browser won't connect after kernel 2.6.4 installation

I upgraded my kernel to 2.6.4 from 2.4.22 in Fedora Core 1, and I lost my ability to browse the internet. The browser just sits there...never displays a page, but never complains that it can't find the address. I can't seem to telnet out of the box either.

Strangely enough, eth0 starts fine. I can ping www.google.com, and it returns normal results. Traceroute also behaves normally.

I've got a Netgear MR814v2 router that is serving as firewall/DSL connection. I can ping it, and I can access the admin webpage on that.

There are DNS servers assigned, and I can ping the DNS servers.

I've also turned off IPv6 by typing the following into modprobe.conf:
alias net-pf-10 off

dmesg shows nothing unusual. It claims eth0 started fine.

lsmod shows the module 8139too has been loaded.



Beyong that....I'm a bit lost as to what I need to investigate to get this up and running....


Thanks.
(Posting this from a separate computer, otherwise i'd include the relevant dmesg and lsmod outputs.)

hmm that is freaky.

are you absolutely sure you do not have a firewall running.
here is an explanation.

you open your browser, your browser sends a request to a computer on the network most likely port 80. the response can be any number of ports in the numbers of:
80
81
82
1080
8080
8088

one of these ports is being blocked if you can ping www.google.com but cannot access it from a browser (Port 80, 8080, 1080)

by a firewall I mean your linux machines firewall service.
if you are using redhat type 'redhat-config-network'
switch to no firewall and try the browser, you may have to restart the computer.

The firewall settings should be checked, but the explanation as to why is not quite correct. When you connect to a webserver you use port 80 unless you (or the link you click) explicitly tells you to use a different port. You would do that by typing something like http://somerandomsite.com:8080 . Then you would connect to port 8080 on somerandomsite.com. When you ping you are not using tcp so that means your tcp ports are irrelevent to ping. You use ICMP when you ping.

Try the firewall suggestion above. Also at the command line type

telnet www.google.com 80

. See what that results in. You will probably get something like this:
Trying 216.239.57.104...
Connected to www.google.com.
Escape character is '^]'.

or you will get connection refused. The first one is good. That means you connected to the google webserver over port 80.

One more suggestion after re-reading your post....
Have you completely unplugged the electricity from your DSL modem and your router? Try that and plug them back in after several seconds. I've seen that work in some cases.

No. No firewall is running. I've disabled it to make sure.


" Telnet www.google.com 80 " results in:
Trying 216.239.51.99....

and just hangs there.

And yeah, I've reset the DSL router....even upgraded the firmware!

Ok. I can SSH into the linux box from my Windows laptop (gotta love OpenSSH).

Here's my output from "route":

Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.0.0 * 255.255.255.0 U 0 0 0 eth0
169.254.0.0 * 255.255.0.0 U 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default 192.168.0.1 0.0.0.0 UG 0 0 0 eth0


I find the "169.254.0.0" route in there a bit odd...but I'm not familiar enough with routing tables to know what to do with that.

So the IP address of your router is 192.168.0.1 , correct? Have you gone through the setting of your router to see if all looks good in there? Could you post the output of ifconfig as well?

Yeah. The router is fine. I'm connected thru it as we speak, and my old 2.4.22 kernel on the linux box can access the outside world. Just not the new 2.6.4 kernel.


ifconfig:

eth0 Link encap:Ethernet HWaddr 00:30:1B:AF:44:23
inet addr:192.168.0.2 Bcast:192.168.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:920 errors:0 dropped:0 overruns:0 frame:0
TX packets:853 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:95872 (93.6 Kb) TX bytes:85302 (83.3 Kb)
Interrupt:185 Base address:0xb000

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)

Is that output you gave me what's going on while the 2.6 kernel is running? If not then ouput that plus iptables -L.

This is an odd problem. DNS seems to be working fine. You can ping locations. Have you gone through your kernel config to make sure you didn't leave anything out of the networking section?

