Confused vmware patch PSOD released July 31, 2014

[JoakimA]

I am a bit confused about vmware's PSOD fixes caused by E1000 adapters. I thought we had a final fix for this issue in patch released i January.

Now we have a new fix for this released July 31. (VMware ESXi 5.1, Patch ESXi510-201407401-BG: Updates esx-base (2079118)).

The article states:

"Virtual machines that use an E1000e virtual network adapter might fail with a purple diagnostic screen if Receive Side Scaling is enabled on the virtual machine. The purple diagnostic screen or backtrace contains entries similar to the following".

Can we trust this is solved now? Im about to upgrade our 4.1 esxi hosts to 5.1 and we have about 200 VM's with E1000 nic. Or do I have to disable RSS on all of them to be sure?

[MKguy]

Looking at the patch notes, these are two similar, but not identical e1000 related issues. With the latest one being not as widespread compared to the one fixed in January (probably because of the RSS dependency), where we saw quite a few people running into this issue.

http://kb.vmware.com/kb/2062300

PR 1042045: ESXi host experiences a purple diagnostic screen with errors for E1000PollRxRing and E1000DevRx when the rxRing buffer fills up and the max Rx ring is set to more than 2. The next Rx packet received that is handled by the second ring is NULL, causing a processing error. The purple diagnostic screen or backtrace contains entries similar to:

[....]

http://kb.vmware.com/kb/2079118

Virtual machines that use an E1000e virtual network adapter might fail with a purple diagnostic screen if Receive Side Scaling is enabled on the virtual machine. The purple diagnostic screen or backtrace contains entries similar to the following:

[...]

Can we trust this is solved now? Im about to upgrade our 4.1 esxi hosts to 5.1 and we have about 200 VM's with E1000 nic. Or do I have to disable RSS on all of them to be sure?

We never know whether other issues will pop up, but if your 5.1 hosts have the latest patches then you shouldn't need to disable RSS on your VMs to mitigate the above issue.

While we're at it, a February patch solved another e1000 bug:

http://kb.vmware.com/kb/2072653

This patch updates the esx-base VIB to resolve the issue where virtual machines using the E1000/E1000E adapter might lose network connectivity when the networking heap memory, netGPheap, reaches the maximum limit. In the vmkernel.log file, you see entries similar to:

[...]

http://kb.vmware.com/kb/2072694

With all the recent bugs I would seriously avoid the e1000 vNIC whereever possible. All modern Linux kernels come with a builtin vmxnet3 module and on Windows VMs I would want to have the VMware Tools installed anyways.