Can't get internet working in Linux

Hello all,

I have a dual boot computer with Fedora Core 1 and
Windows 2000. I have internet access under Windows
but under Linux I can't get internet working. The
ethernet card is working though (8139too.o)

I have tried all possible things as far as my limited
knowledge permits but still no result. What I found
from my efforts were the following:

o It can ping itself, and others
directly connected to it
o It can ping the default gateway
o The resolv.conf file contents are valid
o It cannot ping any of the nameservers
o Traceroute shows that (apparently) it gets stuck
at the gateway (or maybe the next hop)
o The same configuration works flawlessly under
Windows 2000.

(It might seem that the routing table is
not setup correctly. I thought so, but could not
find any flaw with it. Tomorrow I might me able
to post the routing table.)

Could you please suggest whether I should try
anything else. (Next day, when I am at that computer
again, I might me able to send you the results I got
from the various commands that made me draw those
conclusions.)

Regards,
Santanu

santanu wrote:
> Hello all,
>
> I have a dual boot computer with Fedora Core 1 and
> Windows 2000. I have internet access under Windows
> but under Linux I can't get internet working. The
> ethernet card is working though (8139too.o)
>
> I have tried all possible things as far as my limited
> knowledge permits but still no result. What I found
> from my efforts were the following:
>
> o It can ping itself, and others
> directly connected to it
> o It can ping the default gateway
> o The resolv.conf file contents are valid
> o It cannot ping any of the nameservers
> o Traceroute shows that (apparently) it gets stuck
> at the gateway (or maybe the next hop)
> o The same configuration works flawlessly under
> Windows 2000.
>
> (It might seem that the routing table is
> not setup correctly. I thought so, but could not
> find any flaw with it. Tomorrow I might me able
> to post the routing table.)
>
> Could you please suggest whether I should try
> anything else. (Next day, when I am at that computer
> again, I might me able to send you the results I got
> from the various commands that made me draw those
> conclusions.)
>
> Regards,
> Santanu
>
What does natstst -r show?

santanu wrote:
> Hello all,
>
> I have a dual boot computer with Fedora Core 1 and
> Windows 2000. I have internet access under Windows
> but under Linux I can't get internet working. The
> ethernet card is working though (8139too.o)
>
> I have tried all possible things as far as my limited
> knowledge permits but still no result. What I found
> from my efforts were the following:
>
> o It can ping itself, and others
> directly connected to it
> o It can ping the default gateway
> o The resolv.conf file contents are valid
> o It cannot ping any of the nameservers
> o Traceroute shows that (apparently) it gets stuck
> at the gateway (or maybe the next hop)

What is the gateway? Is it a DSL modem or some such? Is it using DHCP
for Windows? Are you setting up the network in linux in the same way?

>From what you describe my assumption would be that the gateway is not
routing. The fact that it works in windows indicates possible firewall
or routing issues that are being solved by the windows settings. Of
course I am not there and have no program output to go by so this is
assumption only.

Hi,

The following are the results of the various commands I tried:

==============================
[root@techno santanu]# /sbin/route -n
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use
Iface
192.168.10.0 0.0.0.0 255.255.255.0 U 0 0 0
eth0
169.254.0.0 0.0.0.0 255.255.0.0 U 0 0 0
eth0
127.0.0.0 0.0.0.0 255.0.0.0 U 0 0 0
lo
0.0.0.0 192.168.10.9 0.0.0.0 UG 0 0 0
eth0

[root@techno santanu]# /sbin/ifconfig
eth0 Link encap:Ethernet HWaddr 00:E0:18:65:92:8F
inet addr:192.168.10.160 Bcast:192.168.10.255
Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:161 errors:0 dropped:0 overruns:0 frame:0
TX packets:20 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:23711 (23.1 Kb) TX bytes:1556 (1.5 Kb)
Interrupt:9 Base address:0x6000

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:72 errors:0 dropped:0 overruns:0 frame:0
TX packets:72 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:3724 (3.6 Kb) TX bytes:3724 (3.6 Kb)

[root@techno santanu]# cat /etc/resolv.conf
nameserver 202.9.145.6
nameserver 202.54.9.1

[root@techno santanu]# ping www.google.com
PING www.google.akadns.net (216.239.57.104) 56(84) bytes of data.


--- www.google.akadns.net ping statistics ---
5 packets transmitted, 0 received, 100% packet loss, time 4000ms



[root@techno santanu]# traceroute 202.9.145.6
traceroute to 202.9.145.6 (202.9.145.6), 30 hops max, 38 byte packets
1 192.168.10.9 (192.168.10.9) 0.842 ms 0.903 ms 0.793 ms
2 static61.11.71-1.dsl-cal.eth.net (61.11.71.1) 37.180 ms 37.484 ms
117.395 ms
3 * * *
4 * * *
5 * * *
6 * * *
7 * * *
8 * * *
9 * * *
10 * * *
11 * * *
12 * * *
13 * * *
14 * * *
15 * * *

=============================

(sorry if the lines wrap)

I have also disabled the firewall settings for now, but still
no effect (no internet, i.e.)

