test_rebuild_server_in_error_state randomly times out waiting for rebuilding instance to be active (cells v1)
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Compute (nova) |
Won't Fix
|
Low
|
Unassigned |
Bug Description
2017-08-10 18:58:35.158151 | tempest.
2017-08-10 18:58:35.158207 | -------
2017-08-10 18:58:35.158221 |
2017-08-10 18:58:35.158239 | Captured traceback:
2017-08-10 18:58:35.158258 | ~~~~~~~~~~~~~~~~~~~
2017-08-10 18:58:35.158281 | Traceback (most recent call last):
2017-08-10 18:58:35.158323 | File "tempest/
2017-08-10 18:58:35.158346 | raise_on_
2017-08-10 18:58:35.158381 | File "tempest/
2017-08-10 18:58:35.158407 | raise lib_exc.
2017-08-10 18:58:35.158436 | tempest.
2017-08-10 18:58:35.158525 | Details: (ServersAdminTe
Looks like this mostly shows up in cells v1 jobs, which wouldn't be surprising if we missed some state change due to the instance sync to the top level cell, but it's also happening sometimes in non-cells jobs. Could be a duplicate bug where we missing or don't get a network change / vif plug notification from neutron so we just wait forever.
Changed in nova: | |
status: | Confirmed → Won't Fix |
Actually the cells v1 job shouldn't be waiting for vif plug events, but it looks like nova-compute is configured for that:
Aug 10 17:49:38.949627 ubuntu- xenial- rax-ord- 10376969 nova-compute[ 20201]: DEBUG oslo_service. service [None req-e2b1deea- 3fbd-4806- 8dc4-fbfde2dbf7 70 None None] vif_plugging_ is_fatal = True {{(pid=20201) log_opt_values /usr/local/ lib/python2. 7/dist- packages/ oslo_config/ cfg.py: 2875}} xenial- rax-ord- 10376969 nova-compute[ 20201]: DEBUG oslo_service. service [None req-e2b1deea- 3fbd-4806- 8dc4-fbfde2dbf7 70 None None] vif_plugging_ timeout = 300 {{(pid=20201) log_opt_values /usr/local/ lib/python2. 7/dist- packages/ oslo_config/ cfg.py: 2875}}
Aug 10 17:49:38.949950 ubuntu-
http:// logs.openstack. org/12/ 491012/ 12/check/ gate-tempest- dsvm-cells- ubuntu- xenial/ 4aa3da8/ logs/screen- n-cpu.txt. gz
cells v1 doesn't support those - oh but cellsv1 jobs run with nova-network, so neutron shouldn't be a problem there.