Create and manage clusters

A cluster is an OpenSearch execution unit composed of multiple nodes (Data Node and Cluster Manager Node) and OpenSearch Dashboards that indexes, searches, and analyzes large-scale data. A cluster provides an integrated search and analytics environment that includes high-performance search processing, data durability, scalability, and security configuration. It enables reliable exploration of data generated by various applications and systems.

The following sections describe how to create and manage OpenSearch clusters in Advanced Managed Search.

Create cluster

Create a new cluster. Depending on the environment, cluster creation can take approximately 10 minutes.

info

• Even if you expand the volume after creating the cluster, the maximum IOPS remains fixed at the initially specified value.
• Set an appropriate volume size for stable performance.

1. Go to KakaoCloud console > Analytics > Advanced Managed Search.

2. In the Clusters menu, click [Create cluster].

3. Enter the information required to create the cluster and click [Create].

   Item	Category	Description
   Basic information	Name	Enter the name of the cluster to create - Duplicate cluster names cannot be used within the same project - Example: my-cluster-01
   	OpenSearch version	Select the OpenSearch version - Supported OpenSearch versions
   Network	VPC	Select the network where the cluster will be configured - Using a VPC with two or more availability zones is recommended to maintain high availability of the cluster - Select an existing user VPC or create a new one and select it
   	Subnet	Select the subnet to which data nodes will be assigned - Select an existing user subnet or create a new one and select it
   	Security group	Configure the security group to apply to the cluster - Select an existing user security group or create a new one and select it - For more information, see Manage security groups
   Data node	Instance type	Select the instance type - Each data node in the cluster runs with the specifications of the selected instance - After the cluster is created, you can use Change instance type to change only to a higher specification
   	Availability zone	Check the availability zones for the selected subnet
   	Number of data nodes per availability zone	Enter the number of data nodes to distribute in each availability zone - Data node quota: up to 9 per project - For more information about the data node quota, see Quota * If you specify 2 availability zones and 3 data nodes, 3 data nodes are placed in each availability zone: Total number of data nodes: 6
   	Volume per data node	Specify the volume size to assign to each data node - Volume type: only SSD is currently available - Volume size: from 50 GiB to 1,600 GiB * If you specify 3 data nodes and a volume size of 50 GiB, 50 GiB is assigned to each data node: Total volume size: 150 GiB
   Cluster manager node	Instance type	Select the instance type - Cluster manager nodes run with the specifications of the selected instance - The instance type cannot be changed after the cluster is created
   	Volume per cluster manager node	Check the volume type and size assigned to each cluster manager node - Volume type: SSD - Volume size: fixed at 50 GiB
   Master user	Master username	Check the master username - For convenience, the master username is fixed as admin
   	Master password	Set the master password - The password must include uppercase letters and special characters, and can contain only letters, numbers, and special characters (8–16 characters)

Manage clusters

View cluster list

You can view the clusters created in the project.

1. Go to KakaoCloud console > Analytics > Advanced Managed Search.

2. In the Clusters menu, view the list of currently created clusters.

   Item	Description	Note
   Name	Name specified by the user when creating the cluster	Click the cluster name to view cluster details
   ID	Resource ID of the cluster
   Status	Current status of the cluster	For detailed status values, see Cluster lifecycle and status - For a cluster that completed rollback, a warning icon appears next to Running. Hover over the icon to view a message
   Detailed status	Current detailed status of the cluster	For more information, see Cluster status and detailed status
   Version	OpenSearch version in use
   Number of availability zones	Number of availability zones based on the network settings configured when the cluster was created
   Number of data nodes	Number of data nodes created when the cluster was created	Total number of data nodes: Number of availability zones * Number of data nodes
   Endpoint	Address for accessing the OpenSearch cluster through the API
   Created at	Date and time when the cluster was created

Cluster status and detailed status

When you expand a cluster or change its instance type, Detailed status is displayed in addition to Status. This lets you identify the current stage of the operation.

