On 02/03/2012 03:55 PM, Bob Hoffman wrote:
On the virtual machine, which is having the issue, this is my eth0 (I have been commenting out things to see if anything changes the issue.)
Looks fine.
on the main machine which experiences no problems.
eth0 and 1 onboard, eth 2 add on card, all are the same except hwaddress, uuid, etc.
Looks fine, assuming of course that the DEVICE and HWADDR are different. :)
(tried playing with bonding mode, but got nowhere there, again, the host machine never has an issue)
Are you bonding for performance or reliability?
Bridge config looks fine, too.
not sure why vnet0 shows up in ifconfig, it is there and sometimes it is not depending on boot....
Are you using KVM? I assume so. Regardless, the 'vnetX' interface is the link between the VM and the bridge. Think of it as your virtual network cable. As such, it exists when the VM exists, and is deleted when the VM migrates or is shut down.
What does 'cat /proc/net/bonding/bond0' show on either node?
I added network and netmask to the eth0 on the vm and I think that may be it. I am surprised it worked right without that in there. Definitely the broadcast ip in ifconfig was wrong.
whee, we shall see if this fixes it.
fyi, the bond0 file worked well with the addresses commented out, something the internet was not sure of. Many said you needed them there even if bridging. I think not.
My issue appears to be with the virtual machine just not working via the network. I can only describe it as when your windows computer has not been used for a while and then the power management takes over and turns it down.
The eth is up, everything is fine and working via the local computer (via ipmi and virtual manager) however the network service has to be restarted whenever I have not done anything with the VM for an extended period of time (20 mins?, not sure).
The only thing I can do at this point is make a cron job that runs every 5 minutes and have a script that runs service network restart over and over and over and over again.
gotta be missing something.
I am thinking adding ethtool opts wol pumbg (wake on lan, all on)
the ethtool for eth0 on my vm said wol = d, never wake up.
so I set it in the ifcfg to just wake on everything.
will see, that may be the issue. Hopefully.
why in the world would the system set up a webserver install with a lan that is not alwys on though?