OpenSearch:Technical O&M

What can I do if inaccurate statistical results are returned due to a large data amount?

• Each search process includes the following steps: Select one million results from the retrieved results to calculate scores and perform a rough sort. Then, select 200 results from the roughly sorted results to calculate scores and perform a fine sort. Return the top 5,000 results to users. The finely sorted results are first returned, and then the roughly sorted results are returned.

• If a large amount of data exists, results are randomly selected and sorted. Then, estimation is made based on the statistics that are collected on the sorted results. Therefore, the value that is obtained based on the statistics is inaccurate.

How is data synchronized in primary and secondary tables?

• In most cases, the data in the primary table can be synchronized within 10 seconds, whereas data synchronization in secondary tables takes longer. Real-time synchronization cannot be achieved.

• If both the primary and secondary tables contain data to be updated, we recommend that you first update secondary tables and then the primary table. You can refresh the primary table data to update the secondary table data and the indexes. This reduces the data synchronization duration.

• Alternatively, you can wait for one second after the secondary table data is updated. Then, you can search for the primary table data in the source data based on the updated data in the secondary tables, and update the primary table data.

How do I process the data of the DATETIME and TIMESTAMP types?

If a table in an ApsaraDB RDS data source contains data of the DATETIME and TIMESTAMP types, OpenSearch automatically converts the data into milliseconds.

If you want to retain the original time formats, we recommend that you convert the data of the DATETIME and TIMESTAMP types to strings before you import the data to OpenSearch.

How do I use PrivateLink to connect to OpenSearch?

1. Log on to the Virtual Private Cloud (VPC) console. In the left-side navigation pane, click Endpoints.

   In the top navigation bar, select a region based on your business requirements. On the Interface Endpoint tab, click Create Endpoint.

2. Create an endpoint.

   On the Create Endpoint page, specify a VPC, a security group, and at least two zones for the endpoint. Keep the value of the Endpoint Type parameter to Interface Endpoint, set the Endpoints Service parameter to Other Endpoint Services, and then enter an endpoint service name in the input box that appears. In this example, com.aliyuncs.cn-hangzhou.opensearch is used. After you enter the endpoint service name, click Verify to check whether the endpoint service is valid. If the endpoint service is valid, click OK.

3. Check the connection status.

   The endpoint details page displays the connection status of the endpoint.

   After the endpoint is created and then connected, the endpoint is in the Available state.

4. Test the connection.

   Copy the domain name in the Domain Name of Endpoint Service section of the endpoint details page. On an Elastic Compute Service (ECS) instance that resides in the specified VPC, run the following command to test the endpoint connection:

   # curl ep-foo.opensearch.cn-hangzhou.privatelink.aliyuncs.com/v3/openapi/apps/foo/search
   {
     "status":"FAIL",
     "errors":[{
         "code":4016,
         "message":"Authentication info was required for the request"
       }],
     "request_id":"d164fa75cec6cb360c13693acd806941"
   }

OpenSearch endpoint service