Upgrade Broke Network IF

[barf]

I have only recently discovered this problem on my Aspire 5020 laptop which was working perfectly with Kernel 2.6.23.1-42.fc8. The laptop is mostly used for running a few apps in Windows that cant run in Linux.

After the last software update the network interface stopped working. It wasn't immediately obvious, I didn't notice until the next time I booted into linux and it complained that it couldn't receive updates. It would appear that the 2.6.24.3-34.fc8 kernel has dropped support for my RTL8169 network card (that only took minutes to write. it has taken hours to figure out where the problem lay) :mad:

So I thought I'd just boot up in the old kernel until this is resolved, bugger, I cant get the X server started anymore. Even tried changing the driver to 'vesa' which always got X started, but not this time seems there have been alot of changes ( just to clarify, boot kernel 24.3-34 get X, boot 23.1-42 no X). I'm not too bothered about the X sever but can anyone point me in the direction of getting the network up again.

[bradchaus]

post output of

ls /etc/sysconfig/network-scripts

[cable_txg]

I'm experiencing the same problem from my previous post. eth0 and wlan0 no longer work! dhclient fails. I could only go online from the previous kernel (no X). :(

[barf]

post output of

ls /etc/sysconfig/network-scripts

Hey thanks for looking in. Could you perhaps tell me what you are interested in. I've got no network to this machine and there's about 30 entries in this folder, can't cut and paste and I'm not the best typer.

[barf]

I'm experiencing the same problem from my previous post. eth0 and wlan0 no longer work! dhclient fails. I could only go online from the previous kernel (no X). :(
Great to know I'm not the only one. Similar problems here http://www.fedoraforum.org/forum/showthread.php?t=184277.

When you look in Network Configutration does your device nickname end in .bak. i didn't mention the issue with the WLAN in my original post as I rarely use it. So full story is both network interfaces no longer work. If you open Network Configuration we have one device for ETH0 with a nckname of eth0.bak which is inactive. Under the hardware tab there are 3 entries.
Broadcom BCM4318 Ethernet eth0 ok
Broadcom BCM4318 Wireless wlan0 system
b43 Wireless wlan0 system

So the first entry is a wlan device identified as a eth connection. What is going on?

[Wizofos]

Ok this is connected to another post. For some reason when you run updates after installing Fedora 8 your NIC's don't come up started. I thought that this might be due to having more than one NIC but removed one and still had the same problem. Found the solution when I re-installed the second NIC on my system.
This is from KDE
1. go to SYSTEM/NETWORK.
2. delete the device
3. check in the hardware tab that you see your NIC's
4, now back to devices and NEW and re-define the NIC's

Tried this and now have both NIC's working when I reboot even after I power cycle.

[barf]

Thanks for the input but it doesn't help in this case. I've deleted all records in Network Configuration, rebooted and when I look into after reboot it is as it was previously which means Fedora is detecting and setting these entries. If I try to set an eth 1 and select a network card when I next look at it it has reverted to a Broadcom Wireless. I thnk this is because for whatever reason Fedora is no longer detecting the eth NIC.

[Wizofos]

Barf,
Try deleting them then adding them without the re-boot in between. The problem seems to be when you re-boot the hardware detector/installer finds them and adds them automatically. You have to add them manually to make this work. From what I can see this may be a problem with the auto installer building a bad config file.

[barf]

:( No it still wont work.
Using the the networkable 23.1-42 (but still no X) kernel I did a yum kernel update ('cause I knew there was 24.3-50 update). This got my RTL8169 NIC back in the game so I could now actually configure the devices but still can't activate them. When I try I get a 'determing IP address' window pop up and the computer locks up.
Having now wasted eons trying to sort this I decided it may be more productive to re- install F8 and start with a clean slate. The vesa driver in kernel 23.1.42 now works :confused: again as does the network (well the LAN any way, not tried the WLAN yet as that needs a bit more work). I then tried to boot kernel 24.3.-50 but it hung up when trying to start the network. So I rebooted 23.1-42 and turned off the start network device on boot option. Boot 24.3-50, it works. Have a look in in Network Configuration tool, all looks good, try to activate the LAN, same as above with the computer locked up. I' think I'll stick with kernel 23.1.42 for now until I get a response from the bug report I submitted.
If you are having the same or similar problems please post here, I'd like to know the extent of this problem.

[steever]

Hi Barf - any news on this? I'm in exactly the same situation as I have the same laptop - actually a 5024 - which has been happily chugging along with Fedora 8 until there was a kernel upgrade. There's something going on with networking in that kernel. And I guess there aren't enough users affected by it to warrant a fix. The big problem is that Fedora 9 is coming and it uses the same kernel.

