your patches yet, I need to look a little bit how ressources
pool work.
- Mail original -
De: Dietmar Maurer diet...@proxmox.com
À: Alexandre DERUMIER aderum...@odiso.com
Cc: pve-devel@pve.proxmox.com
Envoyé: Mardi 3 Septembre 2013 06:37:12
Objet: RE: [pve-devel] [PATCH 2/2] setup
Do you mean, modify bridge configuration ? or change the vm network nic
config ?
The VM config can change completely, for example use vmbr1 instead of vmbr0.
Libvirt do it like this, setup bridges/dhcp/ippool (restart dnsmasq) , and vm
get ip through dhcp-config (without restart)
I
ok, so in this case, the ip shouldn't work anymore. (dhcp listen on a
different
pool of ip on each bridge right? and dhcp need an ip in the range of the
pool).
The ip is also changed. The pool definition is also allowed to change.
So we should deny changing bridge without changing ip.
DERUMIER aderum...@odiso.com
Cc: pve-devel@pve.proxmox.com
Envoyé: Mardi 3 Septembre 2013 09:55:36
Objet: RE: [pve-devel] [PATCH 2/2] setup DHCP server at vm_start
ok, so in this case, the ip shouldn't work anymore. (dhcp listen on a
different
pool of ip on each bridge right? and dhcp need an ip
IMHO, IP assignment to VMs is a completely different topic.
Ok, I really need to test your patches ;)
Yes, but my patches do not anything about IP assignment (you can assign any IP
to the VM).
It just add a framework to read/write /etc/pve/resource.cfg, using
the PVE::SectionConfig
Hi Dietmar, I'm a bit busy today, I'll look at it tomorrow.
(for vlan, I think user need to configure manually bridgevlan vmbrXvY in
/etc/network/interfaces,It shouldn't break the current implementation.
- Mail original -
De: Dietmar Maurer diet...@proxmox.com
À: Alexandre
Hi Dietmar, I'm a bit busy today, I'll look at it tomorrow.
OK, thanks!
(for vlan, I think user need to configure manually bridgevlan vmbrXvY in
/etc/network/interfaces,It shouldn't break the current implementation.
Maybe - but wanted to mention that this is not tested so far.
+ # fixme: howto pass gateway for that pool
--dhcp-option='option:router,192.168.2.1'
+ # fixme: howto pass additional dhcp options
+
--dhcp-option='option:optionnumber,optionvalue'
I'm thinking if restarting the dnsmasq daemon at each vm start is good ?
(what happen if multiple vm
+ # fixme: howto pass gateway for that pool
--dhcp-option='option:router,192.168.2.1'
+ # fixme: howto pass additional dhcp options
+
--dhcp-option='option:optionnumber,optionvalue'
Sure, but where to get that settings? I guess we need to extend the IPPool
to include that information.
Also, how does 'dhcp-script' option helps here?
with dhcp script, you don't need to restart dnsmasq deamon to change the
config, as the dhcp-script can do something like:
#dhcp-script.pl macaddress
parse vm config files
return ip
And what if bridge or vlan changes? Above only works
But As you say, I think that client will try to wait around 30s for a dhcp
response, so restart could be ok too.
(Just need to avoid parallel restart, and restart only on config change)
Would you like to take over development on that now?
I just wonder where we should put the resources
11 matches
Mail list logo