Skip to main content

Disk Operation and Maintenance

Scenario 1: High Disk Usage

Cause of the Problem

It is recommended to check the metric "Node Disk Usage Trend Chart". If the disk usage continuously exceeds the 85% warning line for 30 minutes, it indicates high disk usage. Possible causes include:

(1) A large amount of invalid data assets occupy disk space

(2) Datasets are not cleaned up in time

(3) Unreasonable update methods for large datasets

(4) Cache files, backup files, log files, offline upgrade files, and historical images occupy disk space

Troubleshooting Ideas

We recommend troubleshooting as follows:

(1) Refer to "Disk Usage Distribution of Each Node" to understand disk space usage and find the resources occupying the most space

Optimization Measures

a. If cache files, backup files, log files, offline upgrade files, and historical images occupy a large amount of disk space, it is recommended to contact Guandata for manual cleanup. If business allows, you can also adjust the automatic cleanup cycle for these types of files.

b. If business datasets occupy a large amount of disk space, refer to the troubleshooting ideas below for operation.

(2) Refer to the metrics "Datasets with No Consumption" and "Datasets Generating Invalid Consumption" to identify low-value business under the premise of not affecting business

Optimization Measures

For these resources, we recommend gray decommissioning. Gray decommissioning means setting the dataset update mechanism to "manual" and observing whether it affects business. If there is no impact, proceed to clean up and delete.

(3) Refer to the metric "Top 20 Datasets by Storage Space Occupied" to identify large datasets

Pay attention to datasets that occupy a large proportion of storage space (dataset storage space >5% of disk space). You can click the dataset name to jump.

Optimization Measures

Under the premise of not affecting business, consider the following solutions to control dataset size:

a. Determine whether the dataset needs frequent updates. If timeliness is not critical, reduce the update frequency or adjust the update cycle.

b. Set data cleaning for the dataset. You can clean up and delete data that is no longer needed according to business needs. Note: The "Data Cleaning" function is only available for datasets imported from files or connected to databases. View datasets, direct connection datasets, and entry datasets do not support this function.