Weird problem with Linux and routing table...

[CrazyHelloDeli]

My routing table looks like this for my cable ISP:

216.xxx.xxx.xxx * 255.255.255.255 UH eth0
216.xxx.xxx.xxx * 255.255.255.0 U eth0
127.0.0.1 * 255.0.0.0 U eth0
default 216.xxx.xxx.xxx 0.0.0.0 UG eth0

My problem is that im experiancing extremly slow net browsing when using RH7.2 and Netgear FA311. If I ping anything beyond my default gateway(by hostname or IP) its extremly slow and results in always about %90 packetloss and 30 second long pings. I know its not my ISP because it works great on Windows. Ive also noticed that when printing my routing table via "route" it takes about 30 seconds for my first entry to pop up, the next to pop up instantly, and then the GW takes about 30 seconds. Its as is linux is trying to resolve both of those routing entries or forgot where it put them. Everything works fast if I ping anything on my local network, but it takes forever for Linux to "remember" where to look for anything past the GW. Anyone know what the deal is? Its driving me nuts.

 

[CrazyHelloDeli]

I found out I can use "route -n" to show my routing table without it pausing like I described above, which says to me its hanging trying to map "default" to 0.0.0.0. Ive tried to remove the DG entry and re-add it via:

"route add default gw 216.xxx.xxx.xxx netmask 0.0.0.0 dev eth0"

But for some odd reason it ALWAYS puts the genmask to 255.255.255.0 no matter what I specify, which of course doesnt work. Surely someone knows what the deal is(Nothingman, n0cmonkey?) I cant do anything with %90 packloss and 30,000 pings

 

[n0cmonkey]

Why do you want a netmask of 0.0.0.0 for your gateway? Sorry, I dont know a whole lot about Linux networking... Havent touched the OS in a month or so

 

[marat]

Here is mine

Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.1.0 * 255.255.255.0 U 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default 192.168.1.1 0.0.0.0 UG 0 0 0 eth0


where 192.168.1.1 is my router. So you are supposed to have 216.xxx.xxx.xxx istead of 192.168.1.1 and different netmask on eth0.

Edit: Either delete the first route (with netmask 255.255.255.255) or correct the netmask.

 

[CrazyHelloDeli]

<< Here is mine

Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.1.0 * 255.255.255.0 U 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default 192.168.1.1 0.0.0.0 UG 0 0 0 eth0


where 192.168.1.1 is my router. So you are supposed to have 216.xxx.xxx.xxx istead of 192.168.1.1 and different netmask on eth0.

Edit: Either delete the first route (with netmask 255.255.255.255) or correct the netmask. >>



Ok, I deleted the first route of 216.xxx.xxx.xxx with the 255.255.255.255 genmask and the UH flag. Now here is where it gets screwed up. If I add the DG with:

route add default gw 216.xxx.xxx.xxx eth0

It pops back a promt like it worked, but then when I ask it to show me whats in the routing table with "route" I get a list as follows:

216.xxx.xxx.xxx * 255.255.255.0 U eth0
127.0.0.1 * 255.0.0.0 U eth0
default 216.xxx.xxx.xxx 255.255.255.0 UG eth0

Wtf is up with that?! For my default route im SUPPOSED to have 0.0.0.0 in the Genmask column am I not? Not only does the Genmask always show up incorrect, but like I said above it takes about 30 seconds for the DG entry in the table to show. I have a gremlin in my machine!



<< Why do you want a netmask of 0.0.0.0 for your gateway? Sorry, I dont know a whole lot about Linux networking... Havent touched the OS in a month or so >>



I need the netmask(rather, Genmask) to read 0.0.0.0 to signify "use default route". Whereas if I needed a route to a particular host id need a Genmask of 255.255.255.255. But, im sure you already knew that.

Its at the point where I think ive got a corrupt "route" program on my hands, because the behavior doesnt make any sense. Looks like a reinstall, unless someone can tell me which RPM has the route(and other tcp/ip) configuration programs.

 

[n0cmonkey]

<< I need the netmask(rather, Genmask) to read 0.0.0.0 to signify "use default route". Whereas if I needed a route to a particular host id need a Genmask of 255.255.255.255. But, im sure you already knew that. >>



I had no clue. Im not a linux expert so wierd things like this confuse me

 

[manly]

<< Its at the point where I think ive got a corrupt "route" program on my hands, because the behavior doesnt make any sense. Looks like a reinstall, unless someone can tell me which RPM has the route(and other tcp/ip) configuration programs. >>


I seriously doubt route is corrupt. Unix systems don't have mysterious problems like Windows where a package reinstall would fix the trick.

Finding out which RPM owns an executable is quite easy. As root:

# rpm -qf `which route`

Or likely, as any user:

$ rpm -qf /sbin/route

Read up on rpm; it's a pretty handy tool in the sysadmin's toolkit. I find Linux package management (even an imperfect system like RPM) much better than Windows' undocumented (or InstallShield) approach.

The only suspicious part of your routing table is the route to 127.0.0.1. First off, I don't even have an explicit route to localhost in my routing table. I suspect 127. is handled automatically within the Linux kernel networking code as a special case.

Secondly, it looks like there's a network netmask for a host IP (255.0.0.0 for 127.0.0.1); the flag indicates that is a network route. Third, that route is associated to the Ethernet interface eth0 instead of the loopback device lo.

My guess is that if you remove that route, the problem should go away. Not sure why the "genmask" was weird when you manually installed the default route, but try the other suggestion first.

 

[marat]

Let's say your IP 216.215.214.17 / 255.255.255.0

Let's say your Gatewat IP is 216.215.214.1 / 255.255.255.0

Than you routing table should look the following way:

216.215.214.17----*----------255.255.255.255---UH---0---0---0---eth0 - route to your host from your host
216.215.214.0-----*-----------255.255.255.0-------U---0---0---0----eth0 - your lan
127.0.0.0------------*----------255.0.0.0--------------U----0---0---0----lo - loopback
default----216.215.214.1-----0.0.0.0---------------UG---1---0---0--- eth0 - gateway

 

[DirtylilTechBoy]

How do routing tables work??

 

[slackware1995]

<< My routing table looks like this for my cable ISP:

216.xxx.xxx.xxx * 255.255.255.255 UH eth0
216.xxx.xxx.xxx * 255.255.255.0 U eth0
127.0.0.1 * 255.0.0.0 U eth0
default 216.xxx.xxx.xxx 0.0.0.0 UG eth0

My problem is that im experiancing extremly slow net browsing when using RH7.2 and Netgear FA311. If I ping anything beyond my default gateway(by hostname or IP) its extremly slow and results in always about %90 packetloss and 30 second long pings. I know its not my ISP because it works great on Windows. Ive also noticed that when printing my routing table via "route" it takes about 30 seconds for my first entry to pop up, the next to pop up instantly, and then the GW takes about 30 seconds. Its as is linux is trying to resolve both of those routing entries or forgot where it put them. Everything works fast if I ping anything on my local network, but it takes forever for Linux to "remember" where to look for anything past the GW. Anyone know what the deal is? Its driving me nuts. >>



First off, I am going to assume that your cable ISP is using DHCP.

As root (without quotes):
Run "/sbin/ifconfig eth0 down"
"/sbin/dhclient eth0"

This should reset your routing tables.

do a "route -n" to confirm
also make sure that "/etc/resolv.conf" is correct
you can verify it is correct by "/usr/bin/nslookup"
verify that your ethernet device looks good by "/sbin/ifconfig eth0"

Let us know what that does for you and what info you get. Also, give us your tcp/ip info from windows to compare with.