Manage cluster
A cluster is a collection of nodes provisioned using Virtual Machines. The following describes how to manage a cluster in the Hadoop Eco service.
View cluster list
You can view the list of currently created clusters and their basic information.
-
Go to the KakaoCloud Console > Analytics > Hadoop Eco menu.
-
In the Cluster menu, view the list of clusters.
Item Description Cluster filter Filter to search specific clusters or search by keyword.
- Selected items work with an AND condition, and general keyword searches work with an OR condition.Name The name of the cluster specified by the user. Status The status of the cluster.
- If the cluster status is Terminated (including clusters deleted due to errors), it will be displayed for 90 days.
- Detailed explanations of each status value are available in Cluster lifecycle.Type Core Hadoop, HBase, Trino, Dataflow Node count The total number of instances used in the cluster configuration. Open api key status Status of the Open API key Creation date The date the cluster was created Uptime The time the cluster has been operational More actions - Clone cluster: Clone the cluster regardless of its status.
- Open api key issuance: Issue an API Key (visible when Open API is applicable).
- Adjust worker node count: Increase or decrease the number of worker nodes.
- Delete cluster: Delete the cluster.
View cluster details
You can view detailed information about the cluster, node details, node list, action logs, and monitoring information.
-
From the Cluster menu, select the cluster for which you want to view details.
-
View the information on the cluster's detail page.
Item Description Cluster status The status of the cluster Total instance uptime The total uptime of the instances Creator The requester who created the cluster Creation date The date the cluster was created
Check cluster information
You can view cluster tasks, task scheduling, and service integration information.
Item | Description |
---|---|
Cluster information | All information related to the cluster |
Cluster configuration (optional) | HDFS replication count HDFS block size Cluster configuration settings |
Service integration (optional) | Service integration details - Monitoring agent installation status - Data Catalog integration - MySQL database name - MySQL database ID |
Task scheduling (optional) | Task scheduling details - Task type - Task completion action - Executable file - Scheduling log file storage |
Node information
You can check information about master and worker nodes.
Item | Description |
---|---|
Node type | The type of node (master or worker) |
Instance name | Click the instance name to navigate to the detailed page of the corresponding Virtual Machine instance and perform actions. |
Instance id | The ID of the node instance |
Type | The type of the node instance |
Private ip | The private IP of the node instance |
Public ip | The public IP of the node instance |
Status | The status of the node instance |
Check monitoring
You can check detailed monitoring metrics for HDFS, YARN, Namenode, Resource Manager, Nodes, HBase, and other nodes of the Hadoop Eco cluster.
Monitoring for Trino-type clusters will be supported in the future.
-
From the Cluster menu, select the cluster for which you want to check monitoring metrics.
-
Click the Monitoring tab and view the monitoring metrics for that cluster.
Item Description Data period The period for which monitoring metrics will be retrieved
- Period options: 1 hour (default), 3 hours, 12 hours, 1 day, 7 daysQuery items The monitoring metric items to query
- Default items (if monitoring agent is not installed): HDFS / YARN / Namenode / ResourceManager / HBase / Kafka
- If monitoring agent is installed: Default items + Nodes
- By default, all items are queried, and multiple items can be selected.Auto-refresh interval Set the auto-refresh interval
- Interval options: No auto-refresh (default), 10 seconds, 30 seconds, 1 minute, 5 minutesManual refresh Click to manually refresh the monitoring results Monitoring data provided
Item Monitoring data Description HDFS HDFS usage (%) The usage of HDFS Running datanodes (count) The number of running datanodes YARN YARN memory usage (Bytes) The total available and used memory YARN cpu usage (%) The total vCore usage and used vCore Running nodemanager (count) The number of running nodemanager Running applications (count) The number of running applications Namenode Heap size (Bytes) The total available and used heap memory ResourceManager Heap size (Bytes) The total available and used heap memory Nodes Node cpu usage (%) The CPU usage of node instances
- Available when monitoring agent is installedNode memory usage (%) The memory usage of node instances
- Available when monitoring agent is installedHBase Hmaster heap size (Bytes) The total available and used heap memory
- Available only when the cluster type is HBaseRunning region servers (count) The number of running region servers
- Available only when the cluster type is HBaseKafka Active brokers (count) The number of active Kafka brokers
- Available only when the cluster type is DataflowCreated topics (count) The number of created topics
- Available only when the cluster type is Dataflow
Check action logs
If you have issued an Open API Key and created a Hadoop Eco cluster, you can check the action logs.
