Wrong ha_state, when l3-agent that host the active router is down

[wujun]

In an L3 HA Setup with multiple network nodes, we can query the agent hosting the Master HA router via l3-agent-list-hosting-router.

root@node1:~# neutron l3-agent-list-hosting-router demo-router
+--------------------------------------+-------+----------------+-------+----------+
| id | host | admin_state_up | alive | ha_state |
+--------------------------------------+-------+----------------+-------+----------+
| 58fbfcf3-6403-4388-b713-523595411de6 | node1 | True | :-) | active |
| a74be278-e428-41a4-a375-9888e9b99bcd | node2 | True | :-) | standby |
+--------------------------------------+-------+----------------+-------+----------+

Now, On the node1, I stop the neutron-l3-agent, and then check the state.

root@node1:~# neutron l3-agent-list-hosting-router demo-router
+--------------------------------------+-------+----------------+-------+----------+
| id | host | admin_state_up | alive | ha_state |
+--------------------------------------+-------+----------------+-------+----------+
| 58fbfcf3-6403-4388-b713-523595411de6 | node1 | True | xxx | standby |
| a74be278-e428-41a4-a375-9888e9b99bcd | node2 | True | :-) | standby |
+--------------------------------------+-------+----------------+-------+----------+

You can see that there is no "active" router, but north-south traffic is still though the node1 and the keepalived work normally. I think the ha_state of node1 shoud be "active".

[Victor Morales (electrocucaracha)]

Hi wujun,

Thanks for reporting it, I tried to reproduce this bug but I need some extra information for that, like the release that you're using and the relevant configuration values that you used.

[wujun (wujun)]

you can reproduce this bug by stable/newton release.

[Brian Haley (brian-haley)]

There have been a number of issues fixed in this area the past few releases, closing. If it still happens on a newer release please re-open.