7. Placement

Goal: Add a second worker and see how labels and selectors create two hello allocations.

Prerequisites: Chapter 6 — Deploy and changes.

Step 1 — Add a second worker

Edit workspace/workers.json:

[
  {
    "host": "10.0.0.1",
    "labels": ["worker"],
    "memory": "4096 mb",
    "cpu": "2000 mhz"
  },
  {
    "host": "10.0.0.2",
    "labels": ["worker"],
    "memory": "4096 mb",
    "cpu": "2000 mhz"
  }
]

Test SSH and trust the new host:

ssh -i secrets/worker.key agent@10.0.0.2 echo ok
maand worker trust

Step 2 — Build and inspect allocations

maand build
maand cat allocations --jobs hello

You should see two rows: hello on 10.0.0.1 and hello on 10.0.0.2. Build matched job selectors ["worker"] to each worker's labels.

hello job  →  hello @ 10.0.0.1
          →  hello @ 10.0.0.2

To run a job on only some workers, add labels (e.g. "web") to those workers and set "selectors": ["worker", "web"] in the manifest.

Step 3 — Deploy to both workers

maand deploy --jobs hello

Verify both:

maand run_command "hostname && cat /opt/maand/*/jobs/hello/data/status"

Step 4 — Drain one allocation (preview)

To stop hello on one worker without deleting the job definition, use workspace/disabled.json:

{
  "jobs": {
    "hello": { "allocations": ["10.0.0.2"] }
  }
}

Then maand build and maand deploy. The disabled allocation is stopped and not restarted.

Full walkthrough: Guide: disable and drain.

What you learned

Reference: resources and placement.

Next

Chapter 8 — Go further: pick Guides and Reference for your next task.