> On 30 Sep 2015, at 3:48 pm, John Dewey <[email protected]> wrote:
> 
> Why not run neutron dhcp agents on both nodes? 

Yeah have tried this too.

We don’t do this due to metadata, DHCP adds a static route to the metadata 
agent (which is the same as the network node running DHCP). If the network node 
is down the instances loose metadata if they happened to get a lease from that 
DHCP server. With failover the IP address transfers to the other network node 
and metadata keeps working for all.

Make sense?

Sam


_______________________________________________
OpenStack-operators mailing list
[email protected]
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators

Reply via email to