docker-operator

features

• sets oom_score on containers
• performs system prune operations via Docker API
• processes events from the Docker events API and notify owners on Slack based on the experts container label (comma separated)

container labels

These labels are applied to workload containers read by docker-operator to direct and enrich messages when reporting events

labels:
  # try to have 2+ experts for each service.
  # these are used for event notifications.
  # specify the full email used for slack or specify the username and SLACK_EMAIL_DOMAIN env var.
  - "experts=dustind,jdoe"
  - "description="acme service farms chickens"

These env vars are applied to docker-operator so it can connect to the Slack API (optional) and provide enrichement (optional)

Right now LOG_URL_FORMAT and DEPLOY_URL_FORMAT only supports substituting Docker Swarm service names from the com.docker.swarm.service.name container label, but it could easily support others with minor improvements.

environment:
    - SLACK_BOT_API_KEY_PATH=/run/secrets/docker-operator-slack-bot-token
    - SLACK_WORKSPACE_API_KEY_PATH=/run/secrets/docker-operator-slack-workspace-token
    - SLACK_EMAIL_DOMAIN=indeed.com
    - LOG_URL_FORMAT=https://kibana.internal.net/app/kibana#/discover?query=%[1]s
    - DEPLOY_URL_FORMAT=https://swarm.internal.net/service/ps/%s

restart policy

Set a sensible restart policy so you don't get spammed by alerts if your application is unhealthy

restart_policy:
    condition: on-failure
    delay: 5s
    max_attempts: 3
    window: 120s # monitors the process for this period

todo

• add K8s support
• add fork bomb prevention (via parent cgroup pids.max)
• add other notification mechanisms