Telemetry overview
Telemetry is where your Agent’s history lives: the metrics, logs and traces it keeps for your canisters, and the views you build on them. It needs an Agent workspace, because this browser keeps no history of its own.
Where to start
Section titled “Where to start”| You want to… | Open |
|---|---|
| Chart one metric, or ask a question in PromQL | Metrics |
| Search what your canisters printed | Logs |
| Follow one request across canisters | Traces |
| See how real visitors use your web app | Frontends |
| Keep a set of charts to come back to | Dashboards |
| Share a recurring summary | Reports |
| Choose what is kept, and who sends it | Data connections |
For one canister right now, its own page is faster: its Overview and Timeline draw on the same data.
Where the data comes from
Section titled “Where the data comes from”- Monitor reads every canister in the workspace from the Internet Computer, every 5 minutes by default: cycles, memory, calls, and logs where allowed.
- Your code can send its own metrics, logs and traces. Send telemetry from your code.
- Your web app’s visitors send page views, errors and Web Vitals through a small snippet. Monitor a frontend.
History starts when collection starts. Adding a canister today doesn’t fill in last week.
Missing isn’t zero
Section titled “Missing isn’t zero”A gap in a chart means nothing was recorded, not that the value was zero. A canister the Agent can’t read is named as missing, so a short result never looks like a quiet fleet. Signals and freshness.