--- Begin Message --- "I thought: adding the ceph IP on the bond0 would be a nice and easy way to seperate ceph traffic from the VMs."

If there is a bridge on that bond You will eventually not separate traffic.

It is like if You've put several cables into the same network switch (aka: bridged the ports together).

VLANs or physical network separation are good ways to separate traffic.

On 3/23/2021 12:36 PM, mj wrote:
Hi Dorsy,

Thanks for the quick reply! :-)

On 23/03/2021 11:51, dorsy wrote:
Also, if You look at the examples, they do not use a bond and VM bridge with 2 addresses, in the first, an IP is on the bond, the VM bridge is another physical IF.

The second example shows the VM bridge is over the bond, and the IP is on the bridge IF (no IP on the bond there).

Yes, I realise they are not identical. (that's why I said: freely based on..)

I thought: adding the ceph IP on the bond0 would be a nice and easy way to seperate ceph traffic from the VMs.

I have tried now as you suggested, and that works, yes. Thank you!

iface vmbr0 inet static
?????????????? address 192.168.143.10/24
?????????????? gateway 192.168.143.1
?????????????? bridge-ports bond0
?????????????? bridge-stp off
?????????????? bridge-fd 0
?????????????? post-up /sbin/ip addr add 10.0.0.10/24 dev vmbr0

I just remain curious why it would be so strange to put the IP on bond0. I do see most examples on the net NOT having an IP on bond0. So I understand it's not normal.

But what's wrong with it?

I tried also putting the "post-up addr add" stanza to bond0 config, but it doesn't work as well. (strange, given that adding it works, after boot has finished)

I will use your suggestion, thanks, appreciated.

But still: Why is putting an ip on bond0 considered strange, and why doesn't it work *during* boot, and does it work *after* boot?

MJ


--- End Message ---
_______________________________________________
pve-user mailing list
[email protected]
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pve-user

Reply via email to