Observability
You should not have to add a tool to see what your Project is doing.
Build output, runtime logs, resource use per container, and costs broken down. None of it is an add-on you install, a sidecar you configure, or a plan you upgrade to. It is on because your Project is running.
Included. There is no observability tier.
In one place
The Project is where it adds up.
A component tells you about itself. A Project tells you what the application is doing: every component, the databases and buckets they use, the routes in front of them, and one number for what the whole thing costs this month. You do not assemble that view yourself. It is the page you were already on.
Built in
Each part reports on itself.
Nothing here is a separate product. These views are part of every Project you deploy.
Build output
The log of a deploy as it happens, streamed rather than polled, and kept with the deployment so a failed build can still be read afterwards.
Runtime logs
What the component is printing right now, filterable, without opening a shell or attaching an agent.
Resource use
CPU, memory and storage per container while it runs, on the same screen as the console and the power controls.
Cost, broken down
Month to date and a projection, split by base container, extra replicas, bandwidth and custom domains, per component and for the Project as a whole.
Activity
Who changed what, when. Deploys, variables, members, domains. The record exists whether the change came from the console, the API or an assistant.
The system map
What is connected to what, drawn from the Project as it actually is rather than from a diagram someone drew once and stopped updating.
Told, not discovered
Nobody has to watch a screen.
Observability that only works while you are looking at it is not much use. The events that matter reach you where you are.
- A deploy succeeded.
- A deploy failed.
- A database finished provisioning, or was removed.
- A component was converted to static hosting.
- Your balance is critically low.
They arrive in the console and by email, and the same events are what the spend cap acts on.
On the roadmap
Visitor analytics is coming, and is not here yet.
Page views, referrers and where your users come from are being built, and none of it exists today. We would rather write that here than let you find out after you have deployed.
The edge counts requests per app to decide when to scale, and that counter is kept for five minutes and then discarded. It is not an analytics pipeline wearing a different name.
We will not put a date on it before it is done.
Deploy something and watch it run.
The logs, the cost breakdown and the system map are already there.