SDDC upgrade precheck fails with ESXi error that reads "Directory /locker on host does not support the upgrade for resource, Please check"

Products

• VMware Cloud Foundation

Issue/Introduction

• When performing an upgrade pre-check in SDDC Manager, the process fails with one or more of the following errors:
  
  
  • Directory /locker on host does not support the upgrade for resource, Please check
  • ESXi Temporary Directory Exist
  • ESXi Maintenance Mode


• The /var/log/vmware/vcf/lcm/lcm-debug.log on the SDDC Manager contains entries similar to:
  
  YYYY-MM-DDTHH:MIN:SEC ERROR [vcf_lcm,######,precheck_id=####-###-####,resourceType=ESX,resource_id=####-###-####] [c.v.e.s.l.p.impl.esx.EsxUtils,Precheck-97] ERROR occurred fetching the available disk space for the directory: /locker
java.lang.NumberFormatException: For input string: "Available.



• Running the df /locker command directly on the affected ESXi host returns:

  # df /locker
Unable to statvfs(/locker): 5
Filesystem Bytes Used Available Use% Mounted on

Environment

VMware Cloud Foundation 4.x

VMware Cloud Foundation 5.x

Cause

An inaccessible or unresponsive datastore attached to the ESXi hosts prevents the Life Cycle Management (LCM) service from parsing disk space information for the /locker directory. Because the service cannot retrieve a numeric value for disk space, it encounters a NumberFormatException when it receives the string "Available" instead of the expected integer.

Resolution

This issue is typically environmental and requires the removal of the unresponsive datastore.

1. Log in to the affected ESXi host(s) via SSH.
   
   
2. Verify the /locker directory status by running: df /locker
   
   • If the output returns Unable to statvfs(/locker), an inaccessible volume is present.
     
     
3. Run df -h to identify any volumes showing 0 bytes or those that do not respond to the command.
   
   
4. Unmount the inaccessible volume from all hosts in the cluster (not just the host reporting the error).
   • For detailed unmounting steps, refer to: How to detach a LUN device from ESXi hosts.
     
     
5. After unmounting, re-run df /locker on the ESXi host to ensure the Unable to statvfs(/locker): 5 error is cleared and valid partition information is returned.
   
   
6. Re-run the SDDC Manager upgrade pre-check.

Note: If the issue persists, ensure SSH host keys are up to date. Refer to How to update the SSH host keys on the SDDC Manager.