Re: network problems with kernel-2.6.17-1.2139_FC5

[Roy Erickson <erickson@xxxxxxxxxxxxxxxx>]

On Jun 27, 2006, at 2:20 PM, George Avrunin wrote:

I've just updated to the 2139 kernel and I'm encountering some network

problems.  Connections inside my department seem fine but connections

to anywhere outside (other departments on campus, the rest of the

world) lock up after some fairly small amount of traffic.  The

problems disappear if I reboot back into the 2133 kernel.

The symptoms are very similar to some that I recall from a couple of

years ago with the move to the 2.6.7 kernel where some of the

networking parameters changed.  At that time, there were problems with

various firewalls/routers and the solution was to do

   sysctl -w net.ipv4.tcp_default_win_scale=0

   sysctl -w net.ipv4.tcp_moderate_rcvbuf=0

(or the equivalent in /etc/sysctl.conf).

tcp_default_win_scale doesn't seem to exist anymore, but setting

tcp_window_scaling to 0 seems to work, at least as far as I've been

able to check quickly.  But I don't really understand what these

parameters represent and don't want to mess things up.  

Can the gurus tell me what, in words of one syllable, what changed in

this respect from the 2133 kernel to the 2139 and what the right fix

is?  What am I losing if I set tcp_window_scaling to 0?  I have some

influence on the settings of my department's network equipment, but

the connection to the campus backbone is controlled by the Office of

Information Technology, which generally doesn't respond well to

requests from Linux users in the science and math departments.  So if

it's a question of me changing things on my Fedora boxes or OIT

changing settings on its network equipment, I'll be the one making the

changes.

Try http://kerneltrap.org/node/6723

From: John Heffner [email blocked]

Subject: Re: 2.6.17: networking bug??

Date: Tue, 13 Jun 2006 13:39:42 -0400

Mark Lord wrote:

> John / David: Any ideas on what's gone awry here?

>

>

Yes, you have some sort of a broken middlebox in your path (firewall,

transparent proxy, or similar) that doesn't correctly handle window

scaling. Check out this thread:

<http://marc.theaimsgroup.com/?l=linux-netdev&m=114478312100641&w=2&gt;.

The best thing you can do is try to find this broken box and inform its

owner that it needs to be fixed. (If you can find out what it is, I'd

be interested to know.) In the meantime, disabling window scaling will

work around the problem for you.

-John

From: Linus Torvalds [email blocked]

Subject: Re: 2.6.17: networking bug??

Date: Tue, 13 Jun 2006 10:50:51 -0700 (PDT)

On Tue, 13 Jun 2006, John Heffner wrote:

>

> The best thing you can do is try to find this broken box and inform its owner

> that it needs to be fixed. (If you can find out what it is, I'd be interested

> to know.) In the meantime, disabling window scaling will work around the

> problem for you.

Well, arguably, we shouldn't necessarily have defaults that use window

scaling, or we should have ways to recognize automatically when it

doesn't work (which may not be possible).

It's not like there aren't broken boxes out there, and it might be better

to make the default buffer sizes just be low enough that window scaling

simply isn't an issue.

I suspect that the people who really want/need window scaling know about

it, and could be assumed to know enough to raise their limits, no?

Linus

Thanks,

  George