ESXi host returns "Input / Output error" while opening the log files in /var/log.
Since the logs from the disk location could not be accessed, refer to the kernel messages in the memory by running "dmesg"
It shows SCSI command failure warnings related to the disk repeatedly.
2025-07-10T04:22:44.249Z cpu1:2012367)ScsiDeviceIO: 4167: Cmd(0x45b9e0763f48) 0x2a, CmdSN 0x5ce066b from world 2135457 to dev "naa.<device_id>" failed H:0x1 D:0x0 P:0x0
2025-07-10T04:22:44.249Z cpu1:2012367)ScsiDeviceIO: 4167: Cmd(0x45b9e38f88c8) 0x2a, CmdSN 0x5ce066c from world 2135457 to dev "naa.<device_id>" failed H:0x1 D:0x0 P:0x0
Note: Interpreting SCSI sense codes in VMware ESXi
The naa.<device_id> matches the LUN that backs the scratch partition.
Such situations may lead to :
This issue occurs as the disk that is backing the scratch partition is faulty.
Evacuate the VMs and place the host in maintenance mode.
Involve the hardware vendor to investigate the disk failure.
Refer to Identifying disks when working with VMware ESXi to identify the disk.