On 10.08.2017 15:09, Andrew Moore wrote: > Both of those reports were me. I suspect the issue may be isolated to > the HPE custom implementation of the ESXi 6.5u1 build. I haven't seen > any similar reports of people using the vanilla 6.5u1 build. Not surprising. It wouldn't be the first time HPE horribly botched their ESX custom ISOs. (Which is the prime reason I don't *ever* use custom vendor ISOs from any vendor in the first place.) > Interestingly none of the fixes that have been discussed work with this > build either. This includes disabling the rx-mini buffer (# ethtool -G > <interface> rx-mini 0) and adding vmxnet3.rev.30 = FALSE to the VMs vmx > file. Very strange, indeed. > The only way I've managed to restore stability is by removing vmxnet3 > out of the equation completely and changing to the e1000 NIC type. Using a HW version lower than 13 should also help. Unfortunately the sample size of people reporting failure or success is very small at the time, a conclusive result can't be drawn, I am afraid. Grüße, Sven.
Attachment:
signature.asc
Description: OpenPGP digital signature