12 questions to validate the level: registries, healthcheck and restart, reverse proxy, backups and CI.
Answer everything, then click Check: your score and explanations will appear.
4.1 · Recall
1. What is each registry (Hub, GHCR, private) for?
Public for open source, GHCR for your CIs, private for sensitive internal.
4.1 · Application
2. Idir logs in to the registry in a script. Which command is correct?
The stdin-passed limited token lingers neither in history nor in logs.
4.2 · Recall
3. What does restart: unless-stopped do?
Relaunch after crash or reboot, but respects manual stop: the right lasting default.
4.2 · Application
4. docker compose ps shows a service “Up (unhealthy)”. What does Idir do?
Up = living process, unhealthy = sick service: treat, don't route.
4.2 · Application
5. Idir's API crashed because the database wasn't ready. What does he declare?
The condition waits for true ready proven by the healthcheck, not a mere order.
4.3 · Application
6. Three web applications, a single port 80 available. What does Idir do?
Only one published (the proxy), routing by names: one port, N services.
4.3 · Arbitration
7. The database is currently on the frontend network with the applications. Idir's decision?
Database confined to backend: directly unreachable even if the proxy is compromised.
4.4 · Recall
8. Which backup is valid?
Only a tested restore proves the backup.
4.4 · Application
9. Idir backs up the volume without stopping production. How?
Read-only throwaway, off-volume archive: prod keeps running, the backup is elsewhere.
4.5 · Application
10. Which tag does Idir deploy to production from CI?
Prod deploys an exact, tested artefact, not a moving alias.
4.5 · Arbitration
11. Idir's workflow needs the registry token. Where does he put it?
Repository secrets, never hard-coded nor in logs, minimal permissions.
4.1 · Arbitration
12. Idir's team deploys to production an image whose exact content nobody knows. What does he decide?
Single CI factory and immutable tag: each prod image has a known, traceable origin.
What's next?
Score ≥ 80%: move to Level 5 — orchestration and production, and industrialise your gains (private registry, hardened stacks, single frontend, tested backups, CI factory).
Score 50-79%: revisit the failed sections (registries 4.1, reliability 4.2, proxy 4.3, backups 4.4, CI 4.5), reread the explanations above, then retake the test.
Score < 50%: restart the level from 4.1, practising each move (traceable push/pull, healthcheck + restart, segmented frontend, tested restore, green workflow), then retry the exam with a fresh head.