[barf]

Hi Steever, I've just tried the latest kernel update (24.4-64) which had 'many network updates' but it still doesn't work. It seemes to be a bit better, during boot it now gets past bringing up eth0 but now hangs on assigning ip address. I've tried it with DHCP and a statically assigned address. This is really disappointing.
If you want to look at the bug or add acomment to it you can find it here https://bugzilla.redhat.com/show_bug.cgi?id=43910

[steever]

HI Barf. I think that bug is for something else, something about floppy drives. Can you repost?

Steve

[kntry_boy]

System: HP Pavillion DV9720us with the Atheros ar5006eg wireless network card

Newly installed Fedora 8 64bit, and all was well

I then figured I would get all the updates installed, to make sure I was uptodate.

After installing the available updates, which also included the kernel update
From 2.6.23.1-42.fc8
To 2.6.24.3-50.fc8

Before the update, wlan0 came up and worked just fine
After the update, I was unable to bring up wlan0 at all, and that includes trying to use madwifi, and ndiswrapper

Funny that when I boot from the older kernel, wlan0 comes up and works fine again, appears to be related to the kernel intself, or posibbly the modules the newer kernel is loading

I am up for any suggestions to get this working again, I like to be uptodate, willing to run older kernel for awhile.

[barf]

HI Barf. I think that bug is for something else, something about floppy drives. Can you repost?

Steve
Sorry Seve, dropped a digit,its 439107

[steever]

Any luck Barf?

[EndGame]

Im having the same problem! new to this forum btw, and new to linux alltogether. I updated, and then bam, connections lost. However, when i screw around with it. I see eth0 go active for 2 seconds, then back again. Also i got a random msg at one point saying something about theres a download for new things for it. Something like that. I got mad and had a few beer instead. What do i do??

[barf]

Hi Steever, There have been several updates to the network packages and kernel and the network configuration tool now works as you would expect BUT the computer still locks up after trying to start eth0. I did once get it working when in a moment of frustration I deleted all entries in the configuration tool and the network manager popped up a balloon saying the network was connected. It didn't survive a reboot and I've been unable to reproduce it. I've posted an update to the bug which is still showing as new so no one has taken it on. Pity really as you'd think the developers would be keen to fix things that they've stuffed up.

[barf]

Im having the same problem! new to this forum btw, and new to linux alltogether. I updated, and then bam, connections lost. However, when i screw around with it. I see eth0 go active for 2 seconds, then back again. Also i got a random msg at one point saying something about theres a download for new things for it. Something like that. I got mad and had a few beer instead. What do i do??

To carry on using Fedora with networking you need to boot the old kernel. When you boot the comp at the screen which says 'booting kernel *********** in 5secs' press any key. This will present you with a menu where you can pick which kernel to boot.

As you are new to linux I would suggest that you try a different distro. Ubuntu is a good choice for a beginner however if you choose to stick with Fedora then put more beer in the fridge. :)

[steever]

The most annoying thing is that Fedora 9 will use the same kernel. That means until someone fixes this problem, we're stuck using Fedora 8 with an old kernel. It's really annoying.

I moved to Ubuntu Hardy. Nice distro. I'll come back to Fedora when this problem is fixed.

[EndGame]

I dont mind putting more beer in the fridge. I like Jumping in head first. So.. what should i do?

[barf]

I dont mind putting more beer in the fridge. I like Jumping in head first. So.. what should i do?
Reread my post to you!

[barf]

I've bitten the bullet and installed F9 preview and after much stuffing around I finally got the network to work. First of F9 does not start the network until you actually login (handled by network manger). So you need to get rid of NM and NM Dispatcher and you must not start the network until you have a desktop. Here's how I did it I hope it helps others.

First boot into a terminal. At the grub boot screen press any key, you'll see a menu (might only have one item) move the cursor onto the kernel you want to boot and press 'e'. On the next screen move the cursor onto the line that starts with the word 'kernel' and press 'e'. At the end of the line after 'quiet' type a space followed by 'single' (don't type the single quotes). press 'enter' then 'b'. The system will do some boot procedures then present you with a command prompt. We want Fedora to boot up in run level 3 so we are going to edit /etc/inittab/ using the venerable text editor VI so at the command prompt type vi /etc/inittab. Now press 'i' to enter insert mode and using the cursor keys scroll to the end of the file. Move the cursor so that it is under the '5' and press 'delete' then '3'. Press ESC to go back to command mode then hold down 'shift' and press 'z' twice, this will save your changes and exit VI.
Now we are going to edit the ifcfg-eth0 script. Type vi /etc/sysconfig/network-scripts/ifcfg-eth0 edit this file in the same way as above so that ONBOOT and NM_CONTrOLLED = no ( don't worry if NM_CONTROLLED is missing).
When you are back at the command prompt type 'exit'. The system will finish booting up into a command line terminal. Type 'root' and enter at the prompt and then enter your password when requested. Type 'startx' and enter. You should eventually get a desktop.
Access the services configuration tool and stop and disable the Network Manager and Dispatcher.
Access the network configuration tool and give it an IP address, netmask, gateway and DNS IP now activate the network. I get a period lasting about 2 - 10 minutes were the system is slow to respond but eventually settles down (this is what stuffs my system up and prevents booting up with the eth0 activated).
Now open a terminal and type 'gedit', use this text editor to edit /etc/inittab and change the default run level back to 5 from 3. Now when you reboot you will get a desktop and the network will need to be started manually.

