Re: [pfSense-discussion] acx100 and 1.2 beta

2007-08-07 Thread Paul M
Marius Schrecker wrote:
> Hi,
> 
>   I'm currently running 1.0.1 (developer) with the acx100 native driver
> from kewl.org which I compiled using the recommended patch.  Works okay,
> but I remember it being quite a bit of work.
> 
> What's the status on this driver in 1.2? Will it be built-in, or easier to
> compile, or is there a procedure for using ndiswrapper for this.


I guess the only way to find out is to try the live CD version!



[pfSense-discussion] acx100 and 1.2 beta

2007-08-07 Thread Marius Schrecker
Hi,

  I'm currently running 1.0.1 (developer) with the acx100 native driver
from kewl.org which I compiled using the recommended patch.  Works okay,
but I remember it being quite a bit of work.

What's the status on this driver in 1.2? Will it be built-in, or easier to
compile, or is there a procedure for using ndiswrapper for this.

My thumbs are itching to try 1.2, and this is the only thing stopping me :-)

Cheers!

Marius


Re: [pfSense-discussion] jumbo frames

2007-08-07 Thread Nick Buraglio
On 8/7/07, Eugen Leitl <[EMAIL PROTECTED]> wrote:
>
> On Tue, Aug 07, 2007 at 01:00:55PM +, Nick Buraglio wrote:
> > Is there a reason you're running jumbo frames?  If you're going to
>
> I need more performance on NFS and RDP (assuming, RDP can make
> use of jumbo frames -- I'm not sure, and the the current
> firmware of the switch doesn't show the jumbo diagnostics).



I just had to ask.  I've seen many people "need" it but then not really have
a reason and since many vendors do it differently it can become a pain.
I will likely help with large data transfers using NFS.


> enable it you should run it on all interfaces that touch the segment.
> > It will likely work on lower mtu devices but it is certainly not
> > beat practice to mix and match.  If you are running it internally
> > to gain performance then you may see some improvements depending
> > on what protocol you are using to transfer data (assuming at
>
> Yes, that's the reason. Of course it's a GBit Ethernet network.
> There's very little point with jumbo frames on 100 MBit Ethernet,
> should it even be supported.


Again, had to ask.   Large MTU support is sometimes used where it really
isn't the best solution so I wanted to make sure there wasn't a better
solution to the issue you were trying to address.

> least gig connectivity) but unless you're ISP supports it (and
> > you have like FTTC) you won't see any upstream gain.
>
> The ISP is a vanilla ADSL (PPPoE), so I can't up the MTU there.


It really should be enabled on the entire segment, you may or may not see
weirdness and hard to diagnose problems if you don't.


--
> Eugen* Leitl http://leitl.org";>leitl http://leitl.org
> __
> ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
> 8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
>


Re: [pfSense-discussion] jumbo frames

2007-08-07 Thread Adam Armstrong

Greg Hennessy wrote:

I've just switched to jumbo frames on the home network (enabled
jumbo frames (mtu 9014) on NIC and one switch). I'm running a recentish
(1.2-BETA-1-TESTING-SNAPSHOT-05-11-2007) pfsense on WRAP, with
mtu 1500 there (I don't think WRAP NICs can do jumbo frames).

Should I run into problems? Higher latency, higher CPU load,
more memory? Don't see anything, but the box is not exactly
loaded.



If what I am reading above is correct, you've enabled it on a workstation
and on the switch but not on the firewall ? 
Everything on the same collision domain ? 
  
You mean broadcast domain. A switch/vlan is a broadcast domain, an 
ethernet segment is a collision domain (i.e. a hub or thinnet).
If so, you will have problems. 

Mix and match frame size on the same VLAN is a recipe for grief. 
  

adam.


Re: [pfSense-discussion] jumbo frames

2007-08-07 Thread Eugen Leitl
On Tue, Aug 07, 2007 at 01:00:55PM +, Nick Buraglio wrote:
> Is there a reason you're running jumbo frames?  If you're going to

I need more performance on NFS and RDP (assuming, RDP can make
use of jumbo frames -- I'm not sure, and the the current 
firmware of the switch doesn't show the jumbo diagnostics).
 
