Cluster Commands

Command Shape

robust-ctl cluster [--server <addr>] [--output table|json] <subcommand>

Subcommands

1) status

Check cluster status.

robust-ctl cluster status
robust-ctl cluster --output json status

2) healthy

Check cluster healthy result.

robust-ctl cluster healthy
robust-ctl cluster --output json healthy

3) config

get

robust-ctl cluster config get

set

robust-ctl cluster config set \
  --config-type FlappingDetect \
  --config '{"enable":true}'

config is passed through to server-side config API as-is.

4) tenant

Manage cluster tenants (multi-tenancy support).

list

List all tenants.

robust-ctl cluster tenant list
robust-ctl cluster --output json tenant list
robust-ctl cluster --server 192.168.10.15:58080 tenant list

Sample output (table mode):

+-------------+-------------------+-------------+
| tenant_name | desc              | create_time |
+-------------+-------------------+-------------+
| business-a  | Business A tenant | 1738800000  |
| staging     | Staging env       | 1738900000  |
+-------------+-------------------+-------------+

create

Create a new tenant.

robust-ctl cluster tenant create -n <TENANT_NAME> [-d <DESC>]

Flag	Short	Required	Description
--tenant-name	-n	Yes	Tenant name (1–128 chars)
--desc	-d	No	Description (max 500 chars)

Examples:

# With description
robust-ctl cluster tenant create -n business-a -d "Business A tenant"

# Without description
robust-ctl cluster tenant create -n staging

delete

Delete a tenant.

robust-ctl cluster tenant delete -n <TENANT_NAME>

Flag	Short	Required	Description
--tenant-name	-n	Yes	Name of the tenant to delete

Example:

robust-ctl cluster tenant delete -n business-a

5) node leave

Permanently remove a node from the Raft cluster (scale-in / decommission). This is different from a node going temporarily offline — a temporary outage needs no command; a restarting node recovers automatically.

Prerequisites: stop the target node's process first, then run this; a still-alive node is refused by default (use -f to force). After removal, wipe the node's data directory before it can rejoin. Quorum guard: removal is refused when a group has ≤ 2 voters (at least 3 members are required to remove one).

robust-ctl cluster node leave -n <NODE_ID> [-f]

Flag	Short	Required	Description
--node-id	-n	Yes	ID of the node to remove (≥ 1)
--force	-f	No	Remove even if the node is still alive, default false

Example:

# Stop node 3's process first, then remove it
robust-ctl cluster node leave -n 3

# Force-remove (while the node is still alive)
robust-ctl cluster node leave -n 3 -f

Notes

• Tenants provide logical isolation within a single cluster. Suitable for serving multiple business units or multiple environments (dev / staging / prod) from one deployment.
• Use --output json for scripting and automation.