SOLVED!!!!! thanks CMB!!!

(have 4 nics total on board, all intel i354- and I kept 1 for esxi mgt functions.) (VMX3 works too- I also manually matched physical MACs w. NICs in PF vm options- I don't know if that works better… or what) (the 100 speed is just my old switch- I wasn't sure if the new switch was bad or not.) (will put my other VMs on the LAN vswitch2)

Enclosed are a couple pics (stock, wrong, and working)- so if others have this issue.... they can see how to fix.


Don't do it that way.  For your (bridged) WAN, use an IP on your private network like 192.168.2.1.  Then you for your Intnet1 LAN and Intnet2 DMZ (if you need a DMZ), use 10.0.0.0/24 and 172.16.1.0/24

Did some testing and those are my findings for Virtio NICs on VirtulBox:

2x Virtio (checksum offload disabled):
Transfer: 6mb/s, guest CPU 100% NIC1 + 70% NIC2
Host CPU usage: ~60-70%

2x e1000:
Transfer: 6mb/s  guest CPU 60% NIC1 + 60% NIC2
Host CPU usage: ~50-55%

Transfer: 10mb/s  guest CPU 85% NIC1 + 75% NIC2
Host CPU usage: ~63%

That's exactly how I did it too.

Great. Thanks a lot.

@Moohouse:

Have you tried disabling APM and/or ACPI for the VM, to see if this locates your problem?

Hi

Thanks for answer. If I do that I'm getting a panic. But I don't think the problem is ACPI because I've already seen same setup working. I have another pfsense on the same host which is working. The only difference I see is the bad one is on a "sata" bus and the working one on a "virtio" bus.

Thanks

Sorry for the late reply!

I do appreciate the replies posted thus far and it appears (on the surface) that it will be straightforward providing I assign the 2 Wan and 1 Lan physical ports correctly to the VM and such.

Jeah, that worked, thank you!

Yeah saw that, thanks turning off the txcsum and rxcsum seems to fix the problem..

@johnpoz:

And what card to you have exactly? http://www.amazon.com/dp/B00EKQN2KK/?tag=pcpapi-20 And what is your N router? http://www.linksys.com/en-eu/products/routers/WRT160NL Most cards are 2x2 streams at best.  Do you have AC clients?  If not that card having AC gets you nothing.  Where did you get the card, someone gave it to you for free? My Dad's office motherboard died. Because AC pc cards are not all that cheap.  Never in my life seen a dual band card that would allow you to have both 2.4 and 5ghz up a the same time, etc. This is exactly why Im doing this. To Learn. I did not know this

You do understand 5ghz N and AC is only 5 is lot less coverage than a 2.4..  So sticking some little card in a box you have in the corner is not going to give you much coverage.
I told you. My house is not large and my home office is centrally located on the second floor.

I am all for experimenting..  Please do fire up pfsense its a great product, and I love running mine on esxi.. But wanting to use it as an AP is just not something you should be toying with to be honest, not when you could pick up a dual band N access point for pennies.
Why exactly should I not toy with it? What will it hurt? Im going to be honest with you here. After have my idea being called "beyond stupid" and this post here, I feel Im being attacked for asking a question and wanting to try something. I know I can't read tone and body language through text so I could be taking this wrong but I dont think so.

If your going to want to play on improving your network - trying to stick some pc card into your host and use it a AP is just not the right road to go down..  That is my personal and professional opinion, take it or leave it.

If you're still seeing "Dec  5 08:39:49 pfSense kernel: (ada0:ata1:0:1:0): Error 5, Retries exhausted", you're having a different issue of some sort. Any issue you're having on 2.2.1 at this point is completely different from #4445, that's definitely completely fixed in 2.2.1. Those disk error logs could potentially cause similar symptoms if they're still occurring.

@johnpoz:

Its a few years old, its a HP Microserver N40L, it has AMD Turion II 1.5ghz dual core.  I bumped it to 8GB of ram, using a SSD for datastore and added some nics.  It runs really sweet, uses very little power,  I have 5 disks in it not counting the SSD.  I get 100MBps pulling files from the storage VM that I have raw mounted the other disks too.

Nice.  I'm just hoping my Xeon X3 will perform well in conjuction with my Plex server.  My site to site VPN should be the most taxing thing it has to handle.

I'm curious, what is the recommended method for disabling tx checksums on the hypervisor side for kvm/qemu? I'm running proxmox and it creates tap interfaces for each VM (tap110i0), ethtool doesn't seem to support tap interfaces, it returns:

ethtool -K tap118i0 tx off Cannot change tx-checksumming Actual changes: generic-segmentation-offload: on

I would prefer to not disable it on the bridge interface (vmbr0) due to the performance hit johnkeates mentions. Also, does this only affect virtio, is e1000e unaffected?

SOLVED!

a simple "sudo dhclient eth0:1" solved it…

Probably the same geniuses who made my KVM with a hardcoded command key of Num Lock.

If I boot a VM and it has focus, the Num Lock toggling by the VM causes the KVM to switch. Brilliant.

could you try with a clean install without vmware-tools ? (to try with the builtin vmxnet3 drivers).

most people manage to get around 1.5-2 gbit/s  pushed through when using vmxnet3 adapters  … so something is probably going wrong.
are the cpu's maxed out when checking this from your vsphere client?
see this thread: https://forum.pfsense.org/index.php?topic=87675.0

I can report that the 2.2 releases, both Beta and Release, work fine with Hyper-V, and with CARP.

6 Pf's on different hosts/clusters & I have a CARP array doing OpenVPN Site-to-Sites to other PF's, works fine.

Only thing you need to do i've found is:
Use a Gen1 VM.
Set the disk to be fixed size.
Enable MAC Spoofing on the NICs that will have CARP addresses.
Disable dynamic memory.