MongoDB Atlas

Project health, real historical metrics per process, the Performance Advisor’s index suggestions and slow queries, alerts and events.

MongoDB Atlas in Kubo
A health hero built from open alerts and cluster state, with a tile per cluster.

An Atlas connection monitors one project through the MongoDB Atlas Administration API v2. One programmatic key scoped to a project reaches many clusters and processes, so the project is the connection: project id, credentials, and optional base-URL or CA overrides. It opens with six sections — Overview · Collections · Metrics · Advisor · Alerts · Events — with clusters reached from the Overview rather than a tab of their own.

Two auth modes, because Atlas has two

  • a programmatic API key, authenticated with self-contained HTTP Digest. Atlas API keys use Digest, not Basic or Bearer — the hash implementation is unit-tested against the RFC 2617 example vector;
  • a service account, authenticated with OAuth2 client credentials — client id and secret exchanged for a short-lived bearer token, cached, refreshed before expiry and re-minted on a 401.

Either way the credentials are stored as one reusable credential in the credentials manager, and every request carries the versioned Accept: application/vnd.atlas.<date>+json media type Atlas requires.

Polling

The Administration API has no watch endpoint, so the lists, the overview and the measurements poll while live updates are on, plus pull-to-refresh.

In detail