How to kill an Unresponsive VM (ESXi 5.x)

When a Virtual Machine crashed it might happen that you cannot power it off with the vSphere Client. In that case, you have to kill the VM through the ESXi command-line.

Connect to the ESXi host with SSH. (If a virtual machine crashed in a cluster and you cannot identify where it is running, you have to identify the ESXi Hosts where the VM is running)

Locate the World ID with esxcli vm process list or vmdumper -l

[root@esx1:~] vmdumper -l
wid=462925 pid=-1 cfgFile="/vmfs/volumes/5474e3ca-af7a8515-9aef-001b2193b3b0/vcsa.virten.lab/vcsa.virten.lab.vmx" uuid="56 4d a9 f8 05 4a d6 a6-cb 08 de 39 a6 e7 a4 38" displayName="vcsa.virten.lab" vmxCartelID=462924

[root@esx1:~] esxcli vm process list
vcsa.virten.lab
 World ID: 462925
 Process ID: 0
 VMX Cartel ID: 462924
 UUID: 56 4d a9 f8 05 4a d6 a6-cb 08 de 39 a6 e7 a4 38
 Display Name: vcsa.virten.lab
 Config File: /vmfs/volumes/5474e3ca-af7a8515-9aef-001b2193b3b0/vcsa.virten.lab/vcsa.virten.lab.vmx

The World ID in that case is 462925

Kill the World ID. The kill command has 3 options which should be used consecutively. Verify that the VM is no longer running with vmdumper after each step. If the process is still active (vmdumper output still displays the virtual machine) go to the next step:

1. soft - Give theVMX process a chanceto shutdown cleanly (like kill or kill-SIGTERM)

   [root@esx1:~] esxcli vm process kill -w 462925 -t soft
   [root@esx1:~] vmdumper  -l |grep 462925

2. hard - Kills the process immediately (like kill -9 or kill-SIGKILL)

   [root@esx1:~] esxcli vm process kill -w 462925 -t hard
   [root@esx1:~] vmdumper  -l |grep 462925

3. force - Last resort attempt to kill the VM

   [root@esx1:~] esxcli vm process kill -w 462925 -t force
   [root@esx1:~] vmdumper  -l |grep 462925

4. If the process is still running, reboot the ESXi host.