@patch said in Virtualized PFsense - host (linux) cannot ping Pfsense Wan Interface (and vice-versa) - Not rule related: @spyshagg Yuk As a desperate measure I would prefer: using pass through NICs for pfsense. Monitoring the interfaces within pfsense and resetting them via an pfsense watchdog / interface monitor if required. But whatever works Hardware and software configuration are not possible at this point. The problem manifested itself 3 full weeks after deployment and not in the 2 weeks of internal testing prior to deployment. A simple reset does not fix the issue. The vm must be shutdown and restarted. It appears the problem is indeed with the hypervisor blocking packets. thank guys

Well, I reckon its a firmware version issue. When Proxmox boots up, it installs t5fw-1.26.2.0.bin If I pass the whole card through to pfSense, it installs t5fw-1.25.0.40.bin and the interfaces are available for assignment as expected If I just pass a VF through with fw 1.26.2.0, pfSense ignores the card and doesn't load the driver, even though it is present as a PCI device in the VM Its a bit of a catch 22. I can install 1.25.0.40 by passing the whole card through to pfSense, but after that the host can't see the card until it is rebooted and then it updates to 1.26.2.0. I haven't been able to figure out how to prevent Proxmox from updating the firmware. I used ethtool to downgrade the firmware on the host but the pfSense VM no longer boots with VFs attached. Days and days on this :(

@elewsmer please share with us, how you able to set queues greater than one. on pfsense 2.5 doesn't work: https://forum.netgate.com/topic/138174/pfsense-vtnet-lack-of-queues/9?_=1638715122702

@stefansz said in Pfsense 1 nic: I can purchase usb network adaptors and try to do pass though (if using proxmox), but I don’t know how good they are if going via usb USB network cards are not recommended to be used with pfSense. You may set it up on Proxmox and bridge it to pfSense though, but then it's on Proxmox / Debian, how they perform. I don't know. Anyway, to connect all your devices behind pfSense, you need a switch at all. So get a VLAN capable switch, then you can also go with only one NIC on pfSense and set up WAN and LAN on VLANs.

@viragomann I will try after work. Thank you!

@keyser said in vmxnet 3 only autoselect available - limited to 1.4GiB/s (10G not working): If you choose the Intel 1000 adapter i ESXi, it will also show 1000Base-T as a connect option - because the adapter emulates physical hardware. That's what I would expect. That's what hardware emulation should do. VMXnet3 is a paravirtual adapter that does not emulate specific physical hardware, but is software only - and software speed limited only. virtIO is paravirtualized as well. That is all, what I was saying. Above you mentioned paravirtualization as the only reason for not showing any connection speed. But this seems not to be the whole story. Obviously virtIO signals the operating system that it is a 10Gb hardware, but vmxnet3 does not.

@viragomann Ok. Thanks. I have restarted the server. Problem not solved. The other address i want to ping is a public ip. It is linked to DMZ interface.

It completely depends how you have the VM configured in VBox. If the host NIC is passed to the VM then it would not have a connection using that and would need some other connection via the VM. Steve

@thatsysadmin we tested passthrough of the 10G NIC with almost the same results. Bare metal on the same machine works fine. I think this is an issue with esxi and pfsense. We see very high interrupts on pfsense during bandwidth testing on ESXI even if we passthrough the NIC.

@lostnil Have any update on your Proxmox throughput issue? Were you ever able to determine if you could route gigabit traffic through your pfSense VM?

I was able to disable it from the shell. Using pfSsh.php playback disabledhcpd My main issue is using my own DHCP server within VirtualBox to now assign an IP to the LAN interface

Hmm, higher throughput with higher MTU like that implies something in that local link it hitting a PPS limit. Mismatched MTU will cause horrible throughput issues generally. But, yeah as mentioned, the first thing I would be doing here connecting something basic and baremetal to the connection to prove it can do 1Gbps. That really shouldn't be very demanding though. Steve

@canaparo67 said in Hyper-v processor compatibility fatal trap 1: amd epyc 7452 32-core Don't know for sure, but could be an unanticipated condition running a 2013 OS on a 2020 processor. I, on the other hand, run that 2013 OS on a 2008 processor(Intel). LOL Might need to bust out the credit card. Again. No help, I know. [image: 1633449714993-734f226f-9f1b-4de8-9db3-84e0a46dae3c-image.png] [image: 1633449752761-3cc19c57-739c-4fd7-98a5-13a8cc58ffac-image.png]