Configuration

An app's Configuration tab is where you change it after it is created. The tab is shown to callers with apps/write; the Owner, Contributor and Apps Operator roles have it.

Sections

Section	What it holds
Runtime	Stack, version, compute plan, port, startup project path, persistent storage — see Runtime and compute
Networking & Security	The app's domains, HTTP Only and Let's Encrypt — see Custom domains and HTTPS
Scaling	Manual, Horizontal Pod Autoscaler or KEDA — see Scaling
Environment Variables	Settings and build variables — see Environment variables and secrets
Secrets	References to vault secrets — see Environment variables and secrets
Deployment Source	Repository and branch, or container image — see Deployment source
Sentinel	Security monitoring of the app by Sentinel

Saving

Each section is saved on its own with its Save button; Reset throws away what you have not saved. A section with unsaved changes marks the tab. A save changes only the fields of that section, so two people saving different sections do not undo each other.

Every saved change becomes a new revision of the app, which you can compare and roll back — see Revisions.

What a change does

Change	What happens to the running app
Runtime stack or version, startup project path, build variables (which settings are marked, and their values), branch	The image is built again, then rolled out. Container-image apps are not built.
Other environment variables, secret references, compute plan	New instances replace the old ones, on the same image
Container image	New instances replace the old ones, with the new image
Port	New instances replace the old ones, checked on the new port. The image is not rebuilt — see Port.
Persistent storage on or off	New instances replace the old ones, with or without the disk
Scaling	Instances are added or removed; the running ones stay
Domains, HTTP Only, Let's Encrypt	Routing and the certificate change; the instances stay

When instances are replaced, an app without persistent storage keeps serving: each new instance has to be ready before an old one is removed. If the new ones never become ready, the old ones keep serving and the app shows Degraded: as soon as the platform sees that the new instances crash, cannot be placed on the cluster, cannot get their disk or cannot pull their image, and after ten minutes when they run but their port never accepts connections. An app with persistent storage is stopped first and then started, because its disk can be attached to only one node at a time, so it is briefly down.

A change to a stopped app is saved but leaves it stopped; it takes effect when you start the app.