Cluster Panel

The Cluster panel shows the live topology and health of an ArcadeDB High Availability cluster, the current leader, and exposes membership-management actions on the leader.

Cluster Panel

The header shows:

  • Cluster Name.

  • A Role badge for the local node (Leader or Follower).

  • A Health badge (color-coded).

  • Refresh button.

  • Auto-Refresh dropdown (Off, 5s, 10s, 30s, 1m).

The panel has three sub-tabs.

Overview

Top info row: Leader name, Leader HTTP address, Local Peer identifier, Server count.

Below it, a grid of Node Cards — one per node — each showing status, uptime, hosted databases and key metrics. Each card also lists what that node’s build supports, or, when the leader has not been able to ask it, the reason why nothing is known about it.

A Cluster Readiness banner appears above the cards when a group or API-token change would be refused right now. Those changes are replicated to every node, and a node that cannot understand the change stops rather than skip it, so the leader refuses the whole change until the last node is upgraded and reachable. The banner names which nodes are holding the cluster back and why, so a rolling upgrade that is not finished is visible before you attempt an administrative change rather than after. It is hidden entirely on a single server, on a fully upgraded cluster, and on a follower - only the leader polls the other nodes, so only the leader can answer for them. Available since 26.10.1.

Cluster Overview

Metrics

Summary cards:

  • Elections — number of leader elections since startup.

  • Last Election — timestamp of the most recent election.

  • Uptime — cluster uptime.

Cluster Metrics

Management

Membership and lifecycle actions.

The left column exposes:

  • Cluster Membership — list the current peers and add or remove them. This action is only available on the leader; on a follower the controls are read-only.

  • Bootstrap Baselines — when applicable, shows the per-database fingerprints used to validate new peers.

The right column contains a Danger Zone with the Shutdown This Server button. Shutting a server down is irreversible from the UI; in HA the rest of the cluster keeps running.

Cluster Management