This issue occurs when an external application (for example, EMC ECC) connects to the ESX hosts and pulls the virtual machine inventory information. This issue occurs because not all hosts are configured for the external application and the /var/log/vmkernel log for these hosts are flooded with lock holder messages.
To minimize these messages, check for the presence of third-party or partner processes and products, such as a backup product, that may frequently query the virtual machine inventory. Reducing the polling interval or stopping the monitoring reduces or stops the log spew. You need to restart the hostd service for the changes to take effect.
Note: With the additional logging in ESX 4.1, these messages are also reported by the NFC service (used by the Datastore browser). The NFC service needs to check the virtual machine inventory, virtual machine, and storage information when displaying information about a virtual machine. You can ignore these messages when they are reported by the NFC service.