implemented whitelisting of anonymous docker volumes

This commit is contained in:
Kevin Veen-Birkenbach 2023-11-17 18:31:35 +01:00
parent fb588a346c
commit 83cc40b718
3 changed files with 24 additions and 11 deletions

View File

@ -2,23 +2,24 @@
## Description
This role checks for anonymous Docker volumes that are not bound to a container and may be left over from previous operations. It provides a cleanup mechanism by identifying such volumes and possibly taking action against them.
This role checks for anonymous Docker volumes that are not bound to a container and may be left over from previous operations. It provides a cleanup mechanism by identifying such volumes, excluding any that are whitelisted, and possibly taking action against them.
## Files
- `vars/main.yml`: Variable definitions for the script's directory.
- `vars/main.yml`: Variable definitions for the script's directory and whitelist.
- `handlers/main.yml`: Handlers to reload and restart the systemd service and timer.
- `files/health-docker-volumes.sh`: The script that checks for anonymous Docker volumes.
- `files/health-docker-volumes.sh`: The script that checks for anonymous Docker volumes and excludes whitelisted volumes.
- `tasks/main.yml`: Tasks to create necessary directories, copy scripts, and create systemd service and timer.
- `templates/health-docker-volumes.service.j2`: Systemd service template.
- `templates/health-docker-volumes.service.j2`: Systemd service template, including the whitelisted volumes as a parameter.
- `templates/health-docker-volumes.timer.j2`: Systemd timer template.
- `meta/main.yml`: Meta information declaring dependencies for the role.
## Usage
This role can be included in your playbook. Set the `path_administrator_scripts` variable to determine where the health check scripts should reside.
Include this role in your playbook and set the `path_administrator_scripts` variable to determine where the health check scripts should reside. Define `whitelisted_anonymous_volumes` in `vars/main.yml` with an array of volume IDs that should be ignored by the health check.
The role uses `systemd_notifier` for failure notifications, so ensure this dependency is present in your environment.
Ensure that the `systemd_notifier` dependency is satisfied for error notifications.
## Created with AI
This script was created with the help of AI. The full conversation you find [here](https://chat.openai.com/share/1fa829f1-f001-4111-b1d4-1b2e3d583da2).
This script was created with the help of AI. The full conversation you find [here](https://chat.openai.com/share/1fa829f1-f001-4111-b1d4-1b2e3d583da2).

View File

@ -1,17 +1,29 @@
#!/bin/bash
status=0
# The first argument is a space-separated list of whitelisted volume IDs
whitelist=$1
whitelisted_volumes=($whitelist) # Split into an array
anonymous_volumes=$(docker volume ls --format "{{.Name}}" | grep -E '^[a-f0-9]{64}$')
if [ -z "$anonymous_volumes" ]; then
echo "No anonymous volumes found."
exit 0
exit $status
fi
echo "Anonymous volumes found:"
for volume in $anonymous_volumes; do
container_ids=$(docker ps -aq --filter volume=$volume)
# Check if the volume is in the whitelist
if printf '%s\n' "${whitelisted_volumes[@]}" | grep -q "^$volume$"; then
echo "Volume $volume is whitelisted and will be skipped."
continue
fi
status=1
container_ids=$(docker ps -aq --filter volume=$volume)
if [ -z "$container_ids" ]; then
echo "Volume $volume is not used by any running containers."
continue
@ -29,4 +41,4 @@ for volume in $anonymous_volumes; do
done
done
exit 1
exit $status

View File

@ -4,4 +4,4 @@ OnFailure=systemd-notifier@%n.service
[Service]
Type=oneshot
ExecStart=/bin/bash {{health_docker_volumes_folder}}health-docker-volumes.sh
ExecStart=/bin/bash {{ health_docker_volumes_folder }}health-docker-volumes.sh "{{ whitelisted_anonymous_docker_volumes | join(' ') }}"