Intelligent inspection is a preventive maintenance measure used to check the status of the system and promptly identify potential problems. The intelligent inspection system generates reports based on inspection items, enabling you to detect and address issues early. This helps reduce system failures and downtime, ensuring system stability and reliability.

The following is an overview of the intelligent inspection functionality.

1. Log in to the ByteHouse console and switch to the target environment. On the Diagnostics & Optimization tab, open the Intelligent Inspection page, select the Inspection Tasks tab, and click Create Inspection Task.
   

2. On the Create Inspection Task page, enter basic task information, select the inspection type, and view the inspection content.
   

   warning

   Intelligent inspection will consume resources in the current environment. It is recommended to perform this operation during off-peak hours to avoid impacting business operations.

   • The inspection task configuration

     Parameter item	Parameter descriptions
     Inspection task name	Supports custom inspection task names.
     Inspection types	Supports manual inspection and periodic inspection. Manual inspection: After creating an inspection task, you need to manually trigger execution from the inspection task list. Periodic inspection: You can configure the inspection cycle. During the inspection cycle, the system will automatically execute inspection tasks.

   • Inspection contents description

     Categories	Inspection items	Inspection item descriptions and risk level determination rules
     Group system load	The VW expiration time	Days until subscription-based compute group expiration. The rules for determining risk levels are as follows (unit: days): Normal: [365, 7] Medium risk: [7, 3] High risk: [3, 0]
     	Peak CPU usage in the past day	Peak CPU is defined based on a 30-minute average window. Risk level determination rules are as follows: Normal: [0, 90] Medium risk: [90, 95] High risk: [95, 100]
     	Peak memory usage in the past day	Peak memory is defined using a 30-minute averaging window. Risk level determination rules are as follows: Normal range: [0, 80] Medium risk: [80, 85] High risk: [85, 100]
     	Peak iNodes usage in the past day	The peak iNode is defined using a 1-minute average window. The rules for determining risk levels are as follows: Normal: [0, 90] Medium risk: [90, 95] High risk: [95, 100]
     	Peak cache usage over the past day	Peak cache is defined using a 1-minute average window. The rules for determining risk levels are as follows: Normal: [0, 90] Medium risk: [90, 93] High risk: [93, 100]
     Query loads for compute groups	The insert success rate for the past 1 day	Calculation formula: successful insert queries / all insert queries. Rules for determining risk levels: Normal: [100, 98] Medium risk: [98, 95] High risk: [95, 0]
     	Select success rate (%) (past day)	Calculation formula: successful queries / all queries. The rules for determining risk levels are as follows: Normal: [100%, 98%] Medium risk: [98%, 95%] High risk: [95%, 0]
     Dedicated server	Dedicated server expiration time	Number of days remaining before the expiration date of a subscription-based dedicated server. The rules for determining risk levels are as follows (measured in days): Normal: [365, 7] Medium risk: [7, 3] High risk: [3, 0]
     	Peak CPU usage over the past day	Peak CPU is defined using a 10-minute average window, and the rules for determining risk levels are as follows: Normal: [0, 90] Medium risk: [90, 95] High risk: [95, 100]
     	Peak memory usage in the past 24 hours	Peak memory is defined using a 10-minute average window, and the rules for determining risk levels are as follows: Normal: [0, 80] Medium risk: [80, 85] High risk: [85, 100]
     Data tables	Number of unhealthy partition tables	Count how many tables in the current environment contain unhealthy partitions The rules for determining risk levels are as follows: Normal: [0, 0] Medium risk: [0, 100,000] High risk: [100,000, 100,000] ByteHouse provides a dedicated partition health diagnostic feature. For more information about table partition health, see Partition Health Diagnosis.
     Gateway connection count	Number of gateway connections in the past 24 hours	Calculation formula: current TCP connections / TCP connection limit. The rules for determining risk levels are as follows: Normal: [0, 90] Medium risk: [90, 95] High risk: [95, 100]
     API Key	Remaining validity period	Number of days until all users' API keys expire. The rules for determining risk levels are as follows, measured in days: Normal: [365, 7] Medium risk: [7, 3] High risk: [3, 0]

3. Click OK. The system will create inspection tasks.

4. You can view tasks that have been created in the inspection task list. You can also enter a task name in the inspection task list to view inspection tasks.

For tasks where the inspection type is set to manual, you can manually trigger task execution as needed.
In the inspection task list, click the Execute now button in the operation column. In the popup, click Confirm. The system will perform the inspection.

After the system displays the message "Inspection task executed successfully", you can view the inspection task execution results on the inspection report page.

You can view the generated inspection report on the inspection report page to see the inspection results, impacts of any abnormalities, and recommended actions.

View inspection reports #

On the inspection report page, you can view the list of generated inspection reports. Click Inspection Report ID or the View Report button in the operation column to view report details.

The inspection report displays inspection details, including inspection items, inspection values, severity level, inspection category, abnormal impact, handling recommendations, and other information. You can determine whether it is necessary to handle the exception based on inspection values, severity level, impact, and other relevant information.

Download inspection reports #

On the inspection report page, click the Download Report button in the operations column to download the current report.

You can also click Inspection Report ID to open the details page, and then click Download Report.

(Optional) Delete an inspection report #

To delete an inspection report, go to the inspection report page and click the Delete button in the Actions column to delete the current report.