MAC prefix for each vnet
- Thread starter ns33
- Start date
Not that I know of, maybe room for a contribution here.. However, technically, the MAC is linked to the VM/CT, not to the network it's hooked on, so that may not be a reasonable feature. Imagine a NIC whose MAC would change depending on the switch it's plugged in, I'd be unhappy 
I'm confused, why would it ever change past the creation? Once set, how often is it reconfigured anyway? Its prefix would only apply during creation, just like the current datacenter wide prefix applies. Changing the bridge in an adapter doesn't currently regenerate the mac, even if you change the datacenter level prefix; I wouldn't expect that workflow to change. It's just an on-creation feature.
It would be a huge benefit for anything that is auto deployed and the mac of the network defines the network config.
A contribution would be easy in theory. I'm assuming the sdn is written inRust which I'm just scratching the surface. Correction looks like Perl?
It would be a huge benefit for anything that is auto deployed and the mac of the network defines the network config.
A contribution would be easy in theory. I'm assuming the sdn is written in
Last edited:
Well, unless you actually restore a backup, it won't leave the cluster, that's quite a given. But a NIC can easily be replugged to another bridge.. I guess if it's more of a marker, and if you make sure you don't reconfigure the NIC it's fine, but there are many cases where you'd change that.. Creation (or clone) is easy, but you'd need to either say you don't care after, or deal with what would happen if it switches networks!
And yeah, random_ether_addr is written and mostly called (with prefix) in Perl, so you'd need to hack around it, but maybe ask on here, on pve-devel or on the bugtracker if that's something that would be interesting at all, as I'm not sure this is super relevant, but maybe that's just me.
And yeah, random_ether_addr is written and mostly called (with prefix) in Perl, so you'd need to hack around it, but maybe ask on here, on pve-devel or on the bugtracker if that's something that would be interesting at all, as I'm not sure this is super relevant, but maybe that's just me.
Last edited:
setting mac on vnet could be only usefull if you have an ip on the vnet. (so mainly when vnet is used as gateway for the vms).
It's only possible with evpn, as the ip/mac is the same on each vnet (anycast ip). Don't make sense to add feature for other zone type.
if really needed, you can override in /etc/network/interfaces with
I'll be merged with sdn generated config.
It's only possible with evpn, as the ip/mac is the same on each vnet (anycast ip). Don't make sense to add feature for other zone type.
if really needed, you can override in /etc/network/interfaces with
Code:
iface vnetxxx
hwaddress <mac>I'll be merged with sdn generated config.
Aw, I thought you meant the MAC for the VM/CT on the vnet, not the bridge itself, sorry.. and it could be interesting for any type of bridge then no?