ci: remove docker vfs workaround , fix yum task (#11)

[rptaylor]

continue #11

[ltalirz]

cheers, I see you figured out how to change the image

The error is now

  TASK [cvmfs_contrib.cvmfs_client : CVMFS check setup] **************************
  fatal: [instance]: FAILED! => {"changed": false, "cmd": ["/usr/bin/cvmfs_config", "chksetup"], "delta": "0:00:04.997400", "end": "2023-06-14 22:16:39.123837", "msg": "non-zero return code", "rc": 2, "start": "2023-06-14 22:16:34.126437", "stderr": "", "stderr_lines": [], "stdout": "Warning: autofs service is not running\nError: no public keys in /etc/cvmfs/keys\nError: required parameter CVMFS_SERVER_URL undefined for soft.computecanada.ca", "stdout_lines": ["Warning: autofs service is not running", "Error: no public keys in /etc/cvmfs/keys", "Error: required parameter CVMFS_SERVER_URL undefined for soft.computecanada.ca"]}`

[rptaylor]

@ltalirz yes but I'm still very confused about what is actually running. There is distro: [centos8] in ci.yml (EOL next year, should be Alma or Rocky) but I didn't change that. I changed the geerlingguy image in molecule.yml to Rocky 8 but that doesn't seem to be doing anything. And the runner OS shown in https://github.com/cvmfs-contrib/ansible-cvmfs-client/actions/runs/5272700911/jobs/9535275302?pr=12 is still Ubuntu. What is the OS of the actual container that the CI is running in?

[rptaylor]

@ltalirz I added some debug info in https://github.com/cvmfs-contrib/ansible-cvmfs-client/actions/runs/5272761999/jobs/9535402541?pr=12 to definitely confirm the right packages are installed, and the config also looks correct.
I think the only conclusion can be that autofs/FUSE/CVMFS are not working in the docker container so the CI will not work this way. There are various special considerations required to run CVMFS in a container but I am not very familiar with that. Was the VFS thing a workaround for FUSE/autofs in a container?

[ltalirz]

@ltalirz yes but I'm still very confused about what is actually running. There is distro: [centos8] in ci.yml (EOL next year, should be Alma or Rocky) but I didn't change that. I changed the geerlingguy image in molecule.yml to Rocky 8 but that doesn't seem to be doing anything. And the runner OS shown in https://github.com/cvmfs-contrib/ansible-cvmfs-client/actions/runs/5272700911/jobs/9535275302?pr=12 is still Ubuntu. What is the OS of the actual container that the CI is running in?

