Help Center/ Cloud Search Service_KooSearch/ User Guide/ Enabling KooSearch Document Q&A Service
Updated on 2025-08-22 GMT+08:00
View PDF
Share

Enabling KooSearch Document Q&A Service

On KooSearch, you can create a knowledge base, upload documents to it, and then use this knowledge base for document-based Q&A. However, before you can use KooSearch Document Q&A, you need to enable the service.

KooSearch is available only in CN-Hong Kong and AP-Singapore. KooSearch is in the open beta test (OBT) phase. To trial-use it, submit a service ticket.

Accessing the KooSearch Console

  1. Log in to the CSS management console.
  2. In the navigation pane on the left, choose KooSearch > KooSearch Document Q&A.
  3. Select a document Q&A service created earlier, and click Q&A in the Operation column to switch to the KooSearch console.

Enabling KooSearch Document Q&A Service

  1. On the KooSearch page, click Enable.
  2. On the displayed page, configure the service.
    • Basic Configuration
      Table 1 Basic settings

      Parameter

      Description

      Billing Mode

      Select Yearly/Monthly or Pay-per-use.

      • Pay-per-use: You are billed by actual duration of use, with a billing cycle of one hour. For example, 58 minutes of usage will be rounded up to an hour and billed.
      • Yearly/Monthly: You pay for the service by year or month, in advance. The service duration ranges from one month to one year.

      Required Duration

      Select a duration when Billing Mode is set to Yearly/Monthly.

      If necessary, select Auto-renew to enable automatic renewal of the service.

      Specifications

      Select Agile, Basic, Professional, or Enterprise.

      Region

      Select the region of the KooSearch service from the drop-down list.

      Regions are geographic areas isolated from each other. Resources are region-specific and cannot be used across regions through internal network connections. Select a region near you to ensure the lowest latency possible.

      Service Name

      Set a custom service name.
    • Vector Cluster Configuration

      A vector database stores and retrieves vectorized data. Currently, only Elasticsearch 7.10.2 clusters are supported. Configure cluster parameters. After the document Q&A service is enabled, a cluster is automatically created based on the settings you configure here.

      Table 2 Vector cluster parameters

      Parameter

      Description

      Nodes

      Number of nodes in the vector cluster. Select a number from 1 to 32. You are advised to configure three or more nodes to ensure high availability of the cluster.

      CPU Architecture

      Select x86 or Kunpeng. The architectures actually supported depend on your region.

      AZ

      Select an AZ associated with the cluster's region.

      Flavor

      Select a node flavor for the cluster. For details, see ECS Types.

      Node Storage Type

      Select a node storage type, which can be Common I/O, High I/O, or Ultra-high I/O.

      Node Storage Capacity

      Select a node storage capacity. Its value range varies with different node specifications.

      The node storage capacity must be a multiple of 20.

      Disk Encryption

      Whether to encrypt the data disks of nodes using Key Management Service (KMS).

      Enabling disk encryption enhances the security of the data stored on cluster nodes. By default, disk encryption is disabled.

      After disk encryption is enabled, you need to configure Key Name by selecting an enabled KMS key from the drop-down list. If no key is available, click Create key to go to the Data Encryption Workshop (DEW) console and create a new key or modify an existing key. For details, see Creating a Key.

      NOTE:
      • Only cloud disks support disk encryption. Local disks do not support disk encryption.
      • Only custom keys whose Key Algorithm is AES or SM4 and Usage is ENCRYPT_DECRYPT are supported. KMS keys that are unavailable in the Key Name drop-down list are not supported by the cluster.
      • Disk encryption and decryption do not alter cluster management or O&M processes. However, they do increase the system's processing load, potentially affecting the system's operational performance.
      • Once a cluster is already created, disk encryption cannot be enabled or disabled.
      • After cluster creation, the KMS key cannot be changed.
      • If the KMS key used by the cluster is disabled, the cluster cannot be scaled or upgraded, its node specifications or AZs cannot be changed, and its nodes cannot be replaced (by specifying the nodes that need replacement). To solve this problem, you will have to create a new cluster and migrate your data to that new cluster.

      Security Mode

      Whether to enable security mode for the cluster.

      • The security mode is enabled by default. In security mode, a cluster's communication is encrypted and access to the cluster requires user authentication. This is why the Administrator Username and Administrator Password of the cluster must be configured.
        • The default administrator username is admin.
        • Set and confirm the Administrator Password. This password will be required when you access this cluster.
      • If Security Mode is disabled, a cluster in non-security mode will be created. With such a cluster, access to the cluster will not require user authentication, and data will be transmitted in plaintext using HTTP. Make sure the customer is in a secure environment, and do not expose the cluster access interface to the public network.
    • (Optional) Search Model Configuration
      • Enabled: When the KooSearch Document Q&A service is enabled, a text vectorization model and a search reranking model will be created automatically.
      • Disabled: When the KooSearch Document Q&A service is enabled, these two models will not be created, which will affect the use of knowledge bases. If necessary, configure a search embedding model, search reranking model, and cache generation model on the Model Management page. For details, see (Optional) Creating and Managing Models on KooSearch.
      Table 3 Parameters for search model configuration

      Parameter

      Description

      Text Vector and Reranking Inference Instance

      Select the number of inference instances for the text vectorization and reranking models.

