strange connection problems

Ask questions regarding Gentoox, Gentoo and Linux in general in these forums and we'll do our best to help you!
Nightwing
Newbie
Posts: 6
Joined: Sun Feb 25, 2007 5:41 am
Location: Kelowna, BC, Canada
Contact:

REALLY SOLVED NOW! LOL!

Post by Nightwing »

OKay, again, really trully finally solved for good now! Said screw it, and bouhgt a second IP from my ISP and took a 3rd Linksys and wired it for uplink mode and wolla, VoIP router has it's own ip now, good for it cause I don't care for it anyways, and the good working linky router has it's own, and so yeah, everything's perfect now! I still learnt alot from this thread though and still recommend it to anyone having network problems.

If not for GentooX, I'd still be feeling alien to Gentoo in general, being so use to Debian/Red-Hat based distro's myself. Thanks again!

--Nightwing
<deadmatrix@gmail.com>
MisterB
Newbie
Posts: 3
Joined: Wed Feb 28, 2007 12:42 pm

Post by MisterB »

I have similar strange problems and I think it does have something to do with Gentoox. It seems that every time this problem pops up in the forums people solve it by either buying new hardware or rewiring... Something I still hope to avoid.

My setup is also a xbox connected behind a USR router 8054 (setup as a simple switch) which is in turn connected to the ADSL router/modem. Everything else connected to the USR works fine (windows pc, linux pc and laptops). The weird part is, even the xbox works perfectly when I boot into XBMC (tested with weather info and RSS feeds).

What is even stranger is that pinging the outside world works fine and dns resolution works as well (i.e. I can ping google.com), but any other protocol fails when it's more than two hops. This means I can access the webinterface of my the USR router but not the webinterface of the ADSL router (or any other external webpage for that matter). I'm completely at a loss here. It seems only ICMP and UDP packets go through to the outside world, although I can't begin to understand why.

Any guru's that can shed some light on this?
Nightwing
Newbie
Posts: 6
Joined: Sun Feb 25, 2007 5:41 am
Location: Kelowna, BC, Canada
Contact:

Network Bug in GentooX

Post by Nightwing »

I initally thought the bug was in GentooX as well, as every other machine on the network could recieve all needed settings from the router I figured to be at fault, my VoIP Vonnage Linksys Router, but GentooX could ping http://www.google.ca as well as anywhere else I told it too. And looking at ifconfig and route as well as resolv.conf said that it was getting all the right settings. The only thing different is that the DHCP ip reported from doing an init.d/net.eth0 restart said "192.168.1.105/24" where as when I specified, I stopped after the 105. etc. I dug out my old Gentoo 2006.0 Live CD and tested it on one of my machines on the same router, worked just fine using iface_eth0, so I tried iface on the Xbox as well as default config_eth0, niether one made a difference, and in fact, iface gives a yellow fonted warning to using the newer config_eth0 standard instead but was still green lighted with OK durring restarts of the NIC itself. In my experiences with Linux, I find that specifying the IP more or less works better especially with the stubborn distro's. I would first specify all IP's to Gentoo in /etc/conf.d/net and /etc/resolv.conf then log into all router's and specify ALL of there IP's rather then let them work using DHCP, even specify the IP that your modem gives normally to your lead router through dhcp. Try giving your router a host name and specify your ISP's default domain name.

Ex. for my ISP, I would use: Xbox.ok.shawcable.net

Let us know how you make out and wiether if any of this makes any difference or not.

Personally, I think that something is defenitley up with the TCP/IP stack in GentooX. Is there such syntax as to specify 'emerge TCPIP' either by obvious means at a bash prompt or else via magic?

--Nightwing
<deadmatrix@gmail.com>
MisterB
Newbie
Posts: 3
Joined: Wed Feb 28, 2007 12:42 pm

Post by MisterB »

Sorry for the late reply, forgot to check back. But I'm sorry to say I took the easy way out. Got tired of trying and installed xebian the same day without any network problems. It also had the latest myth frontend saving me some time (or a lot of time if I had to compile). So for now I'm sticking to that, although I did have to brush up my apt-get skills :wink:
Still, you're doing a great job Shallax!
orochi
Gentoox Guru!
Posts: 606
Joined: Sun Oct 24, 2004 8:11 am

Post by orochi »

Doesn't linux have TCP/IP in the kernel unlike windows?
MisterB
Newbie
Posts: 3
Joined: Wed Feb 28, 2007 12:42 pm

Post by MisterB »

Yes, the TCP/IP stack is part of the linux kernel, although I've never seen this problem before. Maybe a recompile of the kernel would fix it, but I couldn't be bothered at the time :wink:
orochi
Gentoox Guru!
Posts: 606
Joined: Sun Oct 24, 2004 8:11 am

Post by orochi »

Why recompile, if you have never ran magic and it updated your kernel, just take the initrd.gz and vmlinuz (you prolly just need vmlinuz) from the install cd you used to your Gentoox version /shrug
Krazy
Adept
Posts: 54
Joined: Wed Sep 10, 2003 8:41 pm
Location: Auckland, New Zealand

External Networking issue with Home 6.1

Post by Krazy »

Hi all,

Yep.... looks like this one is biting me too. :(
Summary: Networking fails to get outside through my ADSL modem/router.

