VMware Cloud Community
JMills
Expert
Expert

Spectacular MONITOR PANIC post-VMotion (ESXi 5.0u1 VM)

Alrighty, I have a fun one:

First-tier VM (VM on physical HW) experiences a spectacular VMM Monitor Panic during resume post-VMotion.

The panic is always the same, but the particular vCPU that hits it "first" varies (vcpu-0 .. vcpu-5 on a 6 vCPU VM)

MONITOR PANIC: vcpu-4:VT launch failed; VMCS valid (error code 7)

...and here's where it gets fun:

The problem isn't reproducible thus-far when the VMM is running in Debug mode.

Phys Hosts are identical (hardware, BIOS, ESXi version, Configuration)

VM is 6 vCPU + HW/MMU virtualization + ESXi 5.0u1

vESXi "host" (inside the VM) is either idle, or has a small number of ESXi 5.0 VMs running

2nd tier VMs are 2 vCPU + Auto-HW/MMU virtualization + ESXi 5.0

I have 24 of these VMs, all seem to experience the problem with a 75%+ likelihood of incidence post-VMotion

vmware.log excerpt just prior to event:

10-10T01:03:33.012Z| vcpu-0| Transitioned vmx/execState/val to poweredOn
10-10T01:03:33.012Z| vcpu-0| Migrate: cleaning up migration state.
10-10T01:03:33.012Z| vcpu-0| MigrateSetState: Transitioning from state 11 to 0.
10-10T01:03:33.064Z| vcpu-0| Monitor has started
10-10T01:03:33.064Z| vcpu-5| Monitor has started
10-10T01:03:33.064Z| vcpu-4| Monitor has started
10-10T01:03:33.064Z| vcpu-1| Monitor has started
10-10T01:03:33.064Z| vcpu-2| Monitor has started
10-10T01:03:33.064Z| vcpu-3| Monitor has started
10-10T01:03:33.075Z| vcpu-0| MonitorInitNumaUnmapVMM
10-10T01:03:33.082Z| vcpu-0| SVGA: Registering MemSpace at 0xd8000000(0xd8000000) and 0xd0800000(0xd0800000)
10-10T01:03:33.108Z| vcpu-0| CPT: vmstart
10-10T01:03:33.108Z| vcpu-1| CPT: vmstart
10-10T01:03:33.108Z| vcpu-3| CPT: vmstart
10-10T01:03:33.108Z| vcpu-2| CPT: vmstart
10-10T01:03:33.108Z| vcpu-4| CPT: vmstart
10-10T01:03:33.108Z| vcpu-5| CPT: vmstart
10-10T01:03:33.115Z| vcpu-1| MONITOR PANIC: vcpu-4:VT launch failed; VMCS valid (error code 7)

Thoughts?

Reply
0 Kudos
1 Reply
JMills
Expert
Expert

JMills wrote:

.

.

.

The problem isn't reproducible thus-far when the VMM is running in Debug mode.

It took some extra beating (more VMotions) but I was able to trigger the same event with the VMM in Debug mode.

vmware.log excerpt just prior to event:

10-10T23:44:08.029Z| vcpu-0| Transitioned vmx/execState/val to poweredOn
10-10T23:44:08.029Z| vcpu-0| Migrate: cleaning up migration state.
10-10T23:44:08.029Z| vcpu-0| MigrateSetState: Transitioning from state 11 to 0.
10-10T23:44:08.072Z| vcpu-2| Monitor has started
10-10T23:44:08.072Z| vcpu-3| Monitor has started
10-10T23:44:08.072Z| vcpu-4| Monitor has started
10-10T23:44:08.072Z| vcpu-0| Monitor has started
10-10T23:44:08.072Z| vcpu-1| Monitor has started
10-10T23:44:08.072Z| vcpu-5| Monitor has started
10-10T23:44:08.085Z| vcpu-0| MonitorInitNumaUnmapVMM
10-10T23:44:08.099Z| vcpu-0| SVGA: Registering MemSpace at 0xd8000000(0xd8000000) and 0xd0800000(0xd0800000)
10-10T23:44:08.146Z| vcpu-0| CPT: vmstart
10-10T23:44:08.146Z| vcpu-5| CPT: vmstart
10-10T23:44:08.146Z| vcpu-3| CPT: vmstart
10-10T23:44:08.146Z| vcpu-4| CPT: vmstart
10-10T23:44:08.146Z| vcpu-3| VM-entry failure (3130): EPTP_EMT(eptp) == MTRR_TYPE_WB
10-10T23:44:08.146Z| vcpu-3| VM-entry failure (3131): EPTP_GAW(eptp) == EPTP_GAW_48
10-10T23:44:08.146Z| vcpu-2| CPT: vmstart
10-10T23:44:08.146Z| vcpu-1| MONITOR PANIC: vcpu-3:VT launch failed; VMCS valid (error code 7)

...anyone? 🙂

Reply
0 Kudos