A couple of weeks ago we had a problem with a VMWare server at one of our customers after an upgrade, think it it was to 6.5 or so. One of the VMWare guests had a NEC usb 3.0 pci card attached for backup purposes, and that caused a purple screen of the entire host once that machine booted up. We already took the machine out of maintenance mode cause all looked oke at first.
So after every reboot we had a purple screen once that guest booted up, like the autostart tells it to do. Eventually we fixed it by re-installing an older version of esxi. We hade to make an adjustment at a memory settings in that vm to make it work again.
But for when ever we encounter something like this again; Is there a possibility to put the host in maintenance mode, or to disable autostart, before esxi is booted up? Like a safe mode for the host....
if I unterstand you right, after you reboot the ESXi server, the server falls in a PSOD reboot.
Normally before you reboot a system you set it in maintenance mode and when you start it up again it's still in maintenance mode until you remove it manually...
That's correct. And the server gives us a psod before we are able to put it in maintance mode or before we could remove a vm from the autostart list. So we have a loop we cannot escape from.
you have to possibility to enter some boot options when you start esxi with shift+O
maybe you can start something like a safe mode...
Unfortunately, there is not option like safe boot on ESXi host.
Query: But for when ever we encounter something like this again; Is there a possibility to put the host in maintenance mode, or to disable autostart, before esxi is booted up? Like a safe mode for the host....
You cannot place the host in maintenance mode before the host completely booted up. Moreover the host will goto the same state what it was before when it was rebooted.
If my understanding is correct , you want to stop the VM from booting which causes the PSOD when you reboot the ESXi host?
1. You can disable autostart of the VM when the host and VM is running fine which will stop the VM to start automatically on further ESXi reboots.
2. If your VM's configuration file (vmx) resides on shared datastore, to avoid reboot loop, you may temporarily rename the VM folder using datastore browser from other host which see this shared datastore, this may prevent the ESXi host to lock the file and boot the VM.
Thank you for your answer. Unfortunately both options don't work out at such a moment.
1. We don't have sufficient time to put the server in maintenance mode, it gave us a psod before we could do that.
2. That is indeed a good tip. Unfortunately in this situation the customer used an internal raid configuration for storage in a proliant server.
maybe you can boot with an usb stick and use kickstart:
Configure Advanced Bootloader and Kernel Options
then I would try to configure the ks.cfg like this:
#Sample kickstart scripted installation
reboot
%firstboot --interpreter=busybox
# SSH and ESXi shell
vim-cmd hostsvc/enable_ssh
vim-cmd hostsvc/start_ssh
vim-cmd hostsvc/enable_esx_shell
vim-cmd hostsvc/start_esx_shell
vim-cmd hostsvc/maintenance_mode_exit
please test this first on an non productive environment