VMware ESX unrecoverable error: (svga)

[TiemenWerkman]

Hey all,

I recently decided to upgrade to esxi 6.5. I first wanted to test my setup as it's somewhat dated.

Now I'm faced with constant crashes when viewing virtual machines in the console, both webconsole and the vmrc.

I've added the last vmware.log and vmmcores.gz files.

The vm that i'm testing is Windows Server 2008 R2 x64. Vmware tools installed succesfully.

I'm not sure what other information is useful, the setup is very basic.

Tiemen Werkman

[EricChigoz]

Hello,

Good morning.

Could you tell us more about your environment?

Does your Hardware Host meet compliance requirments as stated here:

ESXi Hardware Requirements

Last Updated 04/04/2018  16

Add Topic To MyLibrary

< Ask New Question

Make sure the host meets the minimum hardware configurations supported by ESXi6.5.

Hardware and System Resources

To install or upgrade ESXi, your hardware and system resources must meet the following requirements:

Supported server platform. For a list of supported platforms, see the VMware Compatibility Guide at http://www.vmware.com/resources/compatibility.

ESXi 6.5 requires a host machine with at least two CPU cores.

ESXi 6.5 supports 64-bit x86 processors released after September 2006. This includes a broad range of multi-core processors. For a complete list of supported processors, see the VMware compatibility guide at http://www.vmware.com/resources/compatibility.

ESXi 6.5 requires the NX/XD bit to be enabled for the CPU in the BIOS.

ESXi 6.5 requires a minimum of 4 GB of physical RAM. It is recommended to provide at least 8 GB of RAM to run virtual machines in typical production environments.

To support 64-bit virtual machines, support for hardware virtualization (Intel VT-x or AMD RVI) must be enabled on x64 CPUs.

One or more Gigabit or faster Ethernet controllers. For a list of supported network adapter models, see the VMware Compatibility Guide at http://www.vmware.com/resources/compatibility.

SCSI disk or a local, non-network, RAID LUN with unpartitioned space for the virtual machines.

For Serial ATA (SATA), a disk connected through supported SAS controllers or supported on-board SATA controllers. SATA disks are considered remote, not local. These disks are not used as a scratch partition by default because they are seen as remote.

[TiemenWerkman]

Hardware compatibility seems to be in order, seeing as everything works.

ESXi itself works fine, it's just the vm's that are crashing. I've been using ESXi 5.1 for years on this system without any problems at all.

System:

supermicro pdmse+

Intel q6600

4GB ram

Dell perc h700 controller

onboard nics are no longer supported so I am using an intel 82541PI controller.

VT-x is enabled, without it ESXi won't even boot.

ESXi 6.5 update2 is installed with the free license on a 8GB usb drive.

I have been testing for about a week now and everything works well.

The system is reasonably fast considering it's age and much more responsive than ESXi 5.1 which I was running previously.

The html5 web interface works very well.

But from the start my vm's would crash when accessing the console, both webconsole and the vmrc.

Sometimes after a few minutes and somtimes whithin hours.

When  I check the logs they always contain the same lines:

     svga| W115: MainMem: Failed to map physical memory [439790,439791) 1817 ranges, 'Out of memory' (bad0014).

     svga| E105: PANIC: Unrecoverable memory allocation failure

And then a coredump.

I have googled to see if I could find a solution and came across this KB article 2149941.

It's apparantly been solved but I decided to try the workaround, but it did not help.

Tiemen Werkman

[bluefirestorm]

The CPU that you have, Intel Kentsfield Q6600, is not officially supported for ESXi 6.5. Its closest Kentsfield Xeon equivalent X32xx series is also not officially supported for ESXi 6.5.

https://www.vmware.com/resources/compatibility/detail.php?deviceCategory=cpu&productid=9&deviceCateg...

So you are pushing your luck with trying to run ESXi 6.5 successfully on a very old CPU architecture (around 12 years old). The Kentsfield CPU does not have Extended Page Tables (EPT) feature and thus its virtual RAM will be managed by the hypervisor instead of by the Intel CPU VT capabilities and thus resulting in slow VMs. ESXi 6.5 dropped support for many CPUs that don't have the EPT feature.

It looks like the VMware Tools you have is 10.2.1 (10305) which have SVGA 3D driver version 8.15.1.60. You could try updating the VMware Tools to 10.3.2 which has SVGA 3D driver version 8.16.1.0. VMware Tools 10.2.5 has SVGA 3D driver version 8.15.1.65.

https://my.vmware.com/group/vmware/details?downloadGroup=VMTOOLS1032&productId=614

You could also try reducing the screen resolution so that it will use memory.

But if the crashing problem is related to the unsupported CPU, it looks like you may not have much choice. The latest ESXi version you could probably go with based on Kentsfield Xeon is ESXi 6.0U3 or you should be looking to upgrade to newer hardware.

These lines in the vmware.log does not look encouraging.

2018-09-13T14:11:20.419Z| svga| W115: MainMem: Failed to map physical memory [517628,517632) 4 ranges, 'Out of memory' (bad0014).

2018-09-14T05:16:37.232Z| svga| W115: MainMem: Failed to map physical memory [321162,321163) 990 ranges, 'Out of memory' (bad0014).

2018-09-14T05:16:37.637Z| svga| W115: MainMem: Failed to map physical memory [35987,35988) 992 ranges, 'Out of memory' (bad0014).

2018-09-14T05:16:55.497Z| svga| W115: MainMem: Failed to map physical memory [439790,439791) 1817 ranges, 'Out of memory' (bad0014).

2018-09-14T05:16:55.498Z| svga| W115: MainMem: Failed to map physical memory [439790,439791) 1817 ranges, 'Out of memory' (bad0014).

Without EPT in the CPU, it will be up to the hypervisor to map the virtual RAM of the VM to the actual RAM in the hardware and these errors seem to show ESXi 6.5 is failing (for whatever reason).