Regards,
Santanu

> [root@techno santanu]# /sbin/route -n
> Kernel IP routing table
> Destination Gateway Genmask Flags Metric Ref
Use
> Iface
> 192.168.10.0 0.0.0.0 255.255.255.0 U 0 0
0
> eth0
> 169.254.0.0 0.0.0.0 255.255.0.0 U 0 0
0
> eth0

That doesn't look right. If eth0 is on two networks you should be
using aliasing.

> [root@techno santanu]# /sbin/ifconfig
> eth0 Link encap:Ethernet HWaddr 00:E0:18:65:92:8F
> inet addr:192.168.10.160 Bcast:192.168.10.255
> Mask:255.255.255.0

So it is not on the 169.254.0.0/16 network. Where did that route come
from?

> traceroute to 202.9.145.6 (202.9.145.6), 30 hops max, 38 byte packets
> 1 192.168.10.9 (192.168.10.9) 0.842 ms 0.903 ms 0.793 ms
> 2 static61.11.71-1.dsl-cal.eth.net (61.11.71.1) 37.180 ms 37.484
ms
> 117.395 ms

What lives at 192.168.10.9? Is it doing NAT? I bet not.

Where did 169.254.0.0 come from? You have no ip on that network, why
are you routing like you live on it?

There are some ideas...

Noah Roberts wrote:

> Where did 169.254.0.0 come from? You have no ip on that network, why
> are you routing like you live on it?

Here is a guess...

the 169.254.0.0 route comes from your ISP. Something is going wrong
between DHCP and static settings to your card maybe. The router at the
ISP is dropping your 192.168.10.x traffic like it should, it is
expecting traffic from a 169.254.0.0 address. The object living at
192.168.10.9 is just passing information through.

There are problems with this theory though. Since I am not there I
can't know exactly what is going on and the object at 192.168.10.9
obviously got that address from somewhere and is responding to it.
What is this object?

Go into windows and do an 'ipconfig /all' to discover what it's
settings are when it is communicating with the internet. See how they
differ from your Linux settings.

In article <(E-Mail Removed) .com>,
santanu wrote:
>o It can ping itself, and others
> directly connected to it

Pinging yourself uses the loopback, and only means that the network
card could be configured. It doesn't use the network at all.

>o It can ping the default gateway
>o The resolv.conf file contents are valid

Below - this is OK

>o It cannot ping any of the nameservers

Many people disable/block ping on their servers. Ten years ago it was a
useful tool. Now it is not.

>o Traceroute shows that (apparently) it gets stuck
> at the gateway (or maybe the next hop)

Below

>o The same configuration works flawlessly under
> Windows 2000.

except that it's not the same configuration. It's a completely different
network stack, with different options and defaults.

-------------------------

>[root@techno santanu]# /sbin/route -n

Looks fine - the 169.254.0.0 network probably isn't needed, but it's not
hurting anything.

>[root@techno santanu]# /sbin/ifconfig

Looks OK - no errors. Only thing that MIGHT be of concern is that
the interface is set for an MTU of 1500 (correct for Ethernet, but
may be a problem with pppoe). You might try setting it to 1490 and see
if that makes a difference.

>[root@techno santanu]# cat /etc/resolv.conf
>nameserver 202.9.145.6
>nameserver 202.54.9.1

The first one resolves to pdns01.minnambalam.com and appears to be OK.
The second resolves to giascl01.vsnl.net.in, and I get referred to
dns.vsnl.net.in at 202.54.1.30. It may or may not be valid.

>[root@techno santanu]# ping www.google.com
>PING www.google.akadns.net (216.239.57.104) 56(84) bytes of data.

A BIG clue here - notice that the name has been translated to a CNAME,
and that has been resolved to an IP address. Your connection is working
fine at least as far as one of the nameservers.

>[root@techno santanu]# traceroute 202.9.145.6
>traceroute to 202.9.145.6 (202.9.145.6), 30 hops max, 38 byte packets
> 1 192.168.10.9 (192.168.10.9) 0.842 ms 0.903 ms 0.793 ms
> 2 static61.11.71-1.dsl-cal.eth.net (61.11.71.1) 37.180 ms 37.484 ms
> 117.395 ms

Same thing here - notice that the IP address 61.11.71.1 has been resolved
back to 'static61.11.71-1.dsl-cal.eth.net'. No problem here.

