Neutron OVN Metadata agent fails on kolla redeploy/reconfigure with podman
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
kolla-ansible |
New
|
Undecided
|
Unassigned |
Bug Description
When running OVN using podman as the container engine, the metadata service will fail to bind to netns on compute nodes running the metadata service after a kolla deploy/reconfigure of neutron. This happens if the /var/run/netns directory is not empty.
How to reproduce:
1. Deploy openstack with podman as the container engine, ovn as the network, and ``neutron_
2. After the initial deploy, run deploy/reconfigure again and the metadata service on compute nodes should stop responding.
Environment
OS: Ubuntu 22.04 LTS
Kernel: 5.15.0-92-generic
Podman version: 3.4.4
Kolla Ansible version: stable/2023.2
Image source: Official Kolla images
Relevant globals vars:
```globals.yml
neutron_
enable_neutron_dvr: "yes"
enable_
enable_
enable_
enable_
enable_nova_ssh: "yes"
#######
# Neutron - networking options
#######
# Enable distributed floating ip for OVN deployments
neutron_
# Enable DHCP agent(s) to use with OVN
neutron_
```
Podman container log:
```
Mar 05 06:03:28 compute-1c podman[1526183]: 2024-03-05 06:03:28.531860924 +0000 UTC m=+0.081029044 container exec 798503ec5e3d958
4d16e1e0 (image=
240227, kolla_version=
Mar 05 06:03:28 compute-1c conmon[1526172]: ++ . /usr/local/
Mar 05 06:03:28 compute-1c conmon[1526172]: + echo 'Running command: '\''neutron-
tron_ovn_
Mar 05 06:03:28 compute-1c podman[1526118]: ++ . /usr/local/
Mar 05 06:03:28 compute-1c podman[1526118]: + echo 'Running command: '\''neutron-
tron_ovn_
Mar 05 06:03:28 compute-1c podman[1526118]: Running command: 'neutron-
tadata_agent.ini'
Mar 05 06:03:28 compute-1c podman[1526118]: + exec neutron-
t.ini
Mar 05 06:03:28 compute-1c conmon[1526172]: Running command: 'neutron-
tadata_agent.ini'
Mar 05 06:03:28 compute-1c conmon[1526172]: + exec neutron-
t.ini
Mar 05 06:03:28 compute-1c podman[1526183]: 2024-03-05 06:03:28.573008493 +0000 UTC m=+0.122176633 container exec_died 798503ec5e3d958
c38154d16e1e0 (image=
14f190ef39226fc
Mar 05 06:03:28 compute-1c podman[1526183]: unhealthy
Mar 05 06:03:28 compute-1c systemd[1]: 798503ec5e3d958
Mar 05 06:03:28 compute-1c systemd[1]: 798503ec5e3d958
```