Throw Out A Node

Purpose

In some clusters managed by PTK, it is sometimes desirable to throw out the management of a node in a cluster that cannot be uninstalled due to the large amount of data in the cluster. To address this situation, PTK 1.5.0 introduced the throwout command, which can be used to throw out a node in a given cluster.

It is important to note that the primary node cannot be thrown out if there is a non-primary node in the cluster.

PTK sets some restrictions on throwout, for example, if the current cluster has a primary node, a standby node and a cascade node, it is not allowed to throw out either the standby node or the primary node, you need to throw out the cascade node before you can throw out the standby node. Then the primary node can be throwed out.

Usage

The command to throw out a specified node using PTK is as follows:

# Specify IP
ptk cluster -n <CLUSTER_NAME> throwout -H <IP>
# Specify ID
ptk cluster -n <CLUSTER_NAME> throwout -i <ID>

The throwout command is relatively simple, but there are some restrictions to be aware of when using it:

1. When performing a throwout operation, check to see if there is a downstream database on the specified node.
2. Make sure that the specified node is in the current cluster. During the throwout process, there is an interaction that allows the user to confirm whether or not to thrown out the node. As shown in the following example:

$ ptk cluster -n <CLUSTER_NAME> throwout -i <ID>
Nodes to be throwout:
- <IP>
✔ Are you sure you want to throwout them (default=n) [y/n]: █

Of course, PTK provides the -y|--yes parameter to skip the interaction.