OpenSearch:Configure applications

Instance basic information

1. Instance ID: A unique identifier for the instance.

2. Tag: A custom label used to identify and group instances.

3. Instance Creation Time: The time when the instance was created, in YYYY-mm-dd HH:MM:SS format.

4. Billing Method: The billing method for OpenSearch, which can be subscription or pay-as-you-go.

5. Region: The region where the application is deployed, such as China (Hangzhou), China (Shanghai), China (Beijing), China (Qingdao), or China (Zhangjiakou).

6. Edition: The edition of the OpenSearch instance, such as the High-Performance Search Edition.

7. Cluster Preference Specification: A specification based on your use case, such as Shared General-Purpose, Shared Compute-Optimized, Dedicated General-Purpose, or Dedicated Compute-Optimized.

8. Storage Resource: The cumulative size of all documents across all tables in the application. This value excludes field names, and the capacity is calculated based on the content of fields as strings.

9. Compute Resource: An LCU (Logic Compute Unit) is the unit of measurement for search compute capability. One LCU represents 10 millicores of compute capacity in the search cluster. Note: A millicore is a unit of CPU resources, equivalent to 1/1000th of a CPU core.

10. Instance Description: A brief, customizable description of the instance.

Online and offline applications

When you first configure an application and rebuild its index, it is an online application. If you later perform an operation that triggers index rebuilding, such as clearing data or making an offline change, OpenSearch creates an offline application to ensure the stability of your production service. All changes are made in the offline application. After OpenSearch successfully builds the index for the offline application, you can test it. If the tests pass, you can promote it to be the new online application. This process prevents instance operations from affecting your production services.

Status

The application status indicates its current state. A Normal status means the application version is actively serving traffic.

Other statuses for inactive applications include:

1) Pending Configuration: The status of an application before it is configured.

2) Version Initializing

3) Data Initializing

4) Invalid: The application becomes invalid if the total storage exceeds the quota, or if a data source table or field does not exist. You can only delete and then reconfigure an invalid application.

5) Expired: For a subscription instance, the service is suspended when the billing cycle ends. The instance is retained for seven days, during which you can only renew it. All other operations are disabled.

6) Frozen: The status of an application with an overdue payment.

Application basic information

The current application version includes its billing method, edition, storage resource, compute resource, instance ID, region, and application description.

Note: For details about billing rules, see Billing overview for native OpenSearch instances.

API endpoint

Each region has a different API domain name, also known as an endpoint. When using an API or SDK to access OpenSearch, use the API endpoint provided here. Regions provide distinct internal API domain names (for ECS instances in the same region, including VPC environments) and public API domain names (for public network environments). Choose the appropriate domain name based on your deployment, and ping it to verify accessibility before use.

For the China (Hangzhou) region, the public API domain name is http://opensearch-cn-hangzhou.aliyuncs.com, and the internal API domain name is http://intranet.opensearch-cn-hangzhou.aliyuncs.com. Both support HTTPS.

Data source

OpenSearch supports the following data sources: MaxCompute (formerly ODPS), RDS (versions from 5.2 up to and including 8.0 are supported. Versions outside this range are not supported), PolarDB, and data import via API/SDK.

Application structure

1. Schema: The structure of the application's tables.

2. Index fields: Fields that are configured as an index. The field name is used as the index name.

   Example: query=id:'cloud_search'. For information about index analysis, see Text Analyzer.

3. Attribute fields: Fields configured as attributes.

   Usage: You can use these fields in filter, aggregate, sort, and distinct clauses to perform filtering, aggregation, and sorting. For example: filter=id>100000.

Default display fields

Use default display fields to specify which fields are returned in each search request.

CloudMonitor

CloudMonitor displays common metrics for the current application version, such as Storage Capacity (bytes), Storage Usage (%), Total Documents, Query QPS (queries/sec), Throttled Query QPS (queries/sec), Query Latency (ms), Compute Resource (LCU), LCU Consumption per Query, and Daily Peak of Compute Resource (LCU). For metric definitions and alert configuration, see CloudMonitor Alarms.

Index rebuilding

OpenSearch keeps a mirrored copy of uploaded data, including data synchronized from various data sources. If you change the application structure or need to import a full dataset, you must rebuild the index. Two methods are available: 1) Manual index rebuilding, typically used when you modify the application structure or import full user data. 2) Scheduled index rebuilding, typically used for daily full data imports from data sources like MaxCompute. Data from RDS is synchronized by default, so you do not need to configure a scheduled task.

Switch to subscription

Change the billing method from pay-as-you-go to subscription. For details about billing methods, pricing, and procedures, see Modify Instance Specifications.

Error log

The error log displays messages for failed index rebuilding operations and failed data pushes via API/SDK. API/SDK pushes are processed asynchronously, so a successful response from your code does not guarantee a successful push. If data processing fails, an error is reported in the console. For error code descriptions, see Error Code Reference.

Decommissioned application

An online application is the version currently serving your search traffic.

When you need to change an application's configuration, such as its application structure, OpenSearch creates a copy of the online application called the offline application. The offline application starts with the same settings as the online version. You can perform offline changes (for the data source, application structure, and index structure) on an online application, and also modify its query configuration, sort configuration, and other extended features. However, you cannot create another version from an offline application; you can only modify the offline application itself. The online and offline applications have an inherited relationship but are independent entities.

Decommissioned application: When you publish the offline application, it becomes the new online application, and the previous online application becomes a decommissioned application. The new online application has a Published status and cannot be published again unless you initiate another offline change, which creates a new offline application.

Deleting an application

To prevent accidental deletion, you must confirm the action by entering the name of the online application.

OpenSearch prevents you from deleting an application while an online version exists. To decommission the application, follow these steps:

1. Delete the offline application: In the Application Management console, go to the Application Details page, select the offline application, and click Delete Application.

Note: You can skip this step if no offline application exists.

2. Delete the online application: In the Application Management console, go to the Application Details page, select the online application, and click Delete Application.

3. After the application is in the Pending Configuration status, you can delete it.

Note: You can only delete pay-as-you-go instances. You cannot delete subscription instances.

Offline change

When you need to modify the application structure (add or delete fields, change or add an index, or add or delete an attribute) or change the data source, you must perform an offline change.