Running Anchor

See every Anchor service, manage hosts, upgrades and backups, share work across Engines and open maintenance access, from the console or the command line.

Health you can read

System → Health shows every Anchor service and its instances, as cards or as a live map of the connections between them. Each service carries its recent logs, read-only diagnostics, and a redacted support bundle that never contains secrets.

The Health map showing Web, Connect, Core, Engine, Compass, Vault and Evidence and the protected links between them — The Health map: each line is a reported connection, labelled with its protocol and port.

Hosts, upgrades and backups

In the console, Control Core is where the deployment is run. Add a host with a one-time join code, remove one with a preview that refuses anything that would break the deployment, and roll out a new release from the same page. Upgrades take one signed release, checked on Core and again on each host, and move host by host in a safe order after a preview. A host is never moved backwards, and a run that stops can pick up where it left off. Plan an upgrade for a maintenance window: services restart as each host moves.

Control Core, Hosts tab listing acme-anchor01 with every Anchor role on it — Control Core → Hosts: AcmeCorp runs every Anchor role on one host. Larger deployments place each role on its own.

Backups of the credential and evidence databases count only once Anchor has restored them into a throwaway instance and checked the result. Restore drills rehearse recovery the same way, without touching the live database, and each database can stream every change to a standby copy.

The Back up a database panel, choosing between the Vault and Evidence databases — Backing up a database: written on the database's own host, and proven by restoring it into a throwaway instance.

Upgrades and backups don’t have to go through the console. The anchor command line on an Anchor host runs the same operations through the same API, with the same preview and checks:

anchor upgrade preview --release <sha256>
anchor upgrade start --release <sha256> --reason "<why>" --wait
anchor database backup --role vault --reason "<why>" --wait

Engine pools

Credential work runs on Engines. Engines work as a pool: add another and it takes its share of the verify, rotate and reconcile queue. Each Scope has its own Engine operations switch, and one estate-wide control pauses every credential change when you need everything to sit still.

System, Engine showing one connected Engine, the schedule state and each Policy's lane of work — System → Engine: connected Engines, the schedule, and each Policy's lane of work with what is due next.

Maintenance access

When an Anchor host needs hands-on attention, Open maintenance access opens that one host, through its maintenance account, for one to 24 hours. It closes at the deadline.

Tour complete · NextHow It WorksThe components behind the console, and how they connect.