In rare scenarios, a virtual machine snapshot operation may fail when the VM has an I/O filter attached. This failure can cause the VMX process to crash and generate a core dump.
The issue is triggered by a race condition within the FiltMod (I/O Filter module) synchronization logic.
VMware vSphere ESXi 8.x
A race condition occurs between two VMX threads:
One executing FiltModAssignUpcallThread()
Another executing FiltMod_WakeupUpcallThread()
When FiltMod_WakeupUpcallThread() is invoked while filterInfo->upcallWorldSleeping and filterInfo->upcallWorld are being updated in FiltModAssignUpcallThread(), the watchdog may falsely detect a hung thread and stop the VMX process. This results in a VMX crash and snapshot failure.
This issue is resolved in VMware vSphere ESXi 8.0 U3 Patch 06 (Build 24786616).
Workaround: No direct workaround available.
As a precaution, avoid frequent snapshots on VMs with I/O filters until the host is upgraded.