server is in ERROR status after destroying node with active l3 agent for MOS 8.0
Affects | Status | Importance | Assigned to | Milestone | ||
---|---|---|---|---|---|---|
Mirantis OpenStack | Status tracked in 10.0.x | |||||
10.0.x |
Invalid
|
High
|
Timur Nurlygayanov | |||
8.0.x |
Invalid
|
High
|
Alexander Gromov | |||
9.x |
Invalid
|
High
|
Timur Nurlygayanov |
Bug Description
Environment:
3 controllers, 2 computes-cinder VLAN or VxLAN or VxLAN + L2POP
Steps to reproduce:
1. Check on what agents is router1:
neutron l3-agent-
2. If there isn't agent on the primary controller:
neutron l3-agent-
neutron l3-agent-router-add on_primary_agent_id router1
3. Destroy primary controller
virsh destroy <primary_
4. Wait some time until all agents are up neutron-agent-list
5. Check that all routers reschedule from primary controller:
neutron router-
6. Boot vm3 in network1
7. ping 8.8.8.8 from vm3
8. ping between vm1 and vm3 by internal ip
9. ping between vm1 and vm2 by floating ip
Expected results:
1) vm3 is successfully booted and is in ACTIVE status
2) pings are successful
Actual results:
Sometimes server is in ERROR status, sometimes pings are not successful. One of these conditions always doesn't work.
I have used 'test_shutdown_
Reproducibility:
Server is in ERROR state approximately in 30% of cases.
In other cases pings don't work.
tags: | added: area-nova |
Error message for instance: f757-4f12- add0-4c4cef0bd0 1e was re-scheduled: Timed out waiting for a reply to message ID b4eee6c75c274a9 3803f526007752c 26", "code": 500, "details": " File \"/usr/ lib/python2. 7/dist- packages/ nova/compute/ manager. py\", line 1907, in _do_build_ and_run_ instance filter_properties) File \"/usr/ lib/python2. 7/dist- packages/ nova/compute/ manager. py\", line 2060, in _build_ and_run_ instance instance_ uuid=instance. uuid, reason= six.text_ type(e) )", "created": "2016-05- 11T14:28: 30Z"
"Build of instance c2f69199-