Fleet console

Health of every cluster, the live cluster screens, one inbox, search and occupancy History across sites. Sign in with the admin token or SSO.

Open https://fleet.example.com/ (localhost: http://127.0.0.1:8090/). Bootstrap: paste FLEET_ADMIN_TOKEN. SSO: Fleet identity.

This console does not keep a second copy of leases. What you see on a cluster is that cluster, live, through the tunnel.

PagePath
Overview/ — license, per-cluster tiles, hottest subnets
Clusters/clusters — register, tags, enroll tokens
Cluster workspace/clusters/{id} — live cluster console (Addresses, leases, Work, DNS, History, tenants, tokens, NetBox, Help) plus Fleet webhook and registry settings
Inbox/inbox — alerts; title opens the cluster at the lease, subnet, or Work item. Ack or snooze
Search/search — MAC, IP, hostname, subnet, reservation across clusters; hits open the workspace
History/history — live occupancy fan-out (IP, MAC, hostname, client-id, DUID, switch port). Optional cluster_id, as-of / from–to, CSV
Settings/settings — OIDC (fleet_admin)
Audit/audit — who opened what; CSV export
License/license — upload fleet.entitlement.json

Overview

Fleet Overview: six connected clusters, fleet_org license, hottest subnets

Fleet Overview: six connected clusters, fleet_org license, hottest subnets

Each tile: name, slug, mode, agent connected or API reachable, versions, license %, leases and subnets from the last probe. Click a tile to open that cluster. Hottest subnets lists utilization across the fleet; a row opens the live address map on that cluster.

DHCP is still granted on the site. Overview is health, not a lease store.

Live cluster console

Chicago HQ in Fleet: live Addresses tree, cluster picker, Probe now and Break-glass

Chicago HQ in Fleet: live Addresses tree, cluster picker, Probe now and Break-glass

Once a cluster is open, you get the same primary screens as logging into it: Overview, Addresses, Leases, Work, DNS, and More (cluster nodes, classes, audit, History, settings, license, tenants, tokens, NetBox, Help). The picker at the top switches site and keeps you on the same screen.

Writes go to that cluster through the agent (or API pull / SSH). Fleet roles still apply: readonly cannot change a reservation.

Inbox

Fleet Inbox: open alerts from every connected cluster, Ack and snooze on each row

Fleet Inbox: open alerts from every connected cluster, Ack and snooze on each row

Point cluster webhooks at the URL Fleet generated (POST /fleet/v1/hooks/{cluster_id}, HMAC). Lease events and health alerts land here. Click the title to open the matching object on that cluster. Ack or snooze (1 hour / 24 hours). The grant already happened on the site.

Fleet search for MAC 02:00:00:14 hitting reservations and leases across six clusters

Fleet search for MAC 02:00:00:14 hitting reservations and leases across six clusters

Type a MAC, IP, hostname, subnet, or reservation. Hits include the cluster name. Open the hit and you are in that cluster’s workspace.

Agents periodically push search summaries (IP, MAC, hostname, and other lookup fields — not full lease rows). Search uses that index when it is fresher than about 20 minutes; otherwise it asks the clusters live. This is not a replacement for the search bar on a single cluster.

History

Fleet-wide History (/history) fans out GET /api/v1/history live to allowed clusters. Occupancy is not copied into Fleet — as-of needs that cluster’s lease table. Optional cluster_id narrows to one site. CSV includes the cluster name and matches the lookup that succeeded. Same identity rules as cluster History: exactly one of IP, MAC, hostname, client-id, DUID, or option-82 port. History.

Audit

Fleet logs who proxied which path, enrollments, and break-glass. Export CSV. It does not store full lease tables.