Cluster and Node Operations

You can use the Buckit Manager web interface to run operations against a cluster, a set of hosts, or a single node. This page describes the available operations and how they run.

Buckit Manager runs operations over two transports:

Admin API — operations dispatched through the Buckit admin API. These require admin credentials for the cluster.
SSH — operations that run commands on the hosts over SSH, such as systemctl actions, upgrades, reboots, and provisioning. These require SSH credentials for the cluster.

After an operation finishes, it is recorded in History and the cluster view refreshes.

Note

Some operations are available only on Buckit clusters. On clusters still running MinIO Community Edition, these actions are disabled with a hint to migrate the cluster first. See Adding Clusters.

Cluster Actions

Open the Cluster Actions menu on the cluster detail page to run cluster-wide operations. The menu groups operations by transport.

Admin API

The CLI equivalent column lists the matching bm admin command. Replace ALIAS with the alias of the cluster; Buckit Manager creates an alias for each cluster it deploys or imports.

Operation	Description	CLI equivalent
Restart cluster	Sends a restart signal through the admin API. Every node re-execs in place, causing roughly three seconds of cluster unavailability.	`bm admin service restart ALIAS`
Stop cluster	Stops the Buckit service on every node through the admin API. The S3 API and console are unreachable until you start the cluster again. Data and configuration are not affected. If the cluster runs under systemd, the service may respawn immediately — use Stop cluster (systemctl) to leave it down.	`bm admin service stop ALIAS`
Freeze S3 API / Unfreeze S3 API	Freezes or resumes all S3 API calls (reads and writes). The admin API continues to work while frozen. Use only during maintenance.	`bm admin service freeze ALIAS` / `bm admin service unfreeze ALIAS`
Start heal	Triggers a heal scan through the admin API. The scan runs server-side and may take hours on a large deployment. Buckit Manager follows the event stream live; stopping the follow does not stop the scan.	`bm admin heal ALIAS`
Upgrade cluster via Admin API	Uses the Buckit admin API self-update flow. Downloads the selected release binary and restarts every node together. Use this for clusters that are not managed by systemd.	`bm admin update ALIAS`

SSH

Operation	Description
Rolling restart (systemctl)	Restarts the Buckit service one node at a time over SSH, waiting for each node to report healthy before moving to the next. No cluster downtime.
Start cluster (systemctl)	Starts a stopped cluster by running `systemctl start` on every node over SSH.
Stop cluster (systemctl)	Stops the Buckit service on every node over SSH. Unlike the admin API stop, the systemd unit ends up inactive and is not respawned, so the cluster stays down until you start it again.
Upgrade Buckit systemd service	Stages the selected version on every node without restarting, then restarts the cluster once through the admin API after all nodes are updated.
Rotate root password	Rewrites the cluster root password on every node and restarts the cluster. If a node does not already use a separate reloadable environment file, Buckit Manager first moves the current password into one and performs a rolling restart. It then writes the new password to every node, restarts the cluster through the admin API, waits for health, and rolls back on failure. Important Existing S3 clients, applications, and CLI sessions using the old credentials stop working at cutover. Update them before clients reconnect.

Manager

The gear menu next to the cluster name holds manager-level settings that do not change cluster state directly:

Configure SSH credentials — set the SSH credentials Buckit Manager uses to reach this cluster. See Connection and Application Settings.
Set admin credentials — set the admin API root credentials Buckit Manager uses for this cluster.
Remove cluster definition — stop tracking the cluster in this manager. Hosts and data are not changed; you can re-import the cluster later. This action requires typing the cluster name to confirm.

Host Actions

On the cluster detail page, select one or more hosts using the row checkboxes to enable the host action bar above the nodes table. Host actions run over SSH against every selected host:

Operation	Description
Systemctl restart service	Restarts the Buckit service on each selected host, one at a time, waiting for each node to report healthy before moving on.
Reboot host	Reboots each selected host over SSH, one at a time, then waits for the node to rejoin the cluster. Requires typing `REBOOT` to confirm.
Shut down host	Powers off each selected host. The hosts stay down until you power them on manually. Requires typing `SHUTDOWN` to confirm.

Warning

Do not reboot or shut down more than one host at a time, or the cluster may lose quorum.

Node Actions

The Actions menu on the node detail page runs operations scoped to that single node, grouped by category:

Group	Operation	Description
Service	Systemctl restart service	Restarts the Buckit service on this node and waits for it to report healthy.
Service	Systemctl stop service	Stops the Buckit service on this node. The pool continues serving through other nodes, but with reduced quorum margin.
Service	Systemctl start service	Starts the Buckit service on this node and waits for it to report healthy.
Software	Provision replacement node	Brings a clean replacement host online as a full cluster member by mirroring a healthy same-pool peer’s configuration, then installs and starts the matching Buckit version. The replacement host must meet the following requirements: A supported Linux operating system is already installed, matching the same distribution version and kernel as the other nodes in the cluster, and the host is reachable over SSH using the cluster’s SSH credentials. The host uses the same hostname as the node it replaces. The cluster references each node by hostname in its volume configuration, so the replacement must come up under that hostname to rejoin in place. The data drives are formatted as XFS and mounted at the same mount points the other nodes in the pool use. Buckit Manager verifies that each data path lands on a real mount point and fails preflight if an intended drive is not mounted — otherwise data could be written to the system disk. The host is clean: no existing Buckit or MinIO configuration files, and the data drives contain no leftover data.
Diagnostics	View service log	Opens the service log viewer for the node.
Host	Reboot host	Reboots this host and waits for it to rejoin the cluster. Requires typing `REBOOT` to confirm.
Host	Shut down host	Powers off this host. It stays down until you power it on manually. Requires typing `SHUTDOWN` to confirm.