In certain rare scenarios, the system logger and other processes on the ESXi host may not be able to update files in the scratch space. This is typically, though not exclusively, associated with another event that causes vSAN storage operations to fail to process IO from these processes, usually due to a loss of the underlying storage or the storage being filled to 100% capacity. When this issue occurs, it can cause the host to become completely non-responsive, including the inability to interact with the host via a command-line interface.

If this issue occurs, reboot the ESXi host to recover manageability of the host. You may have to resolve the underlying storage issue by migrating VM's off of the datastore with another host if it is full or resolving an issue with the HBA that host is using to connect to the storage if appropriate.

To prevent this issue occurring, do not store system logs or scratch locations on the vSAN datastore.

For more information about configuring system scratch partitions and the syslog configuration, see:

• Creating a persistent scratch location for ESXi 4.x/5.x/6.0 (1033696)
• Configuring syslog on ESXi (2003322)"