I have installed VM 3.2.6 on several different physical boxes, all on the same sub-net
Each physical server only has one network card, so have setup one VM network with all function enabled.
What I am seeing is frequent problems with networking on my VMs running on these boxes.... for example my SSH sessions to the VM will often crash after about 10-15 mins of use (my SSH session to the physical host still works OK) Sometimes I cant even ping the virtual host, although it comes back after about 30 seconds or so.
My guest o/s isOEL 5, so looking at the /var/log/messages I see lots of errors like below:
Dec 3 01:53:15 dhcp-gb-192-168-0-89 avahi-daemon: Invalid response packet from host 192.168.0.89.
Dec 3 01:53:16 dhcp-gb-192-168-0-89 avahi-daemon: Invalid response packet from host 192.168.0.89.
Dec 3 01:53:18 dhcp-gb-192-168-0-89 avahi-daemon: Invalid response packet from host 192.168.0.89.
Dec 3 01:53:28 dhcp-gb-192-168-0-89 avahi-daemon: Invalid response packet from host 192.168.0.89.
Dec 3 01:54:03 dhcp-gb-192-168-0-89 avahi-daemon: Invalid query packet.
Dec 3 01:56:28 dhcp-gb-192-168-0-89 last message repeated 32 times
Dec 3 03:51:23 dhcp-gb-192-168-0-89 last message repeated 13 times
Dec 3 03:57:55 dhcp-gb-192-168-0-899 last message repeated 2 times
1. The servers are all setup with static IP addresses, so not sure what is connecting with DHCP ?
2. Any clues what to check or change to resolve the problem ?
1. How many gateway addresses do you have defined... and where are pinging the guest from? Same subnet or are the packets being routed in any form?
2. Not trying to burst your bubble... but if you only have a single 1GB nic.... then you're not going to be able to run very many VM guests at the same time. Bare minimum, you should have at least 2. The cluster heartbeat should be separated. A single 1GB connection can not pass more than 125mb per sec of data at any given time.
I found that disabling "avahi-daemon" gets rid of the messages in /var/log/messages
I'm still having real problem with network connectivity though. Every 10-15 mins all network activity drops and lose all JDBC, SQLNET, SSH, etc connections from all clients, even connections on same sub-net. For example, I can't even ping the VM server from the host server.
After 60-120 seconds, it all comes back to life and carries on for another 10-15 mins.
I am only expecting to run 1 VM guest on each physical machine and am not planning to do anything like move VMs across physical servers or anything like that. I'm not even using clustering, so would think my network should be up for the task.