Successful instance type change (Scaling Up)

Status	Detailed status	Description
Running	Idle	Cluster operating normally
Scaling Up	Provisioning New Nodes	Provisioning new nodes
Scaling Up	Relocating Nodes	Excluding existing nodes from allocation and draining their shards to the new nodes
Scaling Up	Switching Traffic	Switching traffic to the new nodes and synchronizing the dashboard host
Scaling Up	Deleting Old Node	Deleting the existing nodes
Scaling Up	Done	After the instance type change is complete, the cluster status returns to Running and the detailed status returns to Idle

Successful cluster expansion (Scaling Out)

Status	Detailed status	Description
Running	Idle	Cluster operating normally
Scaling Out	Provisioning New Nodes	Provisioning new data nodes
Scaling Out	Add Node Traffic	Performing a cluster health check and updating the configuration based on all data nodes
Scaling Out	Done	After cluster expansion is complete, the cluster status returns to Running and the detailed status returns to Idle

Rollback after an error during Scaling Up or Scaling Out

If a Provisioning Error occurs in the OpenSearch cluster during cluster expansion or an instance type change (Scaling Out or Scaling Up), the cluster immediately transitions to Rollbacking. Its detailed status then changes through the following stages.

Status	Detailed status	Description
Running	Idle	Detailed status when the cluster is Running, before Scaling Up or Scaling Out begins
Scaling Up/Scaling Out	Provisioning New Nodes	Changing the instance type or expanding the cluster
Rollbacking	Reverting Desired Spec	Reverting the requested specification to the previous specification
Rollbacking	Excluding Surplus Nodes	Excluding new nodes that are no longer needed from allocation
Rollbacking	Deleting Surplus Nodes	Deleting the new nodes
Rollbacking	Reverting Traffic	Restoring the configuration and traffic to the existing nodes
Rollbacking	Verifying Cluster	Verifying that the existing data nodes are Running and all nodes are operating normally
Rollbacking	Done	Rollback is complete, the cluster status changes to Running, and the detailed status changes to Idle

info

After rollback is complete and the cluster returns to Running/Idle, a warning icon appears next to Running in the cluster Status field. Hover over the icon to view the following message:

Cluster expansion or instance type change failed. The cluster was restored to its state before the request and is operating normally.

You can view details about the failed operation in Recent operation history on the Details tab of the cluster details page.

caution

While the cluster is in Rollbacking, you cannot request volume expansion, cluster expansion, an instance type change, or cluster deletion. After rollback is complete, check the failure reason before submitting the request again.

View cluster details

You can view information about clusters created in the project.

1. Go to KakaoCloud console > Analytics > Advanced Managed Search.

2. In the Clusters menu, click the name of the cluster whose details you want to view.

3. View the detailed information on each tab.

Details tab

Cluster information

Item	Description
Name	Name specified by the user when creating the cluster
ID	Resource ID of the cluster
Status	Current status of the cluster - For more information, see Cluster lifecycle and status - A warning icon and message appear next to Running only when the cluster has recovered to its configuration before a failed operation
Detailed status	Current detailed status of the cluster - For more information, see Cluster status and detailed status
Version	OpenSearch version in use
Endpoint	Address for accessing the OpenSearch cluster through the API - To access the endpoint, authentication is required with the master user account set when the cluster was created or another user account with the required permissions
Master username	Master username entered when the cluster was created
Created by	User who requested cluster creation
VPC	User VPC selected when the cluster was created
Subnet	User subnet selected when the cluster was created
Created at	Date and time when the cluster was created
Recent operation history	Operation, failure stage, failure reason, rollback completion time, configuration before the change, and requested configuration - Displayed when cluster expansion or an instance type change fails and is rolled back - For more information, see Recent operation history

Data node

Item	Description
Instance type	Instance specification of the data node selected when the cluster was created
Number of availability zones	Number of availability zones based on the network settings configured when the cluster was created
Number of data nodes per availability zone	Number of data nodes placed in each availability zone and the total number of data nodes - Total number of data nodes: Number of availability zones * Number of data nodes
Volume size per data node	Volume size allocated per data node and the total volume size

Cluster manager

Item	Description
Instance type	Instance specification of the cluster manager node selected when the cluster was created
Number of cluster manager nodes per availability zone	Number of manager nodes and the availability zones where they are placed

Logs

Item	Description
Collected logs	Types of logs being collected
View logs	Go to Log Explorer in the Monitoring service to view logs

Instances tab

Data node

Item	Description
Name	Name created by KakaoCloud when the data node is created - You can check the data node name in the chart legend on the Monitoring tab
ID	Resource ID of the data node
Status	Current status of the data node - For more information, see Data node lifecycle and status - During cluster expansion or an instance type change, new nodes appear in the order Pending → Provisioning → Initializing → Running. A node that fails provisioning appears as Provisioning Error and is deleted during rollback
Instance type	Instance specification of the data node selected when the cluster was created
Availability zone	Availability zone where the data node is placed
Volume size	Volume size allocated to the data node

