Wrong port status after some operations
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
neutron |
Invalid
|
Undecided
|
Yan Songming |
Bug Description
The environment is, openstack master branch, all in one devstack, openvswitch agent, dhcp agent.
After some operations, a nova port turns to wrong status. This issue can be reproduced by following steps:
1. Remove ovs port from ovs bridge using ovs-vsctl del-port command, neutron port status changed to down, that's right;[1]
2. Update name of the port, the port turns to ACTIVE status, that's wrong.[2]
It seems a problem of DHCP provisioning block, when updating port name, a DHCP provisioning block was added, then DHCP agent complete this provisioning block and make the port ACTIVE.
[1] After step 1:
| admin_state_up | True | binding:host_id | c4 | binding:vif_type | ovs | binding:vnic_type | normal
| device_owner | compute:nova
| id | cad2e6a0-
| status | DOWN
| updated_at | 2017-04-
[2] After step 2:
| admin_state_up | True | binding:host_id | c4 | binding:vif_type | ovs | binding:vnic_type | normal | device_owner | compute:nova | id | cad2e6a0-
Changed in neutron: | |
assignee: | nobody → Yan Songming (songmingyan) |
Thanks for filing this bug. Would you please explain how the sequence of steps that you describe above reflect what may happen in a real deployment? Is this something that you have found or experienced in a real deployment?