I wouldn't hold my breath on the Cavium actually working with FreeBSD. Their "drivers" for recent versions of FreeBSD do not hook into the crypto(9) framework, if you can even get your hands on them.

In most cases, not so easy. It depends on what you're trying to do, it may not be feasible in some cases and in other cases it may be possible, but not simple.

@larsN:

I have this when i writing usbconfig dump_device_desct for my TrendNet tew 645ub

Apparently this device s supported by the same Linux driver that supports a device supported by the FreeBSD run driver. But the config output shows a TrendNet Vendor code (0x157e) and it appears that this vendor code is not in the list of supported devices in the FreeBSD run driver. Maybe adding this device's code will be sufficient to get it to work with the run driver, maybe that won't be enough.

@larsN:

If i buy a wireless card, what do you think about this : http://www.ldlc.com/fiche/PB00066238.html (atheros card, the best from JimP's spread sheet ).. but it's not at http://www.freebsd.org/releases/8.1R/hardware.html

Can you confirm ?

That card is a D-Link DWA556 PCI-Express card. If I recall correctly there has been at least one reference to that card in the pfSense forums. A search might turn up a report from an actual user.

@larsN:

I search a wifi card N compatible Pfsense 2.0.1 with an separate antenna , but i don't find.

Are you looking for a WiFi card with a detachable antenna and supported in pfSense? The Alfa AWUS036NH appears to have the same chipset as a device I have that works with the run driver on pfSense 2.0.1. I have seen references to this device in the pfSense forums.

The end device is a Verizon router that only operates as an AP. Currently, I am using an over powered (for the purpose) laptop running under Ubuntu 11.10, and the solution works just fine. I connect the Alfa as a client to the AP down the way and then setup up an adhoc connection using the built-in card to give access the house and it works fine. I wanted to just do this same setup using a low end laptop as opposed to my my i7 machine.

I recently built the following[1]. I would go with the Intel dual port server NIC.

[1] http://forum.pfsense.org/index.php/topic,44269.0.html

Mmm yes it looks to be a mini PCI-e card however accessing it seems quite an involved process!  ::)

Steve

@Hultner:

Could you try testing 1/1Gbs Full Duplex. upload and download simultaneously)? Would be grateful.

Sorry that currently I am not sure how to set up such a test using Iperf and it cannot be conducted using the client I am using now as it is by the ISP, concurrent test cannot be carried out.
In case I managed to make it, I will try to update with you.

I just purchased a new TP-Link modem TD-8840 and seems to work fine now. I think I will stick to the new modem.

Yes, it would be necessary to have a USB drive to run on that, but even with the cost of a USB flash drive included, it is the cheapest low-power device suitable for networking applications that meets the RAM requirement for pfSense.  It would be nice to have pfSense working on a platform this cheap and powerful someday. :)  In my tests, a device with a 400 MHz version of that type of CPU can process more traffic through it (in bridged mode; haven't tested any routing) than an Alix when running the same OpenWrt revision on both (only way I could get the same thing on both).

I don't know about the overlay, but uzip could probably be used as an alternative to squashfs, from what I've heard.  Of course, this probably wouldn't be applicable for pfSense, since you wouldn't be likely to have the root fs on the device's own flash memory.

after play a moment with my 3G usb I got a real information about it, my usb 3g is E173s so it did not support current by PF-Sense.

and I got something from http://www.draisberghof.de/usb_modeswitch/bb/viewtopic.php?t=552

# Huawei E173s DefaultVendor=  0x12d1 DefaultProduct= 0x1c0b TargetVendor=  0x12d1 TargetProduct= 0x1c05 CheckSuccess=20 MessageEndpoint= 0x0f MessageContent= "55534243000000000000000000000011060000000100000000000000000000"

any body have idea? ??? ???

They're unchanged.

@stephenw10:

I expect this machine will now fail!  ::)

Excellent, you've taken your first steps towards reducing your exposure to unexpected outages.

My process goes something like the following:

Assume everything will fail at the absolutely worst possible time.

Determine what the worst possible time is for a failure for each service.

Determine how all the systems that affect a given service can possibly fail.

Determine what, if anything, I can do about the potential failure points.

For any potential failure I can't avoid, create a plan of action and documentation for what I will do when it happens.

Hi,
  Thanks for your reply!
No I haven't done either of those things. I noticed that there is a acpi.ko in /boot/kernel, and I renamed the original file and replaced it with my custom-built one. Probably not the correct way to go about it. I am still quite hazy about the bsd boot process and the difference between builtin "modules", modules loaded from initfs and dynamically loadable modules. As I mentioned in a previous post, I'm more used to Linux.

I'll try modifying the boot configuration to disble acpi, then load my custom acpi.ko later on - I guess none of this will really work if acpi.ko really is built into the kernel. The only solution then would be a completely rebuilt kernel.

Cheers!

Just upgrade the BIOS and it'll come up after you reboot. No need to reinstall.

@stephenw10:

By that logic before Sandy bridge was released no one could use any features of pfSense other than firewalling?  ;)

You need to qualify that statement with a bandwidth requirement. Clearly if your WAN connection is, say, 10/1Mbps you can easily use a Alix board and have plenty of cpu cycles spare.

@jimp:

Indeed I have a number of Atom 330 1U pfSense boxes in the wild running squid, squidGuard, ntop, etc, and they are great performers.

The CPU requirement depends completely on the amount of throughput you need (in PPS, or more generally, Mb/s), along with the number and type of services you want to run.

Sweeping generalizations like that really aren't helpful to anyone.

True, I stand corrected :)

@stephenw10:

Probably the number one reason is that you can build a completely passively cooled Atom system quite easily as it's maximum power dissipation is low. Although the average dissipation of a Sandy Bridge cpu is similar, perhaps even lower, the maximum is far higher so you must allow for that in your cooling solution.

Yes, but can also be done with a SB setup. I still think a SB setup now is a better choice.

I don't see any trace of hostap even in FreeBSD 9 for iwi, so I wouldn't really get your hopes up. That driver you linked to is for Linux, so probably not possible to translate it over.

Certainly less trouble to just get a supported card rather than fight the iwi.

@MMacD:

You were able to get the S1200 with a G620 for $163?  When I look around, I can only see the board itself for that price.  Newegg wants $70 for a 620, bringing the price to ~$230.

If the $163 was just for the board, what persuaded you to go with the G620, considering that it's the high-wattage choice (65W) compared to the D525 (13W) and the diff in total hw cost is minimal?

No, that price was just the motherboard.

I chose the Sandy Bridge chip because it gets SO much more done than an Atom for the same price. TDP should not be used to count real world power usage. It's not even remotely close. Fully loaded, I can't get my G620 to use more than 45 watts.

ups connected and detected properly but refuses to start :(

Thanks for all the info guys.

I tried putting the msk on WAN however it would then just crap out hopefully 2.1 comes out soon.  So it currently is sitting on LAN but I might have another pci nic to replace until the fix.

Excellent! Thank you Stephenw10.

The list brings another question regarding acpi, but I'lll start a new thread for that.

Cheers!