Step 2 — Make it public, and read the queue

About 15 minutes. Branch: step-2.

The one new idea: every action is a row in a queue, and the app's status is not the action's status.

Make it public

Add one line to the web service:

appmecca.yaml @ step-2

    visibility: public

Commit, push, and deploy:

git commit -am "Make the board public"
git push
mecca app redeploy tally -b tutorial

Open the URL in a private window. No login page.

Private is the default, but it is the environment's default

Most environments are set up so a new app is born private, which is the right default for something half-built. Some are not. mecca app show tells you what yours actually is rather than what you expected.

There is also a live switch that needs no deploy at all:

mecca app protect tally -b tutorial   # private
mecca app publish tally -b tutorial   # public

The ALB rule is rewritten in about fifteen seconds, but allow up to ninety for it to converge everywhere: each gateway replica keeps its own sixty-second cache and they expire independently, so for a short window some replicas still serve the old answer. Use this when you need the visibility changed now; use the YAML when you want the repository to be the record.

The ninety seconds matters most going private

Until every replica has refreshed, the app is still reachable to anonymous callers. If you are making something private because it should not have been public, that window is real — wait it out before assuming you are covered.

Note the -b tutorial. Visibility is stored per branch, so flipping one branch leaves the others exactly as they were. On an app with more than one branch both commands refuse rather than guess, and list the branches so you can pick.

Now read the queue

This is the real content of the step.

mecca queue list

Every deploy, build, restart and delete is a row. You asked for something; the platform queued it; an agent claimed it and worked through it. Nothing you type happens synchronously.

mecca queue status <action-id>
mecca queue diagnose <action-id>

status gives you the state. diagnose is the one to remember: it explains in plain sentences why an action has not finished — waiting for a build, waiting for capacity, waiting behind another action on the same app.

The distinction that catches everyone: an app can be running while a deploy action is failed. That is not a contradiction — it is the platform telling you the new version did not make it and the old one is still serving. Which is the behaviour you want, and the reason the next step can break the build without taking your site down.

What you should see

• The board, in a private window, with no login.
• mecca queue list showing at least one completed deploy.
• mecca app show reporting visibility: public.

If it didn't work

Symptom	Try	Usually
Still shows the login page	mecca app show tally -b tutorial	The deploy has not finished, or the change is on a different branch
Deploy sits queued	mecca queue diagnose <id>	Another action on the same app is ahead of it
Changed the YAML, nothing happened	git status	Committed but not pushed — the platform builds from the remote

Reset to the reference

git fetch upstream && git reset --hard upstream/step-2 && git push --force origin tutorial

---

Next: Step 3 — Read the logs before you need them