VMware vSphere Virtual Machine Panic due to Disk Mirroring and Metadata Synchronization Race Condition

Products

• VMware vCenter Server

Issue/Introduction

• A virtual machine may experience an unexpected crash and subsequent vSphere HA restart. The vmware.log will indicate a VMX process panic with a Signal 6 (SIGABRT) error. This issue occurs specifically at the VM process level; the underlying ESXi host and other resident VMs remain stable and unaffected.

• In the vmware.log of the affected VM, you see entries similar to:

2026-04-02T19:48:24Z[+0.000] Wa(03) vcpu-0 - Caught signal 6 -- tid 7041402 (addr 6B7039)
2026-04-02T19:48:24Z[+0.000] Cr(01) vcpu-0 - PANIC: Unexpected signal: 6.
2026-04-02T19:48:24Z[+0.258] Wa(03) vcpu-0 - A core file is available in "/vmfs/volumes/.../vmx-zdump.000

Cause

The crash is caused by a rare race condition between disk mirroring operations and disk metadata updates. When these two internal processes attempt to access the same memory structure simultaneously without proper synchronization, memory corruption occurs, leading to a VMX panic.

Resolution

There is currently no workaround for this issue. Engineering is working on a permanent fix to ensure background disk tasks are properly synchronized.