When Lb agent reconciles due to health check failure , service is in "updating-active" state.

[sangeethah]

Server version - V0.46.0-rc2

Steps to reproduce the problem:

Setup: 3 hosts

Service web1 - scale 2
Service web2 - scale 2
LB service - scale 2

Bring down one of the host which has 1 Lb Agent and 1 instance of service web2.

instance in web2 is not not able to reconcile successfully . The unhealthy instance continues to be in "running" state , since the new instance that was started in the host that is in "reconnecting" state is stuck in "stopping" state forever.

mysql> select id,name,state, health_state from instance where name like "%web2%";
+------+----------------+----------+--------------+
| id   | name           | state    | health_state |
+------+----------------+----------+--------------+
| 1453 | Default_web2_1 | running  | healthy      |
| 1459 | Default_web2_2 | running  | unhealthy    |
| 1468 | Default_web2_3 | stopping | initializing |
+------+----------------+----------+--------------+
3 rows in set (0.01 sec)

mysql> 

Also the LB agent that is marked "unhealthy" reconciles . But the unhealthy instance continues to be in "running" state and the Lb service is stuck in "update-activating" state forever.

| 1455 | Lb Agent | running | unhealthy    | 2015-11-16 18:00:40 |
| 1457 | Lb Agent | running | healthy      | 2015-11-16 18:00:41 |
| 1469 | Lb Agent | running | healthy      | 2015-11-16 18:17:45 |
+------+----------+---------+--------------+---------------------+

[sangeethah]

Tested with latest build from master:

When Lb agent reconciles due to health check failure , service gets to "active" state as expected.