      Model Type

      Select the model language.

      Instance Specifications

      Select Ascend or GP-accelerated.

      AZ

      When Instance Specifications is set to GP-accelerated, select an AZ under the current region.

      Flavor

      Select a node flavor for the search models. For details, see ECS Types.

      Node Storage Type

      When Instance Specifications is set to GP-accelerated, select a node storage type, which can be Common I/O, High I/O, or Ultra-high I/O. The node storage types available vary depending on the selected AZ and instance specifications, as well as the regional environment.

      Node Storage Capacity

      When Instance Specifications is set to GP-accelerated, set a node storage capacity. The value range varies depending on the node flavor selected.
    • (Optional) Search Planning Configuration
      • Enabled: When the KooSearch Document Q&A service is enabled, a search planning model will be created automatically. This model provides intent recognition and query rewriting.
      • Disabled: When the KooSearch Document Q&A service is enabled, this model will not be created, which will affect the use of knowledge bases. If necessary, configure a search planning model on the Model Management page. For details, see (Optional) Creating and Managing Models on KooSearch.
      Table 4 Parameters for search planning configuration

      Parameter

      Description

      Search Planning Inference Instance

      Select the number of inference instances for the search planning model.

      Model Type

      Select the model language.

      Instance Specifications

      Select Ascend or GP-accelerated.

      AZ

      When Instance Specifications is set to GP-accelerated, select an AZ under the current region.

      Flavor

      Select a node flavor for the search planning model. For details, see ECS Types.

      Node Storage Type

      When Instance Specifications is set to GP-accelerated, select a node storage type, which can be Common I/O, High I/O, or Ultra-high I/O. The node storage types available vary depending on the selected AZ and instance specifications, as well as the regional environment.

      Node Storage Capacity

      When Instance Specifications is set to GP-accelerated, set a node storage capacity. The value range varies depending on the node flavor selected.
    • (Optional) Large Model Configuration
      • Enabled: When the KooSearch Document Q&A service is enabled, a preset LLM service will be created automatically.
      • Disabled: When the KooSearch Document Q&A service is enabled, no such service will be created, which will affect the use of the KooSearch Document Q&A service. If necessary, configure the needed NLP model on the Model Management page. For details, see (Optional) Creating and Managing Models on KooSearch.
      Table 5 LLM configuration parameters

      Parameter

      Description

      Model Version

      Currently, koosearch-rag is provided. Developed by fine-tuning an open-source model, this model significantly improves search accuracy through retrieval-augmented generation.

      Generative Model Inference Instance

      Select the number of inference instances for the generative model.

      Model Type

      Select the model language.

      Instance Specifications

      Select Ascend or GP-accelerated.

      AZ

      When Instance Specifications is set to GP-accelerated, select an AZ under the current region.

      Flavor

      Select a node flavor for the LLM. For details, see ECS Types.

      Node Storage Type

      When Instance Specifications is set to GP-accelerated, select a node storage type, which can be Common I/O, High I/O, or Ultra-high I/O. The node storage types available vary depending on the selected AZ and instance specifications, as well as the regional environment.

      Node Storage Capacity

      When Instance Specifications is set to GP-accelerated, set a node storage capacity. The value range varies depending on the node flavor selected.
    • Enterprise Project

      When creating a cluster, you can bind an enterprise project to the cluster if you have enabled the enterprise project function. You can select an enterprise project created by the current user from the drop-down list or click View Enterprise Projects to go to the Enterprise Project Management Service console and create a new project or view existing projects.

  3. Click Next and configure the network settings of the service.
    Table 6 Network settings

    Parameter

    Description

    VPC

    A VPC is a secure, isolated logical network environment.

    Select the target VPC. Click View VPC to go to the VPC management console and check the name and ID of existing VPCs. If no VPCs are available, create one.

    NOTE:

    The VPC must contain CIDRs. Otherwise, cluster creation will fail. By default, a VPC contains CIDRs.

    Subnet

    A subnet provides dedicated network resources that are isolated from other networks, improving network security.

    Select the destination subnet. You can access the VPC management console to check the names and IDs of existing subnets in the VPC.

    Security Group

    Select a security group for the cluster. A security group serves as a virtual firewall that provides access control policies for clusters.

    To select a security group that meets your requirements, click View Security Group to go to the security group list, where you can check the details of each security group.

    The selected security group must allow ports 30275 and 30277 in the inbound direction. Otherwise, the cluster may be inaccessible to external services.
  4. Click Next and confirm the configuration.
  5. Click Confirm.

    The KooSearch Document Q&A page is displayed. The service you enabled is displayed in the service list and its status is Creating. When its status changes to Available, the service is created successfully.

    If the service fails to be enabled, try correcting errors by following the instructions on the web console.