Cluster manager

Item	Description
Name	Name created by KakaoCloud when the cluster manager node is created - You can check the cluster manager node name in the chart legend on the Monitoring tab
ID	Resource ID of the cluster manager node
Status	Current status of the cluster manager node - For more information, see Cluster manager node lifecycle and status
Instance type	Instance specification of the cluster manager node selected when the cluster was created
Availability zone	Availability zone where the cluster manager node is placed
Volume size	Volume size allocated to the cluster manager node

Security group

Item	Description
Name	Security group name
ID	Security group ID

Detailed rules

Item	Description
Inbound rules	View the inbound rules of the associated security group
Outbound rules	View the outbound rules of the associated security group

Topology tab

Item	Description
Topology view	Displays the cluster connection structure across the availability zones of the VPC and subnet

Monitoring tab

Item	Description
Data period	Set the data period to display on monitoring graphs
Query items	Select monitoring query items
View mode	Set the monitoring graph view mode
Auto refresh interval	Set the auto refresh interval
Manual refresh	Refresh monitoring results by clicking the refresh button

Monitoring data

Item	Monitoring data	Description
Cluster	Number of Shards by State	Number of shards in the cluster by state
Cluster	Active Shards Percentage	Percentage of all shards that are active
Cluster	Indexed Documents (5m Interval)	Number of documents successfully indexed in the last 5 minutes
Cluster	Failed Indexed Documents (5m Interval)	Number of documents that failed indexing in the last 5 minutes
Cluster	Search Queries latency	Average response latency of search queries
Cluster	Search Queries per Second	Number of search queries processed per second
Cluster	Total Active Documents	Total number of documents stored in the cluster and currently searchable
Cluster	Thread Pool Queue Size	Number of tasks waiting in the OpenSearch thread pool
Cluster	Thread Pool Rejected Request Count (5m Interval)	Number of requests rejected in the last 5 minutes because the thread pool exceeded its capacity
Cluster	Circuit Breaker Trip Count per Second	Number of times per second the circuit breaker was triggered to protect memory
Data Node	OpenSearch Process CPU Usage (%)	CPU usage of the OpenSearch process on the data node
Data Node	System Memory Usage (%)	Overall system memory usage of the data node
Data Node	System Memory Usage (Bytes)	Total system memory usage of the data node
Data Node	Heap Usage (%)	Usage percentage of JVM heap memory on the data node
Data Node	Heap Usage (Bytes)	Actual JVM heap memory usage of the data node
Data Node	Young GC Count (1m Interval)	Number of Young GC occurrences on the data node in the last minute
Data Node	Old GC Count (1m Interval)	Number of Old GC occurrences on the data node in the last minute
Data Node	Disk Usage Watermark (%)	Disk watermark threshold set in the cluster and current disk usage of the data node as a percentage
Data Node	Disk Usage Watermark (Bytes)	Disk watermark threshold set in the cluster and current disk usage of the data node in bytes
Cluster Manager Node	OpenSearch Process CPU Usage (%)	CPU usage of the OpenSearch process on the cluster manager node
Cluster Manager Node	System Memory Usage (%)	System memory usage of the cluster manager node
Cluster Manager Node	System Memory Usage (Bytes)	Total system memory usage of the cluster manager node
Cluster Manager Node	Heap Usage (%)	Usage percentage of JVM heap memory on the cluster manager node
Cluster Manager Node	Heap Usage (Bytes)	Actual JVM heap memory usage of the cluster manager node
Cluster Manager Node	Young GC Count (1m Interval)	Number of Young GC occurrences on the cluster manager node in the last minute
Cluster Manager Node	Old GC Count (1m Interval)	Number of Old GC occurrences on the cluster manager node in the last minute

Recent operation history

The Recent operation history accordion on the Details tab shows the most recent cluster expansion or instance type change that failed and was rolled back.
If no history is available, the accordion is collapsed and contains no information.

Rollback completion message

Operation	Message
Cluster expansion	Cluster expansion failed and the previous configuration was restored. The cluster is currently running normally.
Instance type change	The instance type change failed and the previous configuration was restored. The cluster is currently running normally.

Operation history fields

Item	Description
Operation	Type of operation that failed - Cluster expansion or instance type change
Failure stage	Detailed status stage where the error occurred - Example: ProvisioningNewNodes
Failure reason	Error message describing the cause - Example: OpenSearchNode is in ProvisioningErr
Rollback completed at	Date and time when the configuration before the request was restored
Configuration before change	Configuration restored by rollback and currently applied - Displays the Currently applied badge
Requested configuration	Configuration requested by the user but not applied - Displays the Failed to apply badge

info

If the failure reason is OpenSearchNode is in ProvisioningErr, contact the Helpdesk and provide the cluster ID, failure stage, and failure reason.

