Ship without fear: blue/green preprod ⇄ prod with one click

The scenario

ajansspor.com runs on Managed Containers (a prod container app + a managed MySQL). The team wants to test a sizable change for days, then go live in seconds with a guaranteed instant rollback. With blue/green we never edit the running prod app: we clone it to a preprod environment, test there, then move the production domain over.

• prod (blue) — the live app serving ajansspor.com.
• preprod (green) — a clone, reachable on its own ca-*.cdn.com.tr subdomain.
• Promote moves ajansspor.com from prod to preprod in seconds, no downtime.
• Rollback moves it back — and, for data changes, restores an automatic DB snapshot.

When the product is more than one app (web + api + worker)

Most real products are a stack, not a single container. Blue/green works on the whole stack as one unit: a project is the set of apps in your account, and one action clones, promotes or rolls back all of them together.

• Create preprod from all apps clones every prod app (web, api, worker, …) into a paired preprod copy under one project.
• App-to-app links are rewired automatically — if web talks to api, the preprod web talks to the preprod api. The preprod stack is self-contained and never points back at production.
• Promote (whole project) moves every app’s production domain to its preprod copy at once; Rollback (whole project) reverses them all. A change that spans several services goes live — or is undone — as a single unit.
• Only one service changed? Use the single-app Create preprod / Promote on just that app.

How videoonly would use this: its apps (front-end, API, workers) live in one account. You’d click Create preprod from all apps (state shared to reuse the live DB/Redis, or clone to test a migration on copied data), validate the whole stack on the ca-* subdomains, then Promote (whole project) to move every production hostname to the new versions in seconds — with Rollback (whole project) as the instant escape hatch.

State: what preprod shares with prod

Pick this when you create the preprod (the --state / dropdown):

• shared — preprod uses the same managed DB/Redis and object-storage buckets as prod (via secure cross-app references; no copy). Best for code/UI changes.
• clone — preprod gets its own managed MySQL with a copy of prod’s data. Safe for data/schema changes; prod data is untouched.
• isolated — preprod starts clean.

On Promote the platform takes an automatic DB snapshot first, so Rollback can revert data too. Static/media on a persistent volume can be shared (RWX) so uploads stay consistent.

Walkthrough A — in the panel

1. Open your CDN account → the Platforms tab → Managed Container Apps. The Blue/Green environments panel sits just above Your apps. (If this account still serves via Pull/Push CDN and has no apps yet, first click Enable Apps Alongside Current Delivery under Platforms.)

   

2. Pick a state and click Create preprod from all apps (whole product) — or Create preprod (single app) if only one service changed.

   

3. The preprod app(s) appear paired with prod under one project (each on its own subdomain). Deploy and test them as long as you need — production keeps serving traffic untouched.

   

4. When ready, click Promote (whole project) → prod. Every production domain switches to its preprod copy in seconds.

   

5. If anything is wrong, click Rollback (whole project) — traffic returns to the previous apps instantly (and the pre-promote DB snapshot is restored). The Rollback button appears on the panel once a promote has run.

Screenshots above are real captures of the live CDN.com.tr panel (Platforms → Managed Container Apps → Blue/Green environments).

Walkthrough B — with cdnctl

The same flow, scriptable end-to-end. Single app:

# 1) Clone prod → preprod (shared add-ons; use --state clone for isolated DB+data)
cdnctl container apps create-preprod --account <uuid> --app <prod_app_uuid> --state shared

# 2) Deploy + test the preprod app on its own subdomain
cdnctl container apps deploy  --account <uuid> --app <preprod_app_uuid>
cdnctl container apps expose  --account <uuid> --app <preprod_app_uuid>
cdnctl container apps status  --account <uuid> --app <preprod_app_uuid>

# 3) Go live: move the production domain to preprod (instant, no downtime)
cdnctl container apps promote --account <uuid> --app <preprod_app_uuid>

# 4) If needed, roll back (traffic + pre-promote DB snapshot)
cdnctl container apps rollback-promotion --account <uuid> --app <prod_app_uuid>

Whole product (multi-app project) — admins/ops run it on the server with artisan:

# Clone EVERY prod app in the account into a paired preprod (one project)
php artisan container-apps:env create-preprod --account <uuid> --all --state shared

# ... deploy & test each preprod app on its ca-* subdomain ...

# Promote the whole project (all production domains move together)
php artisan container-apps:env promote  --account <uuid> --project <project_uuid>

# Roll the whole project back (traffic + DB snapshots for every app)
php artisan container-apps:env rollback --account <uuid> --project <project_uuid>

The same group actions are available in the panel (the “Blue/Green environments” panel: Create preprod from all apps, Promote / Rollback (whole project)) and over the customer API (POST …/platform/container/project/{create-preprod,promote,rollback}). Low-level building blocks: container-apps:switch-origin (move one domain) and container-apps:db clone|snapshot|restore (managed-MySQL data ops).

Why it’s safe

• Production is never edited in place — you build and test on a separate app.
• Instant + reversible — Promote/Rollback is a domain-ownership switch (seconds), not a rebuild.
• Data is reversible too — Promote auto-snapshots the DB; Rollback restores it.
• Per account / per domain — DNS and TLS are unaffected; the production certificate stays.

See also the Blue/green environments guide.