octavia: what is the loadbalancer IP assigned to? - openstack

Question

I am trying to understand how Octavia is put together. I created a loadbalancer on a vlan network. It was assigned an address of 10.40.0.7. When I do openstack loadbalancer list, I see a vip_address of 10.40.0.7 which is not assigned to any amphorae.
I want to understand where the loadbalancer address is mapped. It is not a host. I can't ssh to that address. Perhaps it is the amphora driver but what exactly is that? I can't see that address find it in any namespace. I can't see it assigned to any bridge. What is it assigned to?
Thanks
Ranga

Answer 1

It is not a host.
It is a host! An amphora is just a nova server -- the same thing you get when you run openstack server create. The difference is that the amphora is owned by the service project, so you'll only see it if you were to run (as admin) openstack server list --all-projects. For example:
$ openstack --os-cloud as_me loadbalancer list
+--------------------------------------+---------+----------------------------------+-------------+---------------------+----------+
| id | name | project_id | vip_address | provisioning_status | provider |
+--------------------------------------+---------+----------------------------------+-------------+---------------------+----------+
| 64a6a56d-beeb-4ee2-b495-1cdc26ffd399 | test_lb | 0ac1e30189da48b387cf3c2f5582b2a3 | 10.254.0.6 | ACTIVE | octavia |
+--------------------------------------+---------+----------------------------------+-------------+---------------------+----------+
$ openstack --os-cloud as_admin server list --all-projects | grep amphora
| f6cd75fe-8513-4aae-bee9-af9362525703 | amphora-50dddb41-decf-4b3b-bb7a-f35a751d74af | ACTIVE | lb-mgmt-net=172.24.0.16; test_lb_net=10.254.0.11; test_net1=10.0.1.5; test_net0=10.0.0.4 | octavia-amphora-13.0-20181107.1.x86_64 | octavia_65 |
If you look at that server, you'll see it has several ip addresses:
The one you assigned to it when created the loadbalancer, and
A management network address
Addresses on any subnets to which it is attached
You can ssh into the amphora using the management network address. You should be able to reach it from your controllers. You'll need the appropriate ssh key; where to find that probably depends a lot on how you installed things. I'm using tripleo, and it looks as if the install uses ~/.ssh/id_rsa from the stack user for the amphora ssh key.
[controller ~]$ ssh -i amphora_private_key cloud-user#172.24.0.7
Last login: Thu Nov 15 22:01:16 2018 from 172.24.0.6
[cloud-user#amphora-7d48e10b-5ba4-42c9-bcd5-941d224b2a46 ~]$
Update
The loadbalancer VIP is assigned to an interface inside a namespace on
the amphora. Given the above configuration, I see:
[root#amphora-50dddb41-decf-4b3b-bb7a-f35a751d74af ~]# ip netns
amphora-haproxy (id: 0)
[root#amphora-50dddb41-decf-4b3b-bb7a-f35a751d74af ~]# ip netns exec amphora-haproxy ip a
1: lo: <LOOPBACK> mtu 65536 qdisc noop state DOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc pfifo_fast state UP group default qlen 1000
link/ether fa:16:3e:07:d2:26 brd ff:ff:ff:ff:ff:ff
inet 10.254.0.11/24 brd 10.254.0.255 scope global eth1
valid_lft forever preferred_lft forever
inet 10.254.0.6/24 brd 10.254.0.255 scope global secondary eth1:0
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fe07:d226/64 scope link
valid_lft forever preferred_lft forever
4: eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc pfifo_fast state UP group default qlen 1000
link/ether fa:16:3e:21:9a:d1 brd ff:ff:ff:ff:ff:ff
inet 10.0.0.4/24 brd 10.0.0.255 scope global eth2
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fe21:9ad1/64 scope link
valid_lft forever preferred_lft forever
5: eth3: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc pfifo_fast state UP group default qlen 1000
link/ether fa:16:3e:2a:63:58 brd ff:ff:ff:ff:ff:ff
inet 10.0.1.5/24 brd 10.0.1.255 scope global eth3
valid_lft forever preferred_lft forever
inet6 fe80::f816:3eff:fe2a:6358/64 scope link
valid_lft forever preferred_lft forever