How to Resolve the "Input/output error" Generated When System Commands Are Executed on FusionCompute

Publication Date:  2015-01-26 Views:  280 Downloads:  0
Issue Description
After the FusionCompute runs for a period of time at a site, the execution of many commands returns the "Input/output error", as shown in Figure 1.
Figure 1 Input/output error
Alarm Information
1. Run the lsscsi command to query for the system disk ID. Figure 2 shows an example.
Figure 2 Querying the system disk ID

In this figure, the system disk ID is 0:0:0:0. The actual system disk ID obtained after the fault occurs is 0:1:0:0.
2. Navigate to /sys/class/scsi_device/system disk ID/device. For this fault, navigate to /sys/class/scsi_device/0:1:0:0/device, and check the system status. It is found that the system disk is offline (as indicated in Figure 3). In this state, the system disk cannot read and write data normally. Hence, the execution of system commands returns "Input/output error" since system commands are stored on the system disk.
Figure 3 System disk in offline state

3. Check the log of the RAID card of the disk. The "Data Scrub Corrected Medium Error" is found.
Figure 4 Log of the RAID card

This error indicates that the disk has been damaged and must be replaced.
Handling Process
Replace the disk.
Root Cause