Hardware purchase advice please
-
Yes, I guess.
-
You can add a wifi card at any time. I would recommend starting with the simplest system you can. Adding and testing more interfaces or packages in stages will result in fewer problems.
The biggest advantage of an Atom is that its peak power consumption is low. You can get a system that consumes a very small amount of power most of the time, my own Pentium-M setup is <25W, but because at peak load it consumes more you need much better cooling provision. The Atom can be passively cooled relatively easily for this reason.
If you think that 30W is low, check this out: http://ssj3gohan.tweakblogs.net/blog/8217/fluffy2-59-watt-high-end-desktop-computer.html :)
Steve
Hi Steve,
Thanks for your advice.
I will not be adding the Wifi card right away but I'm curious to which would be the best for general connectivity and range coverage?
I do want the fastest speed possible for wifi.
So I'm looking for the "best" option on what is avail from the shop I will be using below…
Also taking into consideration pfsense support.Here are what I was looking at. At this store....
http://www.mini-itx.com/store/?c=17
I was considering this since it seems to be the better one....
Intel Ultimate-N 633AN Half-Mini PCIe Wireless Card - up to 450 Mbps
I know you'd need all 3 antennas to achieve this.
Which is where perhaps I'd need to drill a few more holes in the case ;)would it be of benefit and most of all is it fully supported by pfsense?
-
Wifi hardware support in pfSense is… limited. ;) Especially under 2.0.X. It's built on FreeBSD 8.1 which was released in July 2010. Anything newer than that is unlikely to be supported. There is no support for 802.11N outside of the specific drivers so although some hardware will work it will only be at 'G' speeds.
There is better support in 2.1 since it's built on FreeBSD 8.3 and some drivers from 9 have been back ported. The best supported cards are those based on Atheros chips-sets.The best source of information is JimP's spreadsheet:
https://spreadsheets.google.com/ccc?key=0AojFUXcbH0ROdHgwYkFHbkRUdV9hVWljVWl5SXkxbFE&hl=enSteve
-
Wifi hardware support in pfSense is… limited. ;) Especially under 2.0.X. It's built on FreeBSD 8.1 which was released in July 2010. Anything newer than that is unlikely to be supported. There is no support for 802.11N outside of the specific drivers so although some hardware will work it will only be at 'G' speeds.
There is better support in 2.1 since it's built on FreeBSD 8.3 and some drivers from 9 have been back ported. The best supported cards are those based on Atheros chips-sets.The best source of information is JimP's spreadsheet:
http://www.draisberghof.de/usb_modeswitch/bb/viewtopic.php?t=983Steve
Hi,
I don't see any spreadsheet.. Am I missing something on the link you provided.
So is there danger of newer cards not been supported at all or just at slower speeds?
-
Nooo, copy and paste fail! :-[
See corrected link above.There is a danger of no support at all. FreeBSD hardware support lags behind other OSs anyway and pfSense lags that by a bit.
Steve
-
I would strongly advise to use a dedicated, separate AP for wifi. pfSense is not really meant to be a wifi router.
-
I would strongly advise to use a dedicated, separate AP for wifi. pfSense is not really meant to be a wifi router.
OK, I think I will skip it anyhow.
Maybe look at it at a later date.thanks
-
I would think the 120W pico-psu will be over kill, I would not expect that box to use more than 30W at any time, mostly a lot less.
Steve
I'm using the Jetway NF99FL-525 with 3 Intel NIC Daughterboard, 1 Compex abg WLAN card and 1 CF attached to SATA via SATA2CF converter. With a standard 300 W ATX power supply it's consuming ~33 W - measured with a power meter. With a picoPSU you will probably reduce power consumption below 30 W. That's why I go along with Steve's opinion: 120 W picoPSU will be overkill :) Of course, my 300 W ATX PSU is overkill as well, but it's currently impossible to get a smaller standard ATX PSU.
-
pvoigt - would this be any better for you?
http://www.ebay.co.uk/itm/HIPRO-100W-ATX-PSU-Power-Supply-Unit-HP-A1463F5-/230822905477
-
With the Sata to CF converter. Should I select IDE or SATA mode in BIOS?
