Skip to content

An Ansible playbook for deploying a Gitlab Runner using Podman

License

Notifications You must be signed in to change notification settings

zedr/gitlab_runner

Repository files navigation

Gitlab Runner with Podman

Ansible Galaxy

A playbook for installing and running a Gitlab Runner instance using Podman using a normal user. The container still needs to run privileged.

Supports Fedora 33, Debian 12.

Installation

From Ansible Galaxy

$ ansible-galaxy collection install zedr.gitlab_runner

From source

Build the collection:

$ ansible-galaxy collection build -vvv -f --output-path ./build

Install the collection:

$ ansible-galaxy collection install -vvv build/*.tar.gz

Usage

Example playbook:

- hosts: all
  collections:
    - zedr.gitlab_runner
  roles:
    - role: gitlab_runner

See the role documentation for additional configuration.

Molecule test dependencies

OS:

  • libvirt-devel

Running the Molecule tests

Start the test suite by creating the VM and converging using the role:

$ make converge

Verify that the test suite passes:

$ make verify

Destroy the VM:

$ make destroy

Test with a local image to speed up things

$ CLOUD_IMAGE_URL=file:///home/zedr/Archive/Fedora-Cloud-Base-33-1.2.x86_64.qcow2 \
make converge

Test by signing into and using a custom registry

$ EXTRA_VARS="custom_runner_image=quay.io/myrepo/gitlab-runner:latest \
              custom_registry_host=quay.io \
              custom_registry_username=my_quay_user \
              custom_registry_password=my_quay_pass" \
MOLECULE_SCENARIO="registry_auth" \
make converge

Vagrant test VM

Use with:

$ vagrant up

Issues

"UNREACHABLE (...) Failed to connect to the host via ssh"

Destroy the VM with make destroy and retry.

"Failed to download metadata" when creating the VM

A sporadic issue causing the error message Failed to download metadata for repo 'fedora-modular': Cannot prepare internal mirrorlist: Status code: 503 seems to be unrelated to the role and its tests. Try launching the tests again.

$ systemctl --user status gives an error

Running systemd ctl commands as an unprivileged user might yield an error Failed to connect to bus: $DBUS_SESSION_BUS_ADDRESS and $XDG_RUNTIME_DIR not defined (consider using --machine=<user>@.host --user to connect to bus of other user)

Try defining the env var XDG_RUNTIME_DIR, e.g. XDG_RUNTIME_DIR=/run/user/$UID systemctl --user status

About

An Ansible playbook for deploying a Gitlab Runner using Podman

Resources

License

Stars

Watchers

Forks

Packages

No packages published