zun-compute container hang in restarting with kolla deploy

[Mario Torrisi]

Hello,

I'm deploying Openstack's services on a virtualised multinode infrastructure that support container provisioning.

Here some relevant information about deployment from globals.yml:

kolla_base_distro: "ubuntu"
kolla_install_type: "source"
openstack_release: "rocky"
...
enable_etcd: "yes"
...
enable_haproxy: "no"
...
enable_horizon_zun: "{{ enable_zun | bool }}"
...
enable_kuryr: "yes"
...
enable_haproxy: "no"
...
The problem is after kolla-ansible deploy completes well, as it seems
controller : ok=32 changed=0 unreachable=0 failed=0
compute : ok=24 changed=2 unreachable=0 failed=0
strage : ok=1 changed=0 unreachable=0 failed=0
localhost : ok=1 changed=0 unreachable=0 failed=

In the compute node zun stuck in 'Restarting' state.
Find attached log files.

[hongbin (hongbin034)]

The logs show that the zun-compute process cannot connect to docker daemon, possibly due to a configuration issue. A few things to check:

* Which version of kolla-ansible you are using (want to make sure you have this patch: https://review.openstack.org/#/c/565240/).
* Add the following to your globals.yaml and see if there is any luck:

  docker_custom_option: " -H unix:///var/run/docker.sock -H tcp://<your_ip_address>:2375"

[hongbin (hongbin034)]

BTW, the option 'docker_custom_option' is for allowing the docker daemon to listen to an IP address that your zun_compute container can access.

[hongbin (hongbin034)]

I am going to close this bug due to no activity within 90 days.