Comment 14 for bug 1837792

Angie Wang (angiewang) wrote :

Hi Mariano,

I took a look the logs you provided. It's not due to compute-kits services(neutron, nova, placement) db-sync, they all succeed.

The armada log shows the error which indicates something wrong with openvswitch ..

2019-10-10 21:32:40.460 25301 ERROR armada.handlers.armada armada.exceptions.k8s_exceptions.KubernetesWatchTimeoutException: Timed out waiting for pods (namespace=openstack, labels=(release_group=osh-openstack-openvswitch)). None found! Are `wait.labels` correct? Does `wait.resources` need to exclude `type: pod`?

-bash-4.2$ grep -R "label" var/log/bash.log
2019-10-07T18:10:19.000 localhost -sh: info HISTORY: PID=19544 UID=42425 system host-label-assign controller-0 sriovdp=enabled
2019-10-07T18:13:03.000 localhost -sh: info HISTORY: PID=19544 UID=42425 system host-label-assign controller-0 openstack-control-plane=enabled
2019-10-07T18:13:05.000 localhost -sh: info HISTORY: PID=19544 UID=42425 system host-label-assign controller-0 openstack-compute-node=enabled
2019-10-07T18:13:12.000 localhost -sh: info HISTORY: PID=19544 UID=42425 system host-label-assign controller-0 openvswitch=disabled
2019-10-07T18:13:18.000 localhost -sh: info HISTORY: PID=19544 UID=42425 system host-label-assign controller-0 sriov=enabled

Looks like you didn't enable "openvswitch" label. By default, vswitch runs in container which requires the
"openvswitch" label to be applied.

You can use the following command to verify you have openvswitch enabled.
[sysadmin@controller-0 ~(keystone_admin)]$ system helm-override-list stx-openstack --long
| openvswitch | [u'openstack'] | [True] |

And also verify your vswitch_type is none.
[sysadmin@controller-0 ~(keystone_admin)]$ system show
| vswitch_type | none |

You need to enable the openvswitch label(require lock/unlock host) and then apply stx-openstack.

-Angie