LAN segments (192.168.200.* and 169.254.2.*) ?

However when I looked at the network configuration in the ASMI it showed that I had 2 ports in the same LAN segment! this is where the problem was. When I had originally set up the system I had plugged one the HMC cables into the company LAN which had automatically assigned an IP address to it. I had also manually configured the other HMC port to allow access from the same LAN segment. When I changed the set up to be an open LAN set up I had should have reset the HMC ports back to their original settings. The new connection for the HMC is using ETH0 (one of the 4 ports on theĀ  ethernet card) not one of the HMC ports, but when the system did its verification of the set up it could see that one of the HMC ports (maybe both?) is in the same LAN segment as the ETH0 port. I used the ASMI reset the HMC ports to the original settings and left the system for 24 hours to see if the attention light reappeared or the notifications came back. We have seen nothing now for over 72 hours so that has probably sorted the issue.

The confusion for me is why the front panel interface (option 30) showed totally different information to the ASMI? It appears it was picking up a couple of ports from the Ethernet card which in itself makes sense but then why does it use the ASMI settings to trigger the error?

Chris…