yahoo-eng-team team mailing list archive
-
yahoo-eng-team team
-
Mailing list archive
-
Message #95682
[Bug 2106357] [NEW] Loadbalancer member list show operating status as ONLINE when member is created for a SHUTOFF state vm
Public bug reported:
Step to reproduce the issue.
1. Create a Load Balancer with:
A listener on port 80 using TCP protocol
A pool with health monitoring enabled
2. Add a VM (which is Active) to the pool as a member:
The load balancer works correctly
The health monitor status is accurate (shows the VM as online)
3. Now add a second VM to the pool, but this VM is in Stopped state:
The health monitor still shows this VM as online
The load balancer continues to send traffic to the stopped VM
The stopped VM should not receive any traffic, but it does.
This breaks the expected behavior of the load balancer because:
The health monitor is not correctly detecting that the VM is down
Traffic is being routed to a non-functional backend
On ovn-sb db it show as
root@ovn-ovsdb-sb-0:/# ovn-sbctl list service_monitor
_uuid : 3e1e2b40-1a5b-4ae4-b1a4-a170b76517ec
external_ids : {}
ip : "192.168.0.11"
logical_port : "635b68bc-5339-46fb-890d-d8a0ab341b50"
options : {failure_count="3", interval="10", success_count="3", timeout="5"}
port : 80
protocol : tcp
src_ip : "192.168.0.194"
src_mac : "56:17:53:f0:6b:7f"
status : online
_uuid : dfcc8077-919e-41a3-8091-d9ea03759ac6
external_ids : {}
ip : "192.168.0.218"
logical_port : "38c022fb-bc84-49d6-9ad1-81a4d572f17e"
options : {failure_count="3", interval="10", success_count="3", timeout="5"}
port : 80
protocol : tcp
src_ip : "192.168.0.194"
src_mac : "56:17:53:f0:6b:7f"
status : []
opnestack CLI shows
openstack loadbalancer member list lb2-pool
+--------------------------------------+------+----------------------------------+---------------------+---------------+---------------+------------------+--------+
| id | name | project_id | provisioning_status | address | protocol_port | operating_status | weight |
+--------------------------------------+------+----------------------------------+---------------------+---------------+---------------+------------------+--------+
| 5eb42fb9-2456-4727-af9e-8e69294672d0 | mem2 | 8d9a93312d624938bff32dda4bd8ae97 | ACTIVE | 192.168.0.11 | 80 | ONLINE | 1 |
| 49be36a6-dbfa-43ea-9e5a-61d187ccc80d | mem3 | 8d9a93312d624938bff32dda4bd8ae97 | ACTIVE | 192.168.0.218 | 80 | ONLINE | 1 |
+--------------------------------------+------+----------------------------------+---------------------+---------------+---------------+------------------+--------+
** Affects: neutron
Importance: Undecided
Status: New
--
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/2106357
Title:
Loadbalancer member list show operating status as ONLINE when member
is created for a SHUTOFF state vm
Status in neutron:
New
Bug description:
Step to reproduce the issue.
1. Create a Load Balancer with:
A listener on port 80 using TCP protocol
A pool with health monitoring enabled
2. Add a VM (which is Active) to the pool as a member:
The load balancer works correctly
The health monitor status is accurate (shows the VM as online)
3. Now add a second VM to the pool, but this VM is in Stopped state:
The health monitor still shows this VM as online
The load balancer continues to send traffic to the stopped VM
The stopped VM should not receive any traffic, but it does.
This breaks the expected behavior of the load balancer because:
The health monitor is not correctly detecting that the VM is down
Traffic is being routed to a non-functional backend
On ovn-sb db it show as
root@ovn-ovsdb-sb-0:/# ovn-sbctl list service_monitor
_uuid : 3e1e2b40-1a5b-4ae4-b1a4-a170b76517ec
external_ids : {}
ip : "192.168.0.11"
logical_port : "635b68bc-5339-46fb-890d-d8a0ab341b50"
options : {failure_count="3", interval="10", success_count="3", timeout="5"}
port : 80
protocol : tcp
src_ip : "192.168.0.194"
src_mac : "56:17:53:f0:6b:7f"
status : online
_uuid : dfcc8077-919e-41a3-8091-d9ea03759ac6
external_ids : {}
ip : "192.168.0.218"
logical_port : "38c022fb-bc84-49d6-9ad1-81a4d572f17e"
options : {failure_count="3", interval="10", success_count="3", timeout="5"}
port : 80
protocol : tcp
src_ip : "192.168.0.194"
src_mac : "56:17:53:f0:6b:7f"
status : []
opnestack CLI shows
openstack loadbalancer member list lb2-pool
+--------------------------------------+------+----------------------------------+---------------------+---------------+---------------+------------------+--------+
| id | name | project_id | provisioning_status | address | protocol_port | operating_status | weight |
+--------------------------------------+------+----------------------------------+---------------------+---------------+---------------+------------------+--------+
| 5eb42fb9-2456-4727-af9e-8e69294672d0 | mem2 | 8d9a93312d624938bff32dda4bd8ae97 | ACTIVE | 192.168.0.11 | 80 | ONLINE | 1 |
| 49be36a6-dbfa-43ea-9e5a-61d187ccc80d | mem3 | 8d9a93312d624938bff32dda4bd8ae97 | ACTIVE | 192.168.0.218 | 80 | ONLINE | 1 |
+--------------------------------------+------+----------------------------------+---------------------+---------------+---------------+------------------+--------+
To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/2106357/+subscriptions
