A quick lesson I’ve learnt this week.
If using VMWare ESXi 5.5 (the most recent release), then make sure you update your VMs (or create new VMs) to use the VMXNET 3 adapter rather than the aged E1000 adapter.
The reason?
An ESXi 5.x host with virtual machines using an E1000 or E1000e virtual adapter fails with a purple diagnostic screen.
The article suggests this may happen with VMWare ESXi 5.0 and 5.1 as well, but I’ve never seen it on those releases. After replacing the adapters with VMXNET 3 ones and reconfiguring the networking on the VMs we’ve had no crashes since.
Funny that you say that. A couple days ago an entire cluster of 4 vsphere pinked screen on us… We are going around the guest and changing the VNic. Amazing that this bug is around since 5.0 and still not fixed.
Agree 100%. It is surprising that VMWare haven’t prioritised this, given it takes an entire infrastructure down.