Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

high availability nova api

Hi. I'm using havana release with openstack. I have 2 controller nodes in HA (active-passive). On this node i have mysql server and rabbitmq with drbd filesystems, keystone, glance (api and registry), neutron-server, 2 vip (1 public, for apache and 1 private for management network) and apache server.

My question is: Should i use High-availability resource agents also for nova-api services? Can i run nova-api in non HA configuration on both controllers? How can i set my_ip flag in nova.conf in this case? and vncserver_proxyclient_address?? I would like to run nova-api on both controllers so when controller1 goes down, all other services are migrated on controller2 and nova-api on controller2 are used.

Is this correct at least in theory or I should use nova-api in HA? If this is correct, i can reuse also for apache (2 web server, one per controller node with openstack-dashboard keystone ip = VIP where keystone binds) Thanks and sorry if this is a bad idea