Assuming the bios on this motherboard will offer that choice? -
With the Sata to CF converter. Should I select IDE or SATA mode in BIOS?
Assuming the bios on this motherboard will offer that choice?Tested so far following CF converters:
http://www.delock.de/produkte/F_283_2-5_91697/merkmale.html?setLanguage=en
http://www.delock.de/produkte/F_283_2-5_91661/merkmale.html?setLanguage=enUsing BIOS IDE mode.
Peter
-
pvoigt - would this be any better for you?
http://www.ebay.co.uk/itm/HIPRO-100W-ATX-PSU-Power-Supply-Unit-HP-A1463F5-/230822905477
You may be right because its power data seems more appropriate. One the other hand I am using a 80+ PSU (Enermax). Only if I can directly compare both PSU I will become more wise :)
I suppose most effective would be using a picoPSU. I once decided to go with the ATX PSU because I wanted to use a standard miniITX case which could be re-used by another machine in the future.
Peter
-
The CF card I have is a 16GB… do I just use the 4GB image with that?
pfSense-2.0.3-RELEASE-4g-i386-nanobsd_vga-20130412-1022.img.gz
Is this the correct image?This will be going on an Atom 1.8 with VGA output only.
-
The CF card I have is a 16GB… do I just use the 4GB image with that?
pfSense-2.0.3-RELEASE-4g-i386-nanobsd_vga-20130412-1022.img.gz
Is this the correct image?This will be going on an Atom 1.8 with VGA output only.
This will work without problems. With this image you will have a VGA console. Mostly, a 4 GB CF will be sufficient for the 4 GB image. However, some CF seem to not provide full capacity so e.g a 4 GB image cannot be successfully written to a 4 GB CF. I've read somewhere in the pfSense forum about it but cannot remember the threads - sorry.
I am mostly using 4 GB CF and write a 2 GB image to it - just to be one the safe side :).
Peter
-
pfSense-2.0.3-RELEASE-4g-i386-nanobsd_vga-20130412-1022.img.gz
With the above image I read about serial port and that I won't see a console?
Will I deffo get a cmd promp with the above image?
And will the CF boot up once I install the image to it using Manuel Kasper's phydiskwrite ?thanks
-
pfSense-2.0.3-RELEASE-4g-i386-nanobsd_vga-20130412-1022.img.gz
With the above image I read about serial port and that I won't see a console?
Will I deffo get a cmd promp with the above image?
And will the CF boot up once I install the image to it using Manuel Kasper's phydiskwrite ?thanks
Well, the above image will provide a vga console. If you want a serial console only, you should use
pfSense-2.0.3-RELEASE-4g-i386-nanobsd-20130412-1022.img.gz - without "vga" in its name. A "cmd prompt" will be provided by both images.I had issues using physdiskwrite and was more successful with "win32diskimager". I used version 0.6 a while ago. Latest version is 0.7:
http://sourceforge.net/projects/win32diskimager/. Alternatively, you can use "dd" under Linux (or Unix).EDIT: We are slowly driving off-topic. If you are still having general installation questions I propose to open a new thread in the corresponding pfSense forum :).
-
So if I have a CMD prompt which I've seen where you pick WAN, LAN etc.
What is the difference between VGA and non-VGA versions specifically?
I thought CMD prompt was what VGA was? -
A command prompt is simply a text prompt where you can enter a command. It appears on the console. The console can be accessed either by keyboard and monitor (VGA) or via a serial connection.
If you want to use a keyboard and monitor directly on the pfSense box use a nanobsd_vga image.
@http://www.pfsense.org/index.php?option=com_content&task=view&id=43&Itemid=44:
Embedded (NanoBSD)
The embedded version is specifically tailored for use with any hardware using Compact Flash rather than a hard drive. CF cards can only handle a limited number of writes, so the embedded version runs read only from CF, with read/write file systems as RAM disks. The NanoBSD platform has two OS slices and a config slice. One OS slice is used to boot from, the other is used for upgrades, and the config slice is where the configuration is held separately.
