Create Empty Cluster

Empty Cluster

Since PTK version 1.5.0, the concept of empty clusters has been introduced.

In some cases, you may want to use PTK to take over a specific node that is not installed by PTK, and providing the concept of empty clusters gives you more flexibility in how you use PTK to manage it.

An empty cluster is a cluster whose metadata only has the cluster name in the cluster managed by PTK. In this empty cluster, there is no instance information, you need to add it by yourself.

Create An Empty Cluster

To create an empty cluster using PTK, execute the following command:

ptk init-cluster [--name|-n CLUSTER_NAME]

• --name|-n: optional parameter, if not added, PTK will automatically create a unique cluster name.
• CLUSTER_NAME: the specified cluster name.

After this command is executed successfully, there will be a log output of "cluster "xxx" init success".

Presentation of Empty Cluster

When executing ptk ls, the empty cluster is presented in the following format:

$ ptk ls
  cluster_name | id | addr | user | data_dir | db_version |     create_time     | comment  
---------------+----+------+------+----------+------------+---------------------+----------
  test         | -  | -    | -    | -        | -          | 2024-06-12T15:37:39 |          

When executing ptk cluster status -n <ClusterName>, the empty cluster is presented as:

$ ptk cluster status -n test
[   Cluster State   ]
cluster_name                   : test

[  Datanode State   ]
  cluster_name | id | ip | port | user | nodename | db_role | state | uptime | upstream  
---------------+----+----+------+------+----------+---------+-------+--------+-----------