Hydra Worker Ops

Hydra Worker Ops

Use this skill when the user wants to:

• start workers for a domain
• scale worker count
• rotate domain worker credentials before rollout
• produce environment values for non-Docker platforms

Primary workflow

1. Ensure ADMIN_TOKEN is available.
2. Run:

ADMIN_TOKEN=<admin_token> ./scripts/start-domain-workers.sh <domain> [scale]

3. Select backend with WORKER_BACKEND:

• docker (default): docker compose deployment
• k8s: updates secret + deployment env + replicas
• bare: prints/exports env and optionally runs BARE_START_CMD
• print: only emits env values

Key env knobs

• API_BASE_URL (default http://localhost:8000)
• WORKER_BACKEND=docker|k8s|bare|print
• Kubernetes:
  • K8S_NAMESPACE
  • K8S_DEPLOYMENT
  • K8S_SECRET_PREFIX
• Bare-metal:
  • BARE_START_CMD

Output expectations

After running, report:

• generated domain credentials (DOMAIN/API_TOKEN/REDIS_PASSWORD flow)
• deployment actions taken for selected backend
• scheduler worker visibility summary from /workers/?domain=<domain>

Guardrails

• Do not reintroduce WORKER_DOMAIN or WORKER_DOMAIN_TOKEN.
• Keep examples consistent with DOMAIN, API_TOKEN, REDIS_PASSWORD.
• If rollout fails, immediately run domain diagnostics skill (hydra-domain-triage).

Source: punassuming/hydra — distributed by TomeVault.