Extend volume

If storage space becomes insufficient as data grows, you can expand the volume size allocated to data nodes to maintain stable cluster operation.
Volume expansion does not interrupt the cluster or restart nodes. You can increase storage without affecting existing data or the search service.

info

Volume expansion increases storage capacity but does not change the CPU or memory resources of the instance type. If you need better search performance, consider expanding the number of data nodes.

1. Go to KakaoCloud console > Analytics > Advanced Managed Search.

2. In the Clusters menu, select the cluster whose volume you want to expand.

3. Click [Extend volume].

4. In Extend volume, expand the volume and click [Extend].

   Item	Description
   Cluster	Name of the selected cluster
   Number of data nodes	Total number of data nodes in the selected cluster
   Current volume per data node	Current volume size per data node and total volume size
   Volume per data node	Volume size to assign to each data node after expansion

Expand cluster

To accommodate increased search traffic or indexed data, you can increase the number of data nodes by expanding the cluster.
OpenSearch automatically reallocates shards during cluster expansion, which might temporarily affect performance.
Expansion can improve search throughput and indexing performance and supports stable operation for large-scale data workloads.

1. Go to KakaoCloud console > Analytics > Advanced Managed Search.

2. In the Clusters menu, select the cluster to expand.

3. Click [Expand cluster].

4. In Expand cluster, increase the number of data nodes and click [Expand].

   Item	Description
   Cluster	Name of the selected cluster
   Number of availability zones	Number and names of availability zones for the selected cluster
   Current number of data nodes per availability zone	Current number of data nodes per availability zone and total number of data nodes
   Number of data nodes	Number of data nodes to assign per availability zone after expansion - Data node quota: up to 9 per project

info

• During expansion, the cluster status is Scaling Out. You can use the detailed status to check the current stage. For more information, see Cluster status and detailed status.
• If provisioning a new data node fails, the cluster transitions to Rollbacking and automatically restores the configuration from before the expansion. When recovery is complete, a warning icon appears next to Running. You can view the failed operation in Recent operation history on the Details tab.

Change instance type

To improve the search processing performance of data nodes, you can change the instance type to a higher specification.
The change is performed without cluster downtime. Cluster availability is maintained, but performance might be temporarily affected.

info

• You can change only to a higher-specification instance type. Downgrades are not supported.
• If the current instance type is the highest supported specification (c2a.8xlarge), no other instance type can be selected.

1. Go to KakaoCloud console > Analytics > Advanced Managed Search.

2. In the Clusters menu, select the cluster whose instance type you want to change.

3. Click [Change instance type].

4. In Change instance type, select the target instance type and click [Save].

   Item	Description
   Cluster	Name of the selected cluster
   Number of data nodes	Total number of data nodes in the selected cluster
   Instance type	Instance type currently applied to the data nodes
   Change type	Select the target instance type - Only instance types with higher specifications than the current type are available - If the current type is the highest specification (c2a.8xlarge), the selection field is disabled and a message is displayed

   Available instance types and change policy

   Current instance type	Available target instance types
   c2a.2xlarge	c2a.4xlarge, c2a.8xlarge
   c2a.4xlarge	c2a.8xlarge
   c2a.8xlarge	Not available (highest supported specification)

info

• During the change, the cluster status is Scaling Up. You can use the detailed status to check the current stage. For more information, see Cluster status and detailed status.
• Immediately after the instance type change is complete, the OpenSearch Dashboards health check might be interrupted temporarily while data nodes are replaced, causing the cluster status to briefly appear as Error. The cluster returns to Running without additional action. If Error persists, contact the Helpdesk and provide the cluster ID.
• If provisioning data nodes with the new specification fails, the cluster transitions to Rollbacking and automatically restores the previous instance type. When recovery is complete, a warning icon appears next to Running. You can view the failed operation in Recent operation history on the Details tab.
• During Relocating Nodes, shards are moved from the existing nodes to the new nodes. This can take a long time depending on the amount of data. If progress appears to have stopped, see Troubleshooting.

Delete cluster

You can delete a cluster that is no longer in use.

caution

Deleting a cluster also deletes its nodes and allocated volumes. Deleted resources cannot be recovered.

1. Go to KakaoCloud console > Analytics > Advanced Managed Search.

2. In the Clusters menu, select the cluster to delete.

3. In the pop-up window, enter Permanently delete and click [Delete].