Just upgraded my XBox to Home 6.1 from Home 4.0, where I had no issues (other than magic tell me to upgrade ;)). I have not changed any wiring, or hardware from that previously working situation. I note that for Home 6.1, the kernel has changed from 2.4.22 to 2.4.32!

Now, on my Home 6.1 networking from my xbox kinda works:
- I've verified Telnet/SSH both work to the same subnet (ie other 192.168.1.x addresses}
- DNS resolves names to addresses
(eg >resolveip shallax.com
IP address of shallax.com is 86.13.233.178)

I noticed this problem by attempting to use "magic", which continously fails after a timeout. From another computer, I can Telnet to the server hosting "magic", and get sensible replies (an error message), so the router is letting the request and response through.

Ok, first thing I did was to attempt to change the hostname. I tried "newhost", but that fails, viz:

Code: Select all

>newhost
Enter new hostname (e.g. saturn): xena
Purging old hostname from hosts file
cat: /etc/hostname: No such file or directory

{no further output; waits indefinitely}
Next I used

Code: Select all

hostname xena
later, I made edits to /etc/conf.d/hostname & /etc/hosts {see below} to get the prompt changed from 'Xbox' to 'xena'.

Things I've tried:
- converted to static IP address from default DHCP
- cable fault? {tried other computers using that cable and they work}
- router firewall on? {yes, but lets other computers access magic, I've also tried with it completely disabled and no change}
- set MTU to 1200 {no difference; did this via ifconfig, I assume you don't need to restart eth0.net for this to take affect as ifconfig shows the changed value immediately}

Here's a dump from ifconfig after a fresh boot

Code: Select all

eth0      Link encap:Ethernet  HWaddr <edited>  
          inet addr:192.168.1.14  Bcast:192.168.1.255  Mask:255.255.255.0
          inet6 addr: fe80::20d:3aff:fe4f:56cb/64 Scope:Link
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:588 errors:0 dropped:0 overruns:0 frame:0
          TX packets:350 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:55495 (54.1 Kb)  TX bytes:42360 (41.3 Kb)
          Interrupt:4 Base address:0x5000 

.... and route:

Code: Select all

Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.1.0     *               255.255.255.0   U     0      0        0 eth0
loopback        *               255.0.0.0       U     0      0        0 lo
default         192.168.1.1     0.0.0.0         UG    0      0        0 eth0

Files:

Code: Select all

# /etc/conf.d/hostname

# Set to the hostname of this machine
HOSTNAME="xena"

Code: Select all

# /etc/hosts:  This file describes a number of hostname-to-address
#              mappings for the TCP/IP subsystem.  It is mostly
#              used at boot time, when no name servers are running.
#              On small systems, this file can be used instead of a
#              "named" name server.  Just add the names, addresses
#              and any aliases to this file...
# $Header: /home/cvsroot/gentoo-src/rc-scripts/etc/hosts,v 1.7 2002/11/18 19:39:22 azarah Exp $
#

127.0.0.1       localhost
192.168.1.14    xena

Code: Select all

# /etc/conf.d/net

# default configuration after install (all uncommented)
#config_eth0=( "dhcp" )
#fallback_eth0=( "192.168.1.13 netmask 255.255.255.0 broadcast 192.168.1.255" )
#fallback_route_eth0=( "default via 192.168.1.1" )

# Changed to Static IP configuration after reading:
# http://www.gentoo.org/doc/en/handbook/handbook-x86.xml?part=4&chap=1
config_eth0=( "192.168.1.14 netmask 255.255.255.0" )
routes_eth0=( "default gw 192.168.1.1" )
... and /etc/resolv.conf ---- on a fresh boot its now empty!! :(
Previously it had:

Code: Select all

nameserver 192.168.1.1
<EDIT>

I replaced the above line in /etc/resolv.conf,
and then restarted the interface (/etc/init.d/net.eth0 restart).
DNS, unsurprisingly is back.

Interestingly, during the interface restart I got messages saying:
* One of the files in /etc/{conf.d,init.d} or /etc/rc.conf
* has a modification time in the future!

Regardless, I rebooted and this time the namerserver setting stuck. Fingers crossed!

I also tried added my ISP's nameserver to this (2 entries) but it made no difference.

</EDIT>

I note from the ifconfig output above that there's an IP6 address; it was probably there for Home 4.0 (I can't remember). Given my router/modem is quite old now, I'm wondering whether its getting confused by any IP6 traffic the xbox may be sending.