> 3 * * *

Whoever is hop 3 is misconfigured, and doesn't feel it necessary to provide
ICMP connectivity (there should be an ICMP type 11 here). My guess is that
they are dropping ALL ICMP because they're sure it's good security. They
should be shot.

>I have also disabled the firewall settings for now, but still
>no effect (no internet, i.e.)

It doesn't look as if that is the problem.

Two most likely problems:

1. Make sure Explicit Congestion Notification is off. Some routers drop all
packets with header bits set that they don't understand. The syntax is

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

2. _IF_ you are on a PPPOE type network, try setting the MTU back slightly.
See RFC1435 and 2923. As root, this can be done with the /sbin/ifconfig
command.

Old guy

In article <(E-Mail Removed) .com>,
Noah Roberts wrote:

>> 169.254.0.0 0.0.0.0 255.255.0.0 U 0 0 0
>> eth0
>
>That doesn't look right. If eth0 is on two networks you should be
>using aliasing.

Actually, it's just a route - very strange, but legal after a fashion.

>So it is not on the 169.254.0.0/16 network. Where did that route come
>from?

He's running Fedora, and it has this wonderful windoze service called
'link local'. If you look inside /etc/sysconfig/network-scripts/ifup
on Fedora (and RH9), you'll find:

# Add Zeroconf route.
if [ -z "${NOZEROCONF}" -a "${ISALIAS}" = "no" ]; then
ip route replace 169.254.0.0/16 dev ${REALDEVICE}
fi

The solution is to set the variable 'NOZEROCONF' in the ifcfg-eth? file
in /etc/sysconfig/network-scripts/.

This is a concept that microsoft invented for win98, where if the idiot
who sets up the DHCP server has it so fscked up that even a windoze box
can't get assigned a dynamic IP, the host will grab an IP address out
of it's ass and use that. Lessee...

http://www.ietf.org/internet-drafts/...nklocal-17.txt

That is the seventeenth attempt to come up with an RFC, but no one seems
to be buying. It should have expired in early January (drafts are only
valid for six months).

Several Linux distributions (I've seen it with Red Hat 9, Fedora, and
late releases of SuSE at least) have added a route for this network, though
I'm not at all sure why, as the only thing it will do is to stop the
kernel from screaming about 'Martian packets' should some windoze box
come up in that mode on this network.

Bottom line: it has no effect what-so-ever on this problem.

=====================

>Here is a guess...
>
>the 169.254.0.0 route comes from your ISP. Something is going wrong
>between DHCP and static settings to your card maybe. The router at the
>ISP is dropping your 192.168.10.x traffic like it should, it is
>expecting traffic from a 169.254.0.0 address. The object living at
>192.168.10.9 is just passing information through.

Nope. If you read my reply to the OP, you'll notice he has no trouble
resolving hostnames - his connection to the DNS servers (at least) is
working just fine. More likely, it's ECN, coupled with an ISP run by
total incompetents.

Old guy

Noah Roberts wrote:
.....
> > 169.254.0.0 0.0.0.0 255.255.0.0 U 0 0
> 0
> > eth0
>
> That doesn't look right. If eth0 is on two networks you should be
> using aliasing.

> > [root@techno santanu]# /sbin/ifconfig
> > eth0 Link encap:Ethernet HWaddr 00:E0:18:65:92:8F
> > inet addr:192.168.10.160 Bcast:192.168.10.255
> > Mask:255.255.255.0
>
> So it is not on the 169.254.0.0/16 network. Where did that route
come
> from?

I have no idea where that line is coming from. Nowhere in
/etc/sysconfig
can I find that ip address. But still it comes up whenever I do
'/etc/init.d/network start'

>
> > traceroute to 202.9.145.6 (202.9.145.6), 30 hops max, 38 byte
packets
> > 1 192.168.10.9 (192.168.10.9) 0.842 ms 0.903 ms 0.793 ms
> > 2 static61.11.71-1.dsl-cal.eth.net (61.11.71.1) 37.180 ms
37.484
> ms
> > 117.395 ms
>
> What lives at 192.168.10.9? Is it doing NAT? I bet not.

Actually, the entity 192.168.10.9 is somoething at the ISP's end.
How do I find out whether it is doing NAT ? All I can say is that
it is not allowing any Linux machines to get through to the Internet,
only alowing Windows machines!!

>
> Where did 169.254.0.0 come from? You have no ip on that network, why
> are you routing like you live on it?
>
> There are some ideas...

I would be very happy if only I could remove it without manually
using 'route del'.

Regards,
Santanu

Noah Roberts wrote:

> Go into windows and do an 'ipconfig /all' to discover what it's
> settings are when it is communicating with the internet. See how
they
> differ from your Linux settings.

Will do and report the results on Monday, when I am at that
computer again.

Regards,
Santanu