A datastore fails to mount although
- the backing LUN is fully accessible (paths available, no connectivity issues, or SCSI warnings)
- the LUN has a valid VMFS partition table
- hexdump of the LUN confirms LUN is readable and VMFS metadata appears intact
- a snapshot of the original LUN may have been presented to the host or other changes may have been made to the LUN at storage array level
- logs reports entries similar to: "LVM: 8445: Device naa.xxxxxxxxxxxxxxxxxxxx:1 detected to be a snapshot"
VMware vSphere 6.7
VMware vSphere 7.0.x
VMware vSphere 8.0.x
When mounting a datastore, ESXi calculates the signature of the LUN (based on basic LUN characteristics) and compares it with the signature calculated when the datastore was created and written into the metadata of the VMFS filesystem. When the calculated signature does not match that written in the metadata the LUN is detected to be a snapshot by ESXi.
By design a LUN detected as snapshot will not automatically be mounted.
The LUN will need to be resignatured (or, for testing purposes, force mounted) as per Troubleshooting LUNs detected as snapshot LUNs in vSphere
In rare scenarios, two snapshots of the same LUN might be presented to the same ESXi host. For more information see: Unable to mount a VMFS volume to an ESXi Host