VM replication fails during VM reconfiguration task due to ConfigStore "database is locked" error on ESXi Host
search cancel

VM replication fails during VM reconfiguration task due to ConfigStore "database is locked" error on ESXi Host

book

Article ID: 437337

calendar_today

Updated On:

Products

VMware vSphere ESXi

Issue/Introduction

  • VM replication is initiated by third party software. 
  • The reconfigure starts normally and enters the reconfiguring state, but fails when attempting to add the new VNIC details.
  • When hostd attempts to write the new NIC configuration, the ConfigStore repeatedly fails to start a transaction due to the database being locked:

var/run/log/hostd.log

YYYY-MM-DDTHH:MM:SS Hostd[2101348]: Task Created :haTask-4-vim.VirtualMachine.reconfigure-#####
YYYY-MM-DDTHH:MM:SS Hostd[2101340]: State Transition (VM_STATE_OFF ->VM_STATE_RECONFIGURING)
YYYY-MM-DDTHH:MM:SS Hostd[2101340]: VmReconfigure: adding opaquenetwork vnic
..
..
YYYY-MM-DDTHH:MM:SS Hostd[2101340]: error [ConfigStore] Exec failed:'database is locked'
YYYY-MM-DDTHH:MM:SS Hostd[2101340]: error [ConfigStore] Failed tostart transaction
YYYY-MM-DDTHH:MM:SS Hostd[2101340]: Add VNIC info to config store failed: Failed to start a transaction. (retry 0)

  • After exhausting retries, the reconfigure operation aborts and the VM exits the reconfiguring state without applying the NIC changes


YYYY-MM-DDTHH:MM:SS Hostd[2101340]: Add VNIC info to config store failed: Failed to start a transaction. (retry 4)
..
YYYY-MM-DDTHH:MM:SS Hostd[2101340]: Unexpected exception during reconfigure
YYYY-MM-DDTHH:MM:SS Hostd[2101340]: State Transition(VM_STATE_RECONFIGURING -> VM_STATE_OFF)


/var/log/vmware/vpxd/vpxd.log

YYYY-MM-DDTHH:MM:SS error vpxd[3836563]:vim.VirtualMachine.reconfigure failed with vmodl.fault.SystemError
YYYY-MM-DDTHH:MM:SS error vpxd[3836563]: -- ERROR task-16923208 --vm-#### -- vim.VirtualMachine.reconfigure

Environment

ESXi version 8.0.3 

Cause

A vNIC memory leak within the nsxt_vnic_cache is causing unbounded growth of the vm_vif_state object. This unchecked growth consumes a disproportionate amount of the ConfigStore ramdisk and results in a vNIC leak issue.

Resolution

This issue is resolved in VCF 9.0.

There is currently no functional workaround on ESXi 8.0.3 version.