It is definitely running in rocky linux now

2023-06-14T22:28:01.0666298Z PLAY [Create] ******************************************************************
2023-06-14T22:28:01.0669599Z 
2023-06-14T22:28:01.0673462Z TASK [Set async_dir for HOME env] **********************************************
2023-06-14T22:28:01.1922065Z �[32mok: [localhost]�[0m
2023-06-14T22:28:01.1922261Z 
2023-06-14T22:28:01.1922424Z TASK [Log into a Docker registry] **********************************************
2023-06-14T22:28:01.2730494Z �[36mskipping: [localhost] => (item=None) �[0m
2023-06-14T22:28:01.2734822Z �[36mskipping: [localhost]�[0m
2023-06-14T22:28:01.2738826Z 
2023-06-14T22:28:01.2743078Z TASK [Check presence of custom Dockerfiles] ************************************
2023-06-14T22:28:01.7627337Z �[32mok: [localhost] => (item={'command': '', 'image': 'geerlingguy/docker-rockylinux8-ansible:latest', 'name': 'instance', 'pre_build_image': True, 'privileged': True, 'volumes': ['/sys/fs/cgroup:/sys/fs/cgroup:ro']})�[0m
2023-06-14T22:28:01.7627754Z 
2023-06-14T22:28:01.7627917Z TASK [Create Dockerfiles from image names] *************************************
2023-06-14T22:28:01.9171902Z �[36mskipping: [localhost] => (item={'command': '', 'image': 'geerlingguy/docker-rockylinux8-ansible:latest', 'name': 'instance', 'pre_build_image': True, 'privileged': True, 'volumes': ['/sys/fs/cgroup:/sys/fs/cgroup:ro']})�[0m
2023-06-14T22:28:01.9178018Z �[36mskipping: [localhost]�[0m
2023-06-14T22:28:01.9183261Z 
2023-06-14T22:28:01.9188332Z TASK [Synchronization the context] *********************************************
2023-06-14T22:28:02.0041267Z �[36mskipping: [localhost] => (item={'command': '', 'image': 'geerlingguy/docker-rockylinux8-ansible:latest', 'name': 'instance', 'pre_build_image': True, 'privileged': True, 'volumes': ['/sys/fs/cgroup:/sys/fs/cgroup:ro']})�[0m
2023-06-14T22:28:02.0045231Z �[36mskipping: [localhost]�[0m
2023-06-14T22:28:02.0057836Z 
2023-06-14T22:28:02.0061669Z TASK [Discover local Docker images] ********************************************
2023-06-14T22:28:02.8187910Z �[32mok: [localhost] => (item={'changed': False, 'skipped': True, 'skip_reason': 'Conditional result was False', 'item': {'command': '', 'image': 'geerlingguy/docker-rockylinux8-ansible:latest', 'name': 'instance', 'pre_build_image': True, 'privileged': True, 'volumes': ['/sys/fs/cgroup:/sys/fs/cgroup:ro']}, 'ansible_loop_var': 'item', 'i': 0, 'ansible_index_var': 'i'})�[0m
2023-06-14T22:28:02.8189991Z 
2023-06-14T22:28:02.8193692Z TASK [Build an Ansible compatible image (new)] *********************************
2023-06-14T22:28:02.9346630Z �[36mskipping: [localhost] => (item=molecule_local/geerlingguy/docker-rockylinux8-ansible:latest)�[0m
2023-06-14T22:28:02.9353143Z �[36mskipping: [localhost]�[0m
2023-06-14T22:28:02.9371505Z 
2023-06-14T22:28:02.9377139Z TASK [Create docker network(s)] ************************************************
2023-06-14T22:28:02.9707436Z �[36mskipping: [localhost]�[0m
2023-06-14T22:28:02.9718162Z 
2023-06-14T22:28:02.9722414Z TASK [Determine the CMD directives] ********************************************
2023-06-14T22:28:03.1071191Z �[32mok: [localhost] => (item={'command': '', 'image': 'geerlingguy/docker-rockylinux8-ansible:latest', 'name': 'instance', 'pre_build_image': True, 'privileged': True, 'volumes': ['/sys/fs/cgroup:/sys/fs/cgroup:ro']})�[0m
2023-06-14T22:28:03.1076127Z 
2023-06-14T22:28:03.1082878Z TASK [Create molecule instance(s)] *********************************************
2023-06-14T22:28:04.1946904Z �[33mchanged: [localhost] => (item=instance)�[0m
2023-06-14T22:28:04.1952754Z 
2023-06-14T22:28:04.1959287Z TASK [Wait for instance(s) creation to complete] *******************************
2023-06-14T22:28:25.6683437Z �[1;30mFAILED - RETRYING: [localhost]: Wait for instance(s) creation to complete (300 retries left).�[0m
2023-06-14T22:28:25.6691176Z �[1;30mFAILED - RETRYING: [localhost]: Wait for instance(s) creation to complete (299 retries left).�[0m
2023-06-14T22:28:25.6697275Z �[1;30mFAILED - RETRYING: [localhost]: Wait for instance(s) creation to complete (298 retries left).�[0m
2023-06-14T22:28:25.6703172Z �[1;30mFAILED - RETRYING: [localhost]: Wait for instance(s) creation to complete (297 retries left).�[0m
2023-06-14T22:28:25.6710285Z �[33mchanged: [localhost] => (item={'failed': 0, 'started': 1, 'finished': 0, 'ansible_job_id': '418950918395.135', 'results_file': '/github/home/.ansible_async/418950918395.135', 'changed': True, 'item': {'command': '', 'image': 'geerlingguy/docker-rockylinux8-ansible:latest', 'name': 'instance', 'pre_build_image': True, 'privileged': True, 'volumes': ['/sys/fs/cgroup:/sys/fs/cgroup:ro']}, 'ansible_loop_var': 'item'})�[0m

[ltalirz]

@ltalirz I added some debug info in https://github.com/cvmfs-contrib/ansible-cvmfs-client/actions/runs/5272761999/jobs/9535402541?pr=12 to definitely confirm the right packages are installed, and the config also looks correct. I think the only conclusion can be that autofs/FUSE/CVMFS are not working in the docker container so the CI will not work this way. There are various special considerations required to run CVMFS in a container but I am not very familiar with that. Was the VFS thing a workaround for FUSE/autofs in a container?

I'll need to look back into this some time, I don't have time now unfortunately.

Besides the autofs service there are some other errors - these are not relevant?

• Error: no public keys in /etc/cvmfs/keys
• Error: required parameter CVMFS_SERVER_URL undefined for soft.computecanada.ca

[rptaylor]

@ltalirz sorry for the mess, I closed the other PR so I could work on this branch but I don't think CVMFS is functional at all in the container.

Yes I believe it's a red herring ; I think any repository that we try to access would not be mountable because autofs/FUSE is not working; if you try one that has keys/URLs provided directly in the config package (e.g. atlas.cern.ch or the config repo itself cvmfs-config.cern.ch) as opposed to getting the keys/URLs via the config repo there will probably be a different error that more directly indicates the real problem.