Skip to main content

Manage Cluster Using Eloqctl

In this document, we will illustrate how to use eloqctl to manage EloqKV cluster.

Please ensure the cluster is already provisioned. For how to deploy EloqKV cluster, please refer to Deploy Cluster.

Check Cluster Status​

Cluster name is set in the configuration file when you provision the cluster. The default cluser name is eloqkv-cluster. You can also check the cluster name using command below.

eloqctl list

The expected output will be:

+----------------+---------+---------+---------+-------+
| name           | product | store   | version | user  |
+----------------+---------+---------+---------+-------+
| eloqkv-cluster | EloqKV  | rocksdb | 0.7.4   | rocky |
+----------------+---------+---------+---------+-------+

Use the following command to query the cluster status:

eloqctl status ${cluster_name}

If you see Success in the output for every node , then the cluster runs normally.

Start Cluster​

Use the following

eloqctl start ${cluster_name}

To start specified nodes, use the following

eloqctl start --nodes node1_ip:node1_port,node2_ip:node2_port ${cluster_name}

Stop Cluster​

  1. Graceful Shutdown. The following command will stop the transaction cluster and log cluster gracefully which means that all the data in memory will be flushed to persistent storage before the processes exit. In this case, the next launch of cluster is fast since there is no redo logs to be replayed.
eloqctl stop ${cluster_name}
  1. Force Shutdown. You can also stop the cluster immediately by passing -f option. It actually send SIGKILL signal to all the transaction processes and log processes.
eloqctl stop ${cluster_name} -f
  1. Stop Persistent Storage. When you deploy a decoupled persistent storage, i.e. Cassandra, the Cassandra cluster will not be stopped automatically by eloqctl stop. You should pass -a option to tell eloqctl to stop every components except monitor.
eloqctl stop ${cluster_name} -a
  1. Stop Monitor. Prometheus and grafana will be stopped by the following command:
eloqctl monitor stop ${cluster_name}
  1. If cluster is deployed with a password, all of the above commands needs to add --password option
eloqctl stop ${cluster_name} --password xxxxx
  1. To stop specified nodes, use the following
eloqctl stop --nodes node1_ip:node1_port,node2_ip:node2_port ${cluster_name}

Update Cluster Configuration​

EloqKV offers various configurations, some of which enable features. For example, enable_data_store activates persistent data storage, and enable_wal enables the Write-Ahead Log for durability. Other configurations are performance-related, such as core_number for specifying the number of worker threads, and node_memory_limit_mb to set the memory limit.

You can easily adjust these settings using eloqctl. The process is as follows:

  1. Edit the configuration file located at $HOME/.eloqctl/upload/${cluster_name}/{node_ip}/EloqKv-tx-{port}.ini to update config for corresponding node. In the example below, core_number is set to 8, and both the persistent data store and Write-Ahead Log are enabled.
[local]
path=data
ip=127.0.0.1
port=6379
core_number=8
enable_data_store=on
enable_wal=on
[cluster]
[store]
[metrics]
enable_metrics=on
  1. Use eloqctl update-conf to update the configuration file across the cluster and restart it with a single command.
eloqctl update-conf ${cluster_name} --restart

Scale Cluster​

Cluster expansion and shrinkage using eloqctl will be available soon!