There are two variations of the NanoBSD platform: The default version which uses a serial console, and another that supports using a VGA console. Each of those variations also comes sized for different sizes of CF cards.
The filename for NanoBSD downloads is laid out as follows: pfSense-2.0.3-RELEASE-size-arch-nanobsd.img.gz. In that filename, size can be one of 512mb, 1g, 2g, or 4g. The arch, as above is for i386 or amd64. The nanobsd part can is either nanobsd for serial console, or nanobsd_vga for the VGA-enabled version.
Steve
-
So if I have a CMD prompt which I've seen where you pick WAN, LAN etc.
What is the difference between VGA and non-VGA versions specifically?
I thought CMD prompt was what VGA was?The VGA images are useful only for boards with a vga adapter. There are boards like most boards of the Alix series from PC Engines which do not have a VGA adapter. You cannot use them with a VGA image. You need the non-VGA images instead and you get your console via serial line.
Peter
-
ok cool. so with the jetway board I have would you say I have the correct image?
Sounds like it to me.
-
ok cool. so with the jetway board I have would you say I have the correct image?
Sounds like it to me.
Yes. I'm using a VGA image for my Jetway NF99FL-525 as well. However, I've decided to go with the corresponding AMD64 2G image:
pfSense-2.0.3-RELEASE-2g-amd64-nanobsd_vga.img.gz.Peter
-
Are you able to run 64-bit pfSense on Jetway NF99FL-525? On the 525 Atom CPU? I wasn't aware it's 64-bit capable.
-
ok cool. so with the jetway board I have would you say I have the correct image?
Sounds like it to me.
Yes. I'm using a VGA image for my Jetway NF99FL-525 as well. However, I've decided to go with the corresponding AMD64 2G image:
pfSense-2.0.3-RELEASE-2g-amd64-nanobsd_vga.img.gz.Peter
Is there a reason you went 64bit and are there any advantages?
Thanks
-
I'd be interested to hear your reasoning too because my advice would be to always use the 32bit image unless you really need 64bit. You can use more than 4GB of RAM but the Atom can't so no advantage there. There may be some performance advantage for specific tasks but it's so negligible you'd need some careful testing to see it. Mostly the 32bit image is better tested, especially on an Atom.
Steve
-
I'd be interested to hear your reasoning too because my advice would be to always use the 32bit image unless you really need 64bit. You can use more than 4GB of RAM but the Atom can't so no advantage there. There may be some performance advantage for specific tasks but it's so negligible you'd need some careful testing to see it. Mostly the 32bit image is better tested, especially on an Atom.
Steve
I've once had two reasons:
1.) My system has 4 GiB RAM.
2.) I wanted to test the AMD64 version.The AMD64 image runs very stable over months without any problems. It has turned out, however, that my pfSense installation newer uses >= 3.3 GiB. That's why I could have gone with the 32 bit image without any disadvantages :).
Peter
EDIT: Do you think there are significantly more i386 installations than AMD64? If so, I agree with you that the feedback from those users may make pfSense i386 more stable. Furthermore, many older Atoms were 32bit systems. Do you know about other reasons why the i386 image may be more stable?
-
The last time the question was asked I believe the figures showed more 32bit installs by some way. I imagine that more and more people are using 64bit though. I can't find it now. :-
Of course the more people who use 64bit the quicker an bugs will be found and squashed. ;)I don't know what the figures are for FreeBSD, would be interesting to find out.
Steve
-
At least for a few years, there will probably be more i386 installs. I'm not sure if we have a way to track that accurately though.
The reason i386 is still more common is because of embedded devices, i.e. ALIX and its cousins, and re-purposed old machines that aren't 64-bit.
As the hardware in that area catches up and becomes 64-bit capable, only then would I expect it to be more common.
That said, it is definitely picking up from what I've seen with customers. New server-grade hardware and VMs are almost always using amd64 now.
-
Got it all up and running…
Well almost...
I've got WAN and LAN working.
But when I set up opt1, opt2, opt3.
They don't seem to do anything.
I can't pull up the web panel.Do I need to set a firewall rule?
I did the set up via the CMD setup
-
Do I need to set a firewall rule?
Yes.
