Nicira plugin: get_port reports wrong port status for extended switches
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
neutron |
Fix Released
|
Medium
|
Salvatore Orlando | ||
Grizzly |
Fix Released
|
Medium
|
Salvatore Orlando |
Bug Description
When using the provider networks extension with the nicira plugin, a single, large, network might be backed by several NVP logical switches.
When reading the logical status for a port, the plugin invokes nvplib.
The Plugin leverages the fact that nvp logical switch uuid == quantum network id.
However, this is not always true in the case of extended switches. The operation then migth a 404 error if the port was being sought on the first logical switch of the chain, whereas it might be on a different logical switch. Its state is then set to ERROR whereas it should be 'ACTIVE' or 'DOWN'.
Steps to reproduce:
1) edit /etc/quantum/
2) restart quantum-server
3) quantum net-create bridged --provider:
4) quantum subnet-create bridged 192.168.0.0/16
5) quantum port-create bridged --name bridged-2; quantum port-create bridged --name bridged-3
bridged-3 will be created on a 2nd logical switch (the 3rd port on the 1st ls is the dhcp port)
6) user@os-
+------
| Field | Value |
+------
| admin_state_up | True |
| device_id | |
| device_owner | |
| fixed_ips | {"subnet_id": "1e568b32-
| id | 691d2cbc-
| mac_address | fa:16:3e:a9:db:0e |
| name | bridged-3 |
| network_id | 6ec622d4-
| port_security_
| security_groups | f0750245-
| status | ERROR |
| tenant_id | d6ddcd5d5d6f4c3
+------
Status should not be 'ERROR' but 'DOWN'
Changed in neutron: | |
status: | Fix Committed → Fix Released |
tags: | added: grizzly-backport-potential |
Changed in neutron: | |
milestone: | havana-2 → 2013.2 |
tags: | removed: grizzly-backport-potential in-stable-grizzly |
Fix proposed to branch: master /review. openstack. org/32155
Review: https:/