As I'm not really familiar with networking, or linux tools that come standard with the Home 6.1 package, I don't know what to do next.

Suggestions gratefully received.
v1.1 xbox, stock HD, soft-modded [uxe+ dualboot: UnleashX/GentooX], with Pro2.0
v1.4 xbox, stock HD, Xecuter3 (X3 3294), running headless with Home 6.1
Krazy
Adept
Posts: 54
Joined: Wed Sep 10, 2003 8:41 pm
Location: Auckland, New Zealand

Post by Krazy »

Ok, because the last post was a long one, I want to clarify a few points.

1) I've fixed my DNS issue

2) External network access is still broken!!! :(
This is MAJOR because I can't:
- run magic
- emerge sync

Here's my evidence:

eg Telnet from XBox

Code: Select all

telnet shallax.com 2000
Trying 86.13.233.178...
telnet: Unable to connect to remote host: Connection timed out
eg Telnet from another computer on my lan

Code: Select all

telnet shallax.com 2000
Trying 86.13.233.178...
Connected to shallax.com.
Escape character is '^]'.
get / HTTP/1.1

HTTP/1.1 400 Bad request
Date: Tue, 20 Mar 2007 07:56:20 GMT
Server: ShAPAChE v2.03 (Linux)
Last-Modified: Sun, 19 Dec 2004 22:06:37 GMT
Accept-Ranges: none
Content-Length: 207
Connection: Close
Content-Type: text/html

<html>

<head>
<title>ShAPAChE: 400 - Bad request</title>
</head>

<body>
        <h1><font face="Verdana">ShAPAChE Error: <font color="#FF0000">400</font> - 
        Bad request.</font></h1>
</body>

</html>
Connection closed by foreign host.
Note: Don't get distracted by the fact I got an error message above.
I believe the fact that I got a response proves that firewall on my router is Ok!

What to do? Help please..... I don't want to go to the dark side (debian etc)
v1.1 xbox, stock HD, soft-modded [uxe+ dualboot: UnleashX/GentooX], with Pro2.0
v1.4 xbox, stock HD, Xecuter3 (X3 3294), running headless with Home 6.1
Krazy
Adept
Posts: 54
Joined: Wed Sep 10, 2003 8:41 pm
Location: Auckland, New Zealand

Post by Krazy »

What to do? ...
Ok, I've unsuccessfully tried:
- returning to DHCP
{Yes, DHCP works, I get an address allocated... but external networking is still stuffed}
- setting MTU to 1000
- setting MTU to 1500
{This seemingly has no effect}

:( :(
v1.1 xbox, stock HD, soft-modded [uxe+ dualboot: UnleashX/GentooX], with Pro2.0
v1.4 xbox, stock HD, Xecuter3 (X3 3294), running headless with Home 6.1
RapCitSD
Newbie
Posts: 2
Joined: Sat Mar 03, 2007 7:18 pm
Location: Rapid City, SD

Post by RapCitSD »

I was having similar problem, what I did was comment out the default fallback lines, and only kept the "dhcp"
jeytr
Newbie
Posts: 2
Joined: Thu Apr 12, 2007 8:35 pm
Contact:

Post by jeytr »

My same exact problem, erasing fallback doesnt work. I would be puzzled if it did though.
Krazy
Adept
Posts: 54
Joined: Wed Sep 10, 2003 8:41 pm
Location: Auckland, New Zealand

Post by Krazy »

Hi all,

Thanks to Shallax and Jetyr who solved this :) I hadn't had time to delve into this myself.
And the solution is...

Login as root.

Code: Select all

cd /usr/src/linux
make menuconfig
Go to "Networking Options ---> IP: TCP Explicit Congestion Notification Support" and disable this option by pressing 'n' on your keyboard.
Exit the menuconfig system and choose to save the config.

Code: Select all

make dep
make bzImage modules modules_install
cp arch/i386/boot/bzImage /mnt/fatx/e/vmlinuz
reboot
v1.1 xbox, stock HD, soft-modded [uxe+ dualboot: UnleashX/GentooX], with Pro2.0
v1.4 xbox, stock HD, Xecuter3 (X3 3294), running headless with Home 6.1
orochi
Gentoox Guru!
Posts: 606
Joined: Sun Oct 24, 2004 8:11 am

Post by orochi »

Lol, I might have had this problem years ago, if this problem even existed a few previous versions of Gentoox.


viewtopic.php?t=3032&highlight=connect


I'm glad you were able to get it fixed.
jeytr
Newbie
Posts: 2
Joined: Thu Apr 12, 2007 8:35 pm
Contact:

Post by jeytr »

In the end it all appears to just be people using certain routers that don't support the protocol. Making the bug more sporadic and hard to catch. Since it is fixed lets see what we have in common, lol. I have a Linksys BEFSR41 v4.
Post Reply