In scope of a migration process, I was using vSphere Converter to convert some VM's from one to another hypervisor (ESXi 6.5). In that process I used special chars which is not advised (gave issue when trying to replicate them with VEEAM). In order to convert them, I stopped the VM, renamed it (actually converted it to the new name), checked the new VM and removed the old one.
This worked fine for most VM's, but one had severe issues.
When stopping the VM named "[DEV] fhml-srv019 Ubuntu Maarten" (Ubuntu 14.4 upgraded to 16.4) via the ESXi (6.5) gui with the stop button, it took quite long to shutdown. Actually the gui didn't show me that it was stopped. When I checked in Converter after a few minutes, I didn't see this VM at all! But I saw suddenly another (stopped) VM named "/vmfs/volumes/541abbb4-c091ccfe-7be6-40a8f02a99b8/Ubuntu Maarten dev/Ubuntu 14.04 vmx09_1.vmx". After a refresh the ESXi gui showed the same. This file didn't exist on the datastore! The folder had the name of the VM before the previous conversion!?! And in that folder only 1 file was present:
[Roger@FHML-VM002:/vmfs/volumes/541abbb4-c091ccfe-7be6-40a8f02a99b8] ls "Ubuntu Maarten dev" -l
We were able to create a new VM (for development purposes) from a template and were able to restore the vmkd definition file (VMware Knowledge Base ). The vmdk flat file contained all information and was mounted, so all required data could copied to the new VM. So after some extra work no data was lost and work could continue within a few hours 🙂
But I'm a bit concerned as I don't have a clue what caused this issue (corrupting a VM when stopping it via ESXi) and have still some VM's to go. I couldn't find an answer when googling.
Has anyone experienced this issue as well? Or has anyone an idea what could have caused this behaviour and how to prevent it?