By default only the LAN interface has rules in place to allow access to anything. You will have to add appropriate rules to the additional interfaces.Steve
-
Another reason for i386 images are that not all motherboards are capable of housing over 4GB of RAM. Typical installs of 4GB is way more that sufficient for pfSense and some resource hungry packages.
When I first started using pfSense, Snort was the killer and would hog up almost 80% of the 4GB RAM. The package now has gone through several cycles of fine tuning and refinements. It barely takes 20% of my 4GB RAM. With Snort, Squid, Dans, pfBlocker and OpenVPN all combined my RAM usage hovers around 35% and steadily increases by 2% everyday. pfSense reloads the cache after some days and memory usage drops down. So not even 2GB of RAM is being used.
Hats off to the developers who have made such a fine UTM product.
-
All I want to do is set up each LAN port like any other simple setup for a home network.
So… when I go to Opt1 ENABLE...then set to STATIC, leaving all else default then I come to Static IP address.
Is this not the same as WAN which in my case is 192.168.1.2 ? for all LAN ports?
Or do they have to be set like 192.168.1.2, 192.168.1.3 etc? which does not make sense.
Actually subnet is showing as 192.168.1.0 so maybe that is right?Then I go to Firewall.. set to..
Pass.
Opt2
ANY
Destination ---> tick NOT then select "Any".
Place a description and SAVE?Then Services ---> DHCP Server...
Select Opt2.
Port range same as LAN ? 192.168.1.10 - 192.168.1.245
then SAVE?From the Googling I've done,,, does this look correct?
thanks
-
Hmm, a few problems there I think. ;)
All I want to do is set up each LAN port like any other simple setup for a home network.
Do you mean like a SOHO router with 4 LAN ports?
So… when I go to Opt1 ENABLE...then set to STATIC, leaving all else default then I come to Static IP address.
Is this not the same as WAN which in my case is 192.168.1.2 ? for all LAN ports?
Or do they have to be set like 192.168.1.2, 192.168.1.3 etc? which does not make sense.
Actually subnet is showing as 192.168.1.0 so maybe that is right?The usual way this would be set up is that each interface is a separate subnet. So for example you could use:
LAN is 192.168.1.1/24 (the default configuration)
OPT1 is 192.168.2.1/24
OPT2 is 192.168.3.1/24
OPT3 is 192.168.4.1/24If your WAN interface is using a private IP, like 192.168.1.2 as you say above, then you would have to choose something else because the WAN interface must use a different subnet.
Then I go to Firewall.. set to..
Pass.
Opt2
ANY
Destination –-> tick NOT then select "Any".
Place a description and SAVE?If you want to allow traffic from devices connected to OPT2 out to the internet or to other interfaces you need a rule more like:
Pass
OPT2
Protocol: any
Source: any
Destination: anyThis is a very permissive rule though.
A rule that has destination 'NOT any' will never match traffic. ;)Then Services –-> DHCP Server...
Select Opt2.
Port range same as LAN ? 192.168.1.10 - 192.168.1.245
then SAVE?The IP range would be different because OPT2 is not the same subnet as LAN. So for the above example it could be 192.168..3.10 - 192.168.3.254
That would leave 192.168.3.2 - 192.168.3.9 for any static IP assignments you wanted to use.Steve
-
Hi Steve,
I'm used to DD-WRT. So I guess allot of the settings are kept simple since all ports on a 4 port router
like the Asus N16 is done automatically.That is pretty much as simple as I wish to keep it.
So I'd go..
192.168.2.1
192.168.3.1rather than
192.168.1.2
192.168.1.3
?Thing is… with any commercial router no matter which of the 4 ports I connect to.
They are all given ips on the same like 192.168.1.*
not 192.168.*.2So all my devices on my Asus with DD-WRT was like 192.168.1.101, 192.168.1.102 etc
Hmm, a few problems there I think. ;)
All I want to do is set up each LAN port like any other simple setup for a home network.
Do you mean like a SOHO router with 4 LAN ports?
So… when I go to Opt1 ENABLE...then set to STATIC, leaving all else default then I come to Static IP address.
Is this not the same as WAN which in my case is 192.168.1.2 ? for all LAN ports?