Yeah, that output from the 2.6.4 kernel. I'm at work right now, but I'll post the output of " iptables -l " tonight... BUT, I did an "iptables --flush" or "iptables -f" (can't recall the exact syntax now...) to make sure it was empty and not blocking anything.

I walked thru the "make xconfig" step-by-step last night, looking for some oddity. I'm currently running a custom-built kernel, but based off the .config from the RPMs via people.redhat.com/arjanv. But, the kernel straight out of the arjanv repositories (I've tried 2.6.2, 2.6.3, and 2.6.4 RPMs) have all had the same problem: can ping, but can't browse.

Is there any tool out there that I can use to dive into where the block is? Will nmap help? Any idea of what logs I should be looking at?

I'm pretty stumped as well.

You can type /etc/init.d/iptables stop and that will end all iptables. I'm not really convinced that is the problem however. If I were you I would ditch the old .config file and do it myself from scratch. Download the kernel from kernel.org. I downloaded that one, did the config myself and have no problems with it.

SUCCESS!!!!

As I was going thru make xconfig, I found the following blurb:

IP: TCP Explicit Congestion Notification support (INET_ECN)

Explicit Congestion Notification (ECN) allows routers to notify
clients about network congestion, resulting in fewer dropped packets
and increased network performance. This option adds ECN support to
the Linux kernel, as well as a sysctl (/proc/sys/net/ipv4/tcp_ecn)
which allows ECN support to be disabled at runtime.

Note that, on the Internet, there are many broken firewalls which
refuse connections from ECN-enabled machines, and it may be a while
before these firewalls are fixed. Until then, to access a site
behind such a firewall (some of which are major sites, at the time
of this writing) you will have to disable this option, either by
saying N now or by using the sysctl.

If in doubt, say N.


SO, I popped into a shell, and typed:

echo 0 > /proc/sys/net/ipv4/tcp_ecn

And BINGO!!!! I am now posting this from my 2.6.4 Kernel-running-Fedora box.

Thanks again for all your help, benji. Your suggestion to dive into the .config and do it myself was indeed the ticket.

Sweet! Glad it worked out.

Originally posted by wintermute222 SUCCESS!!!! As I was going thru make xconfig, I found the following blurb: IP: TCP Explicit Congestion Notification support (INET_ECN) Explicit Congestion Notification (ECN) allows routers to notify clients about network congestion, resulting in fewer dropped packets and increased network performance. This option adds ECN support to the Linux kernel, as well as a sysctl (/proc/sys/net/ipv4/tcp_ecn) which allows ECN support to be disabled at runtime. Note that, on the Internet, there are many broken firewalls which refuse connections from ECN-enabled machines, and it may be a while before these firewalls are fixed. Until then, to access a site behind such a firewall (some of which are major sites, at the time of this writing) you will have to disable this option, either by saying N now or by using the sysctl. If in doubt, say N. SO, I popped into a shell, and typed: echo 0 > /proc/sys/net/ipv4/tcp_ecn And BINGO!!!! I am now posting this from my 2.6.4 Kernel-running-Fedora box. Thanks again for all your help, benji. Your suggestion to dive into the .config and do it myself was indeed the ticket.

I have the same problem. My /proc/sys/net/ipv4/tcp_ecn is 0 originally in the 2.6.6 kernel installed (debian sid).
As many posts suggested, I turned off the ipv6 in modprobe.conf, however it is still no joy...
Some suggest me to change DNS server but I just do not know a "good" one.
I haven't use my linux install for about 4 months because of this!!!
I suspected my router (mr814v2), but here yours works well with it.

A tcpdump indicates that my browser attempts to send out ipv6 request and it takes 2 min to time out...
If I disable ipv6 in firefox, it works well, but this does not prevent other applications from querying ipv6 domain name.

How can I solve this? I still want to work on Linux... Help!!