Docs

Elastic Compute Service

High instance load issue

Troubleshooting and solution for high disk BPS/IOPS load of ECS instance with Linux system

This topic describes how to use the iostat and Atop tools to troubleshoot issues caused by high disk BPS (bytes transferred per second) and IOPS (input/output operations per second), such as system lag, disk I/O throttling, and increased disk I/O latency that degrade the performance of your business processes.

Procedure

Execute the following commands to install the iostat tool.

CentOS

sudo yum install sysstat -y

Ubuntu, Debian

sudo apt-get install sysstat -y

Execute the following command to query disk usage via the iostat tool.

tip

Functions of command arguments are described below. Adjust based on your needs.

-d: Displays disk usage.

-k: Displays data in KB units.

3Statistical time interval. Adjust based on your needs.

5Number of statistical times. Adjust based on your needs.

iostat -d -k 3 5

Sample output

Parameter description

Parameter	Description
Device	Disk name.
tps	Number of I/O requests sent per second.
Blk_read/s	Number of blocks read per second.
Blk_wrtn/s	Number of blocks written per second.
Blk_read	Total number of blocks read.
Blk_wrtn	Total number of blocks written.

Use the Atop tool to view the disk usage ranking.

Make sure that the Atop tool is installed on the target instance. If it is not installed, execute the following commands to complete the operation.

CentOS

sudo yum -y install atop

Ubuntu, Debian

sudo apt update
sudo apt -y install atop

Execute the following command to view the real-time disk usage ranking.

atop -d

Sample output

Based on RDDSK (amount of data read per second), WRDSK (amount of data written per second), WCANCAL (number of disk activities per second), DSK (disk utilization), and CMD (process name), confirm whether the process with high disk usage meets expectations, and obtain the PID (process identifier) of the target process.

If the process meets expectations

Optimize the program processing logic for this process to reduce disk usage. You may also upgrade your instance configuration, scale up the EBS Volume, or Modify the EBS Volume specification to ensure the program is not constrained by EBS Volume performance, preventing normal operation. For operation details, you can view Modify the instance type, Scaling up the EBS Volume, Modify the EBS Volume specification.

If the process fails to meet expectations

You can execute the following commands to manually terminate the abnormal process.

warning

Terminating the process may cause data loss of the target process.

Terminating key operating system processes may cause the system to restart. Please proceed with caution.

On the Atop monitoring page, press the k key to enter process termination mode.

Enter the target process PID, and press the Enter key to confirm.

Enter 9 and press the Enter key to confirm and forcibly close the process.

Last updated: 2026.06.18 17:49:51