Connection works but not for me...
 

Hi. Just to clarify things - I'm a newbie. Been using RHL8.0 for about four months. I have it up and running on its own box. That's how I can get to the web. :D

Now, for current issue. I am running Mandrake 9.1 on a Gigabyte GA-8SMMLP mobo. The network adapter (integrated on-board) is a RealTek RTL-8139. I know it's working because I can ping the card at its IP address and through the loopback. Plus it connected to get the on-line updates for my distro. I know I have a configuration setting out of whack somewhere, but I don't know where. It all looks good to me.

Like I said, I'm a newbie, so I don't know what information I have that might be useful so I don't know what to post. I know the IP address is good, 'cuz it's the same one I use when I boot to Windows. All the rest of the settings (gateway, broadcast, etc.) are the same as my Windows configuration. I can't figure it out. Can anyone either help or ask for the right info to help?

:study: :Pengy: :study:

could you post your settings because there's not much to go on.
try ifconfig for your ipadress, route -n for the route table and cat /etc/resolve.conf for your dnsservers.

GJB

ip address: 192.168.1.1
dnsservers: 206.156.230.1 206.156.230.2
route table:
Dest Gateway Genmask Flags Metric Ref Use Iface
192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
127.0.0.0 0.0.0.0 255.0.0.0 U 0 0 0 lo
0.0.0.0 192.168.1.4 0.0.0.0 U 0 0 0 eth0

domain tranquility.net
nameserver 127.0.0.1
nameserver 206.156.230.1

I hope I got it all

:confused:

okay your default gateway is 192.168.1.4, can you ping that box from this box ?
if you can ping that box can you ping 1 of the dns servers ?

GJB

Nope. I can ping the IP address of eth0 and lo on that box only. I can't ping ANY other computer on my net or the internet.

do you have a firewall installed on the mandrake box ?

No. I have a dedicated box for my firewall, so I left the Mandrake box fairly open.

If you gave an iptables -F on that box and then try to ping ?

GJB

[QUOTE]Originally posted by GJB
If you gave an iptables -F on that box and then try to ping ?

GJB [/Q]

-bash: iptables: command not found

not much help to me. does it tell you anything?

I notice that the last entry in the route table has type U, not UG. AFAIK that means there isn't a default route pointing to the gateway. Not being a RedHat user, I don't know the file to edit to fix this, but one of you RH types surely do.

I have checked the route table for my RH box (is working) vs. the route table from my Md box (posted above, not working) and all of the entries in each table are identical - including the Flags for all three entries being "U" rather than "UG". Since my problem is not only unable to ping gateway, but unable to ping any other hosts on the same hub as this one. I'm not discounting this as the solution, I will look into it. But, given that it's configured (on the routing table, at least) identical to a working box, I don't hold out much hope that this will solve my problem.

Okay, I gotcha. Best bet seems to me to snoop the net from a working box (one of the ones you can't ping) to see if anything is getting across the wire. Ethereal will show you ICMP (what a ping is).

By the way, this may be a stupid question, but what does ifconfig -a show?

Sorry, but you've lost me now. What exactly do u mean by "snoop the net"? Does that mean check my internal network? Or check around the internet? And for what? What is Ethereal? Sorry for all the questions, but I am a newbie. Don't know most of the lingo.

ifconfig (-a) is pretty much identical on both boxes, except the IP addresses. One ends in .1 and the other ends in .254. The RH box ends in .254.

As for pinging, I can ping all the boxes on the network (and the internet) from all boxes EXCEPT my Md box WHEN running Md. I boot this same box to Win2k and it works wonderful. When Md is loaded, I can't ping out nor ping the Md box from anywhere else on the network.

Does any of that help?

Yep, that's sort of where I was going. When running mandrake, I am guessing that if config -a will show that eth0 has an IP address assigned. This would mean that 'ifconfig eth0' would show something like this on it's second line:

inet addr:192.168.1.1 Bcast:192.168.1.255 Mask:255.255.255.0

If you don't see that line, thats a problem, but it sounds like you do. If so, look at the next line; it should say UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1

Look at the lines begining RX packets and TX packets. What do they say?

Ethereal is a really nifty GUI tool that shows you everything that's going on on your local network. Google for it, and you will find rpms for your RH box. By 'snoop the net' I meant running ethereal or something like that to see what's happening.

The part that has me puzzled is that this box runs fine with W2K, but not with Mandrake. If in fact under Mandrake eth0 is correctly configured and up, why no pinging?

