| Re: Even worse thing when migrating online [message #36988 is a reply to message #36978] |
Thu, 06 August 2009 15:50   |
jeronimo
Messages: 4
Registered: August 2009
Location: Czech Republic
|
Junior Member |
|
|
Hi Swindmill,
| swindmill wrote on Wed, 05 August 2009 23:34 | This sounds like an issue I've seen where a bridge's HWaddr changes when interfaces are added or removed from it.
Other machines on the network still have the old HWaddr in their arp cache and cannot communicate with the bridged machine for some period of time after the change occurs.
My workaround is to force the bridge to share it's HWaddr with the physical network interface on the host (ethX) and update the HWaddr to match upon adding or removing an interface.
|
thank you very much! That was exactly the problem -- now, the online migration succeeds (previously, the ssh connection became lost and thus it has failed due to the MAC change) and everything is working fine.
More precisely, almost fine, since now I have to somehow persuade the switches to automatically and quickly learn the location of the migrated VE (currently, even if the VE has been migrated, all the pings coming outside of the VE's segment are "routed" to the source HW node -- as soon as I ping anything from the migrated VE, the switches learn this new location and "route" all the packets as suspected.
Once again, thank you for solving the issue.
Tom.
[Updated on: Thu, 06 August 2009 15:51] Report message to a moderator |
|
|
|
OpenVZ Forum
Members
Search
Help
Register
Login
Home