-
From the Cluster menu, select the cluster for which you want to view action logs.
-
Click the Action logs tab and check the action log entries.
Item Description Action log filter Select filter attributes or enter a keyword to search for action logs matching the conditions Request id The request id for the cluster Status The status of the cluster Task result The result of the cluster task User The email address of the user for the cluster Uptime/instance uptime The time the cluster has been operational
- Hover over the time to see when the cluster was requestedAction log details Click the icon to view detailed information for each action log
- Worker node count: The number of worker nodes set
- Worker volume size: The size of the worker volume in the action log
- HDFS replication count: The replication count in HDFS
- HDFS block size: The block size in HDFS
- User task options: The options the user set during the task
Clone cluster
You can clone a cluster with the same configuration.
You can clone a cluster regardless of its status, but only clusters that have information retained for 90 days after creation can be cloned.
(Note: Clusters of unsupported versions cannot be cloned.)
-
From the Cluster menu, select the [More] icon for the cluster you want to clone > Clone cluster.
-
In the Clone cluster popup, check the cluster information and select whether to clone the task scheduling settings, then click [Confirm].
- Task scheduling settings are only visible if the original cluster is Core Hadoop.
Item Description Cluster configuration Cluster version/type/availability Cluster availability Provides two availability types for operational stability: Standard and High availability Master node settings Node instance type, disk volume, hostname Worker node settings Node instance type, disk volume
Delete cluster
You can delete a cluster that is no longer in use.
Once a deleted cluster's resources are terminated, they are fully reclaimed and cannot be restored. The HDFS data will also be deleted along with the cluster's resources and cannot be recovered.
- From the Cluster menu, select the [More] icon for the cluster you want to delete > Delete cluster.
- In the Delete cluster popup, confirm the cluster to delete, enter
permanent deletion
, and click [Delete].
Api key
Issue new api key
You can issue an API key for the Hadoop Eco cluster when the cluster type is Core Hadoop
and the following conditions are met.
Conditions for issuing an api key for Core Hadoop
type clusters
- Task scheduling is enabled for Hive or Spark.
- The cluster status is Terminated(User Command) or Terminated(User).
When Open API is enabled, two cluster status values are displayed:
Pending
: The state when Open API is enabled and Hadoop Eco creation requests are possible.Processing
: The state when Open API is enabled and Hadoop Eco creation and job scheduling are in progress.
-
From the Cluster menu, select the cluster for which you want to issue an API key.
You can search for clusters by filtering the Open API status or by searching for the cluster for which you want to issue the API key.Open api status Description Not applicable Open API authentication is not exposed Available Open API authentication is exposed In progress Open API authentication is unavailable -
Click the [More] icon for the cluster > Open api authentication.
-
In the Open api authentication popup, click [Issue].
-
Copy the issued API key and save it separately.
The API key cannot be confirmed again, so please copy it from the Open api authentication popup and save it separately. If lost, click [Reissue] to issue it again.
- If a cluster with the same name exists, the Open API key will not be issued.
- If you no longer need to create clusters through Open API, click [Delete] to delete the API key.
Reissue api key
You can reissue the API key for Hadoop Eco service.
When reissuing the Open API key, the old API key will no longer work, and the newly issued Open API key must be used for API calls.
- From the Cluster menu, select the cluster for which you want to reissue an API key > [More] icon > Open api authentication.
- In the Open api authentication popup, click [Reissue].
- In the Reissue Open api key popup, click [Reissue].
Delete api key
You can delete an API key. When deleting the API key, the Action logs tab of the cluster details page will also be deleted.
- From the Cluster menu, select the cluster for which you want to delete the API key > [More] icon > Open api authentication.
- In the Open api authentication popup, click [Delete].
- In the Delete Open api key popup, click [Delete].
Adjust worker node count
You can change the number of worker nodes.
You can only change the node count when the cluster status is Running
.
- From the Cluster menu, select the cluster for which you want to change the worker node count > [More] icon > Adjust worker node count.
- In the Adjust worker node count popup, check the current instance type and count, then enter a number greater or smaller than the current node count and click [Save].