Or do they have to be set like 192.168.1.2, 192.168.1.3 etc? which does not make sense.
Actually subnet is showing as 192.168.1.0 so maybe that is right?The usual way this would be set up is that each interface is a separate subnet. So for example you could use:
LAN is 192.168.1.1/24 (the default configuration)
OPT1 is 192.168.2.1/24
OPT2 is 192.168.3.1/24
OPT3 is 192.168.4.1/24If your WAN interface is using a private IP, like 192.168.1.2 as you say above, then you would have to choose something else because the WAN interface must use a different subnet.
Then I go to Firewall.. set to..
Pass.
Opt2
ANY
Destination –-> tick NOT then select "Any".
Place a description and SAVE?If you want to allow traffic from devices connected to OPT2 out to the internet or to other interfaces you need a rule more like:
Pass
OPT2
Protocol: any
Source: any
Destination: anyThis is a very permissive rule though.
A rule that has destination 'NOT any' will never match traffic. ;)Then Services –-> DHCP Server...
Select Opt2.
Port range same as LAN ? 192.168.1.10 - 192.168.1.245
then SAVE?The IP range would be different because OPT2 is not the same subnet as LAN. So for the above example it could be 192.168..3.10 - 192.168.3.254
That would leave 192.168.3.2 - 192.168.3.9 for any static IP assignments you wanted to use.Steve
-
I thought you might say something like that.
Most SOHO routers, such as the Asus N16, are in fact a two port router with a 4 port switch (5 if you count the internal one) on the same board.
With your setup you have 5 completely independent interfaces which allows you much better control over different network segments. Each interface can have different firewall rules. This is a far more powerful configuration but is also more complex to setup.
The problem is that if you are just substituting this for the N16 then you might have issues with devices not seeing each other.
It is possible to setup the interfaces to behave exactly like they would on the N16 by bridging them together.
http://doc.pfsense.org/index.php/Interface_Bridges
Due to the fact that traffic still has to be moved between each interface in software this will not be as fast as a real switch. Probably not what you wanted to hear. :-\Anyway I'm sure we can come to a configuration that meets your requirements. I think you said earlier that you wanted to remove as many devices as possible. What devices are you actually connecting?
Steve
-
Hi Steve,
Ok, so if I have different IPs hooked up to the LAN interfaces like.
192.168.1.2
192.168.2.2Etc (reason I'm using ..*.2 is because the dd wrt I would connect purely for wireless needs.
And that uses 192.168.1.1. I probably should change this to 192.168.6.2So what your saying is anything on 192.168.1.2 won't be able to communicate with anything on 192.168.2.2?
And so on without bridging them?In the house the following is what is connected.
2 full size pcs.
Netgear NAS
2 networked laser printers.
The 3 below are networked via home plugs…
1 plasma tv
1 Mini ITX home theatre pc.
Ps3Wifi.
Nintendo 3ds
2 laptops
iPad -
So what I think I'm going to do is…
Set up pfsense so it reflects...
192.168.2.1 wan (main)
Then do the rest 192.168.3.1 etcLeave the dd wrt 192.168.1.1
Sounds good thus far?
-
So what your saying is anything on 192.168.1.2 won't be able to communicate with anything on 192.168.2.2?
And so on without bridging them?In general they will be able to communicate, as long as you have allowed it with firewall rules. Just as your desktop PC can talk to a random web server in a completely different IP range, pfSense will route the traffic between it's different interfaces.
However some software will not work across subnets, specifically often media server/client programs. Things that use DLNA or equivalent protocols to automatically find servers often only look locally. Some clients do not even allow for manual entry of the server address (a massive oversight IMHO!). It is possible to extend the reach of some of these using the IGMP proxy between two interfaces but this is a bit hit-and-miss. If you're not using this feature you may not have any trouble.192.168.2.1 wan (main)
Then do the rest 192.168.3.1 etcDo you mean WAN here? What is your WAN connection? The fact that it's a private address implies you have some upstream router.
Steve
-
So what your saying is anything on 192.168.1.2 won't be able to communicate with anything on 192.168.2.2?
