ESX/ESXi host disconnects from vCenter Server randomly

Products

• VMware vSphere ESXi

Issue/Introduction

Symptoms:

• The ESX/ESXi host goes into a Not Responding or Disconnected state and then connects back to vCenter Server after few seconds
• In vpxa logs or hostd logs, you see backtrace similar to:
  
  [YYYY-MM-DD info 'App' opID=eacf42c3-4b] [VpxLRO] -- BEGIN task-internal-42 -- -- vpxapi.VpxaService.searchDatastore -- 52421f84-1d50-b56a-c835-6a355f639d85
  [YYYY-MM-DD verbose 'App' opID=eacf42c3-4b] [VpxaDatastoreContext] Resolved URL sanfs://vmfs_uuid:4ce680af-0a04ae9a-692e-842b2b46cde6/ to localPath /vmfs/volumes/4ce680af-0a04ae9a-692e-842b2b46cde6/
  [YYYY-MM-DD info 'Libs' opID=eacf42c3-4b] Unicode_AllocWithLength: Error: Couldn't convert invalid buffer [FDM-0903\xDC/\xEB\xF5\xFF\xF5_|0-48D1-A1DB-0B67B2C4BA88-7-55f2002-VCENTER] from UTF-8 to Unicode.
  [YYYY-MM-DD error 'App' opID=eacf42c3-4b] Failed to generate core dump: Fork failed: 38
  [YYYY-MM-DD panic 'App' opID=eacf42c3-4b]
  

Environment

VMware ESX Server 3.5.x
VMware ESXi 3.5.x Installable
VMware ESXi 3.5.x Embedded
VMware ESX 4.0.x
VMware ESXi 4.0.x Installable
VMware ESXi 4.0.x Embedded
VMware ESX 4.1.x
VMware ESXi 4.1.x Installable
VMware ESXi 4.1.x Embedded
VMware vSphere ESXi 5.0

Cause

This issue occurs if there are non-ASCII characters or corrupt folders in the datastore.

Resolution

To resolve this issue you must delete the corrupt folder.

If you cannot delete the folder from the command line or the datastore browser, migrate the virtual machines from the datastore, wipe the datastore, and then recreate the datastore.

Note: You can find the corrupted folder and the corresponding datastore from the backtrace entry similar to:

[VpxaDatastoreContext] Resolved URL sanfs://vmfs_uuid:4ce680af-0a04ae9a-692e-842b2b46cde6/ to localPath /vmfs/volumes/4ce680af-0a04ae9a-692e-842b2b46cde6/
Error: Couldn't convert invalid buffer [FDM-0903\xDC/\xEB\xF5\xFF\xF5_|0-48D1-A1DB-0B67B2C4BA88-7-55f2002-VCENTER]