I'll have to get back to you on this. I appreciate (1) your willingness to help me with this and (2) your immediate availability. Unfortunately, the box (along with the rest of MY network) is on the other side of town. It'll be about 45 minutes (from now) before I can get there and check this stuff. I hope you're still around when I get that done. I'll post when I get more info.

:)

Okay, musrum, I hope you're still there. If not, I hope somebody else can pick up where you left off. And many thanks, either way!

I ran the ifconfig -a on both boxes, to compare. Aside from the obvious (MAC and IP addresses) the only differences were that the Md box shows 0 packets for RX and TX - no traffic here - and the Interrupt is 17 on Md box and 5 on RH box. The IRQ (17) seems very suspect at this point, what do you think?

Still here, but not for long, I hope!

The interrupt isn't suspicious on a modern box, but it does bring up another thing to look at:

cat /proc/interrupts

You should see a list of all mapped interrupts. Are there any interrupts on 17? And is it tied to eth0?

Another thing to try:

/sbin/lsmod

This should show you the loaded modules. One of these is likely '8139too'. If it is, and the 'Used by' column says '0 (unused)', then try 'rmmod 8139too' followed by 'modprobe 8139too debug=255'. Then look in /var/log/debug and /var/log/messages and /var/log/system for any interesting messages.

Thanks for all of your help. cat /proc/interrupts does show IRQ 17 mapped to eth0, although I'm not sure what everything else means. See:

17: 0 IO-APIC-level eth0

I'll try the other and see what it provides.

/sbin/lsmod shows Used:1, so I guess that means that it's loaded and being used, even though I can't control it? Hmmmmm.

Would this have anything to do with the Mandrake network config utility requiring a domain entry (host.domain)? Just curious, since there was no problem downloading the install updates, but nothing since I tried to configure the adapter for my use...

The line from /proc/interrupts means that your card is not generating interrupts. Assuming (and I think from your earlier posts it's safe to assume) that the network cable, hub, switch. whatever is good, the my second suggestion is even more pertinent.

BTW, I don't remember if you posted this - does the card have a link light? Is it lit? (You know, the little green led on the card that lights when you plug in the ethernet cable.)

I haven't *physically* verified the link light is on, but the darned thing works a-okay under Win2k, so I'm figuring it's still there and working. Along with the rest of the hardware.

So, I'm going to do the rmmod... stuff and see what comes up. Hopefully (for me) you'll still be here when I get some results.

rmmod gave a busy indication (already installed and running)

I looked in the logs, but I found nothing interesting. What I found in relation to eth0 is that it keeps timing out on the TX and resetting to 100M.

Alright. Now we have something to try - let's try telling the module to run at the media speed your network is really running at; which I am assuming is 10.

In your /etc/modules.conf file you need to add a line to pass an argument to the 8139too module when it loads. To do this add the following line:

options 8139too '"media=\x0210 debug=255"'

The quotes are single quotes outside of double quotes. That will turn on debugging (I hope) and set the media speed to 10 (again, I hope - I may have the syntax for expressing a hex number wrong. Looking at the logs should tell you. Other form would be x0210, without the leading backslash.)

options 8139too '"media=\x0230 debug=255"'

That line should force the media to 100. To see the goodies available you can look at the source (what I did cause I don't have that module built, or type the following:

modprobe -p 8139too

That should list all the arguments you can pass to the driver.

Why are you assuming that the media is running at 10Mb rather than 100Mb? The interface (on all of the boxes, for that matter) is 10/100, as is the switch they're all plugged into. Not arguing, just a newbie question.

???

From your earlier post, "keeps timing out on TX and resetting to 100M". Don't know either way, but thought it might be worth a try.

The only other things I can think of are trying to boot with noapic or acpi=off.

That was suggested by another source - turning off ACPI (in BIOS) - but I can't find a setting to turn that off in my BIOS.

I'll look into the noapic setting. Do you have any suggestions where I might find/set this?

Okay, I seem to be doing something incorrectly.
When I try to run

modprobe -p

I get an error indicating -p is an invalid switch. What gives?

Sorry, but what about etc/host & /etc/resolv.conf & etc/sysconfig/network files?

The correct command to get the info for your module is modinfo -p 8139too

options 8139too media=0x10 debug=255 (0x10 is 100mbit half duplex)

What you could also try is use a good old m$ dos bootdisk, put rset8139 on it
ftp://210.51.181.211/cn/nic/rtl8139a...et8139-503.zip

and set your hardware to 100mbit full duplex that way it won't try to autonegotiate (which sometimes causes problems)

Maybe that could solve it.

GJB