And so on without bridging them?In general they will be able to communicate, as long as you have allowed it with firewall rules. Just as your desktop PC can talk to a random web server in a completely different IP range, pfSense will route the traffic between it's different interfaces.
However some software will not work across subnets, specifically often media server/client programs. Things that use DLNA or equivalent protocols to automatically find servers often only look locally. Some clients do not even allow for manual entry of the server address (a massive oversight IMHO!). It is possible to extend the reach of some of these using the IGMP proxy between two interfaces but this is a bit hit-and-miss. If you're not using this feature you may not have any trouble.192.168.2.1 wan (main)
Then do the rest 192.168.3.1 etcDo you mean WAN here? What is your WAN connection? The fact that it's a private address implies you have some upstream router.
Steve
So what your saying is anything on 192.168.1.2 won't be able to communicate with anything on 192.168.2.2?
And so on without bridging them?In general they will be able to communicate, as long as you have allowed it with firewall rules. Just as your desktop PC can talk to a random web server in a completely different IP range, pfSense will route the traffic between it's different interfaces.
However some software will not work across subnets, specifically often media server/client programs. Things that use DLNA or equivalent protocols to automatically find servers often only look locally. Some clients do not even allow for manual entry of the server address (a massive oversight IMHO!). It is possible to extend the reach of some of these using the IGMP proxy between two interfaces but this is a bit hit-and-miss. If you're not using this feature you may not have any trouble.192.168.2.1 wan (main)
Then do the rest 192.168.3.1 etcDo you mean WAN here? What is your WAN connection? The fact that it's a private address implies you have some upstream router.
Steve
At the moment I have WAN: 192.168.1.0 (from what I can see)
LAN1: 192.168.1.1
LAN2: 192.168.2.1
LAN3: 192.168.3.1
LAN4: 192.168.4.1I've added the same firewall rule that pfsense automatically added to LAN1 and copied to all other LANs after I enabled them.
I've also in DHCP Server,
LAN1: 192.168.1.10 - 192.168.1.245
LAN2: 192.168.2.10 - 192.168.2.245
LAN3: 192.168.3.10 - 192.168.3.245
LAN4: 192.168.4.10 - 192.168.4.245And changed nothing else. Which is the same as what was already set for LAN1 from what I can see.
I can access the internet from all 4 LAN ports now.
But as you'd said… If I use a switch and hook everything up I'm having success them all seeing each other at the moment.
But when I connected the NAS drive to another LAN port i.e LAN2 on pfsense. The NAS drive was not accessible from windows
even after rebooting the NAS.So Windows was 192.168.1.1 range and NAS was 192.168.2.1 range.
How is this fixable?
-
If you have firewall rules on each interface that are the same as the default LAN rule then you should be able to access anything from any interface. From a client on LAN1 you could access the NAS box on LAN2 directly by entering it's IP address. You can also access it by name if you have the right options selected in DNS Forwarder depending if the NAS has a static lease.
When you say you are not 'seeing' the NAS drive what do you mean? It doesn't magically appear in Windows Explorer?
You can always bridge those two interfaces such that they will share a single subnet.
Steve
-
If you have firewall rules on each interface that are the same as the default LAN rule then you should be able to access anything from any interface. From a client on LAN1 you could access the NAS box on LAN2 directly by entering it's IP address. You can also access it by name if you have the right options selected in DNS Forwarder depending if the NAS has a static lease.
When you say you are not 'seeing' the NAS drive what do you mean? It doesn't magically appear in Windows Explorer?
You can always bridge those two interfaces such that they will share a single subnet.
Steve
Hi Steve,
thanks for your reply :)
Yes, before it was listed as a networked drive as nas-0A-70-F1: ReadyNAS Duo, and I was able to locate it under "network".
I'm liking the new way of doing things thus far. I actually think it's much better than the commercial routers.
I even got OpenVPN working with HideMyAss.. and the results was great with speeds.
I just now need to figure IPVanish's settings.As for the the nas-0A-70-F1: ReadyNAS Duo not showing.
It would be great to know how to bridge them so I can do that as that is probably the one thing I'd want to change at the moment.thanks