core dump

You may have noticed new warning message after installing ESXi 7.0, this can happen if your ESXi installation is on a smaller local disk (Nested ESXi Appliance would also be affected) which does not have enough capacity for setting up a core dump target or if ESXi is booting from USB which also does not support core dumps by default.

No coredump target has been configured. Host core dumps cannot be saved.

It is certainly recommended that you have a core dump target configured, especially for Production systems. However, if you wish to suppress the warning, there is an ESXi Advance Setting which you can toggle called UserVars.SuppressCoredumpWarning (can be configure using Embedded ESXi Host Client or vSphere UI in vCenter Server) as well as ESXCLI. Below is the command to suppress the warning:

esxcli system settings advanced set -o /UserVars/SuppressCoredumpWarning -i 1

https://twitter.com/virtualdominic/status/488570929383890945

While catching up on my Twitter feed this morning, I saw an interesting question from Dominic Rivera. I remember a few years back I was troubleshooting an issue and required to panic a Virtual Machine, I did not recall the steps that GSS had walked me through but I figure if anyone would know, it would be my buddy Paudie who used to work in GSS. Sure enough, he had the answer which was using a nifty little tool on the ESXi Shell called vmdumper.

UPDATE: It turns out there is an even simpler thanks to Frank Buechsel who is also from GSS. Frank mentioned you can send an NMI request simply using either the vSphere Web/C# Client. In the vSphere Web Client, you just need to right click on the VM and go to All vCenter Actions->Export Systems Logs. To be able to perform this action, you will need the Global->Diagnostic privilege for the the given user.

Next, expand the menu and select Send_NMI_To_Guest option.

If you wish to perform this operation using the vSphere C# Client, you just need to select the VM and then at the top menu and click File->Export System Logs and you will have the option of sending the NMI request this way.

Using the vmdumper utility, you can create a BSOD (Blue Screen of Death) for a Windows or Kernel Panic for Linux Virtual Machines by sending an NMI (Non-Maskable Interrupt) request.

To send an NMI or any other request, you need to first identify the VM's World ID, this can be done using either ESXCLI or vmdumper utility itself.
Using ESXCLI:

esxcli vm process list | grep -A1 '^[VMNAME]'

Using vmdumper:

vmdumper -l | grep [VMNAME]

In the example above, I have a VM called vcenter55-2 running Windows 2008 R2 and the World ID is 23100118. To send the NMI request, you will now run the following command:

vmdumper [WORLD-ID] nmi

If we now take a look at the VM Console, we will see that it successfully BSOD the Windows VM:

Here is a screenshot for sending an NMI request to a Linux VM. It seems the Linux system handles the NMI in a much more graceful manner and the system was still responsive 🙂

After learning about the vmdumper utility, I also performed a quick search online and found VMware KB 2005715 that provides some additional information. It looks like the KB only references ESXCLI as a way to locate the World ID, I will see if we can get the KB updated to also include vmdumper method of identifying the VM's World ID.

Quick Tip - Suppress new core dump warning in ESXi 7.0

Quick Tip - How to BSOD/Panic a Virtual Machine in ESXi?