# Cluster Startup

> Learn how to start up a CubeCOS single-node or multi-node cluster, including bootstrapping behaviour and monitoring guidance.

The following procedures are used to start up a single CubeCOS node or a CubeCOS cluster.

The startup process involves observing system behaviour, including the bootstrapping sequence and cluster message flow.

## Cluster startup overview {#cluster-startup-overview}

The following startup behaviour applies to the following deployment types:

- AIO
- HCI
- Web scale

CubeCOS automatically bootstraps cluster services during cluster node startup.

The following steps outline the bootstrapping process and how to monitor its status in multi-node cluster environments.

### Perform a CubeCOS cluster startup {#perform-a-cubecos-cluster-startup}

1. Power on all nodes.

   Start all machines that are part of the CubeCOS cluster.

2. Wait for the master node to initialize.

   The master nodes initiates the bootstrapping process. All other nodes remain in a waiting state until it completes.

   
      
   

3. Monitoring the bootstrap process.

   You can monitor the bootstrap process from the console of other nodes.

   While waiting for the master to finish bootstrapping, the console displays `Waiting for master control to finish bootstrap-cube [x min]`.

   
      
   

4. Cluster initialization after master bootstrap

   :::info

   For deployments with converged control nodes (AIO, single-node), the startup process begins and completes on the master node. Afterwards, all other nodes join the cluster.

   For deployments with dedicated control nodes (HCI, web scale), the startup process begins and completes on the master node. Then, the two additional control nodes join the cluster, followed by the remaining compute or storage nodes.

   :::

   After the master node finishes bootstrapping and displays a `Starting cluster` message, all child nodes proceed to join the cluster.

   
      
   

   Other nodes will begin their bootstrapping process and join the cluster.

   
      
   

5. Cluster services check and repair after bootstrap.

   CubeCOS automatically performs service check and repair as the last step of cluster bootstrap. The service check and repair ensures all modules have properly started and no warnings or errors occurred during startup.

   Wait until services to become ready and the console displays `cluster checked and repaired`.

   
      
   

6. Verify cluster service and check results.

   After the cluster services check and repair. Verify that `services ready` is displayed on the node consoles.

   To verify service status, press `F2` to open the console, log in with the `admin` account, and run the `cluster check` command. Ensure all services return an `ok` status.

   
      
   

## Single node startup overview {#single-node-startup-overview}

CubeCOS automatically bootstraps cluster services during startup. The following steps outline the bootstrapping process and how to monitor its status in single-node deployments.

### Perform a single node startup {#perform-a-single-node-startup}

1. Power on the single node.

   Start the machine that hosts the single-node CubeCOS deployment.

2. Single node automatically initiates bootstrapping modules.

   Bootstrapping will be handled automatically, while bootstrapping, the console displays a `services n/a` message.

   
      
   

3. Service readiness and health checks.

   During bootstrapping, CubeCOS automatically checks the health of system services and attempts to repair any that are degraded or unavailable. Wait until all services are initialized. Services are ready when the console displays the `Service Ready message`.

   
      
   

4. Cluster ready.

   Press `F2` to enter the console, log in as `admin` and run `cluster check` to verify that all services are `ok`.