What Can I Do If Disk Scale-Out Fails When There Is Heavy I/O Workload for SCSI Disks?

Symptom

After the capacity of SCSI disks (more than 10) of Linux ECSs is concurrently expanded online, the capacity of some disks remains unchanged after the expansion.

For example, after you increase disk capacity from 1 GB to 2 GB, the output of the lsblk command shows that the disk capacity of sdb/sdc/sdd/sdr remains unchanged.

Figure 1 Error message

Possible Cause

When the I/O workload of SCSI disks is heavy, the disk queue cannot process the capacity expansion requests in a timely manner, so the capacity of some disks is not changed.

If a write I/O operation is triggered on the disk, the Linux ECS will check the capacity again to make the expansion take effect.

Solution

On the Linux ECS, perform a write I/O operation on the disk whose capacity remains unchanged to make the expansion take effect. For example:

Run the following command to create an empty file in the mount point directory of the disk and perform a write I/O operation:

touch file