Re: [Rdo-list] DNS resolver problems w/ instack-virt-setup
On Fri, Apr 17, 2015 at 06:15:37PM +0200, Marius Cornea wrote:
192.168.122.1 is set on the virbr0 interface on the host (default
libvirt net) where under/overcloud VMs are running.
It is also the address of the virbr0 interface *on the overcloud
nodes*.
(192.0.2.0/24 subnet). The overcloud nodes have one interface in
the
brbm bridge and route the traffic through the undercloud node. You can
check that default gw on overcloud nodes is 192.0.2.1 (eth1 of instack
VM).
That confirms what I said in my previous email:
That's not a useful address for either of these nodes, on which
external connectivity -- at least on the controller -- is via
eth0/br-ex on the 192.0.2.0/24 network.
The undercloud node masquerades all traffic coming from
192.0.2.0/24 so the overcloud nodes can get external connectivity,
including to 192.168.122.1 which handles the dns queries.
It doesn't. First, because 192.168.122.1 is set on the virbr0
interface on the overcloud controller node, traffic to this address
never leaves the host.
While the undercloud node does have masquerade rules in place:
# iptables -t nat -S | grep -i masquerade
-A POSTROUTING -s 192.0.2.0/24 -o eth0 -j MASQUERADE
-A BOOTSTACK_MASQ -s 192.0.2.0/24 ! -d 192.0.2.0/24 -j MASQUERADE
It doesn't have ip forwarding enabled:
# sysctl -a | grep ip_forward
net.ipv4.ip_forward = 0
No forwarding, so no masquerading.
--
Lars Kellogg-Stedman <lars(a)redhat.com> | larsks @ {freenode,twitter,github}
Cloud Engineering / OpenStack | http://blog.oddbit.com/
Attachments:
- signature.asc (application/pgp-signature — 819 bytes)