> enable it you should run it on all interfaces that touch the segment.  
> It will likely work on lower mtu devices but it is certainly not 
> beat practice to mix and match.  If you are running it internally 
> to gain performance then you may see some improvements depending 
> on what protocol you are using to transfer data (assuming at 

Yes, that's the reason. Of course it's a GBit Ethernet network.
There's very little point with jumbo frames on 100 MBit Ethernet,
should it even be supported.

> least gig connectivity) but unless you're ISP supports it (and 
> you have like FTTC) you won't see any upstream gain.

The ISP is a vanilla ADSL (PPPoE), so I can't up the MTU there.

-- 
Eugen* Leitl http://leitl.org";>leitl http://leitl.org
__
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE


Re: [pfSense-discussion] jumbo frames

2007-08-07 Thread Nick Buraglio
Is there a reason you're running jumbo frames?  If you're going to enable it 
you should run it on all interfaces that touch the segment.  It will likely 
work on lower mtu devices but it is certainly not beat practice to mix and 
match.  If you are running it internally to gain performance then you may see 
some improvements depending on what protocol you are using to transfer data 
(assuming at least gig connectivity) but unless you're ISP supports it (and you 
have like FTTC) you won't see any upstream gain.





nb

-Original Message-
From: "jason whitt" <[EMAIL PROTECTED]>

Date: Mon, 6 Aug 2007 20:19:35 
To:discussion@pfsense.com
Subject: Re: [pfSense-discussion] jumbo frames

On your home router it shouldn't matter  whered you get a gig
connection ran to your house?


On 8/6/07, Eugen Leitl <[EMAIL PROTECTED]> wrote:
>
> I've just switched to jumbo frames on the home network (enabled
> jumbo frames (mtu 9014) on NIC and one switch). I'm running a recentish
> (1.2-BETA-1-TESTING-SNAPSHOT-05-11-2007) pfsense on WRAP, with
> mtu 1500 there (I don't think WRAP NICs can do jumbo frames).
>
> Should I run into problems? Higher latency, higher CPU load,
> more memory? Don't see anything, but the box is not exactly
> loaded.
>
> --
> Eugen* Leitl http://leitl.org";>leitl http://leitl.org
> __
> ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
> 8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
>


Re: [pfSense-discussion] jumbo frames

2007-08-07 Thread Eugen Leitl
On Tue, Aug 07, 2007 at 08:59:30AM +0100, Greg Hennessy wrote:

> If what I am reading above is correct, you've enabled it on a workstation
> and on the switch but not on the firewall ? 

IIRC WRAP NICs don't have jumbo fram support. It seems to be working,
it must be fragmenting the oversized MTUs internally when
accessing WAN.

> Everything on the same collision domain ? 

On that WRAP box, thre's one WAN and one LAN NIC,
the third NIC is currently unused.
 
> If so, you will have problems. 
> 
> Mix and match frame size on the same VLAN is a recipe for grief. 

I don't use a VLAN on that setup yet. 

-- 
Eugen* Leitl http://leitl.org";>leitl http://leitl.org
__
ICBM: 48.07100, 11.36820 http://www.ativel.com http://postbiota.org
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE


RE: [pfSense-discussion] jumbo frames

2007-08-07 Thread Greg Hennessy

> 
> I've just switched to jumbo frames on the home network (enabled
> jumbo frames (mtu 9014) on NIC and one switch). I'm running a recentish
> (1.2-BETA-1-TESTING-SNAPSHOT-05-11-2007) pfsense on WRAP, with
> mtu 1500 there (I don't think WRAP NICs can do jumbo frames).
> 
> Should I run into problems? Higher latency, higher CPU load,
> more memory? Don't see anything, but the box is not exactly
> loaded.

If what I am reading above is correct, you've enabled it on a workstation
and on the switch but not on the firewall ? 
Everything on the same collision domain ? 

If so, you will have problems. 

Mix and match frame size on the same VLAN is a recipe for grief. 


Greg