After updating (yum update, couldn't get the tool to work)) I am now running F9 Sulphur 2.6.25-14.fc9.x86_64???!! (thought it wasn't released until teh 15th)

[EndGame]

Ok i got the fedora 9 preview myself. Im so new at this it makes me sick. I did everything you said. However i never had to do startx or anything. It just went right into login screen. Where do i put the ip addres netmask and gateway stuff?

[EndGame]

nvm ignore that. How do you do yum update?

[EndGame]

and.. .networks not working ? it says active? im just not getting internet.

[barf]

You do a yum update by opening a terminal and as root type'yum update'.
If you have no internet you can't do a yum update.

If you have followed my instructions properly you should have a connection. Have you given a valid DNS IP address to the interface? Ask your ISP for one if you're not sure.

A bit of general advice as you're new go hear and learn http://rute.2038bug.com/index.html.gz

[barf]

I recently found that it works with no problems in KDE. :)

[Perty]

I don't know if this is the same issue but my wireles (3945) work the first time I booted after I installed FC 9. The second boot it didn't work. After some investigations I deleted the wireless config in NetWrkmanager and afer that it worked again.

I discovered that the macadress was empty and I have set my router to only serve some mac-adresses and when copying my macadress in the empty field it worked.

[EndGame]

well i tried KDE... it didnt work either. This is really getting irritating. and i havent even gotten to the hard stuff yet. why is it being so difficult?

[scotty38]

What's the output of chkconfig --list look like? Check for an instance of NetworkManager (near the top) and network (alphabetically).

Try getting rid of NetworkManager by:

chkconfig NetworkManager off

and adding network (if not already there) by:

chkconfig network on

and see how you go.

You can revert using the obvious commands......

[hyperspace]

This is the same type of problem/issue people are having with F9. Like someone said, "same kernel". I did what scotty38 said, disable NM, to get wired and wireless working under F9.

[barf]

There are obviously a number of problems with Network manager however they are not relevant to this thread. The issue being discussed here is not that you cant connect to the network but rather that your computer locks up when starting the interface (although its taken a bit of work to identify this as being the problem). The cause is the tickless kernel and its not just Fedora users experiencing this problem (unexplained system lock ups). The solution is to boot with nohz=off. If you don't know how to do that follow the instruction on the line starting with 'First boot into a terminal' in my post above that starts with 'I've bitten the bullet' and replace ' 'single' ' with nohz=off .
If this works for you you can make it permenant by adding it to that line in grub.conf.

[barf]

Ok i got the fedora 9 preview myself. Im so new at this it makes me sick. I did everything you said. However i never had to do startx or anything. It just went right into login screen. Where do i put the ip addres netmask and gateway stuff?

Lets try and solve your problem. If you did everything I said you would not be able to go to a login screen. So try it again.

If you have solved the issue please say so and I'll stop wasting my time.

[jganci]

I'm also having this problem. Realtek RTL-8169 on K8T NEO 2 motherboard. My last fedora kernel that has ethernet connectivity is 2.6.23.15-137.fc8. I've applied every update since then, but have yet to get connected. This includes installing fedora 9. I've been following another fedora forum thread about this:

http://forums.fedoraforum.org/forum/showthread.php?t=182994&page=1&pp=15

I just found this thread last night. I did cross-reference this thread there.

It's well over 3 months without a solution. Does anyone know if the problem is being actively worked?

[barf]

See post 32 above. Boot with nohz=off

[jganci]

See post 32 above. Boot with nohz=off

Thanks for your reply.

I did boot with nohz=off (several times). Still hangs during boot at the "Determining IP information for eth0 ..." message.

Some additional information. When I installed fedora 9 I followed the instructions at

http://www.mjmwired.net/resources/mjm-fedora-f9.html

to disable NetworkManager and enable network. I've been using network for as long as I can remember with no problems until the kernel update that replaced kernel 2.6.23.15-137.fc8.

I booted fedora 9 again, with nohz=off. This time I let the boot finish (mine eventually does). Via System->Administration->Services I verified that NetworkManager was not running. Via System->Administration->Network I verified that my eth0 was there with "Controlled by NetworkManager" not checked, "Activate device when computer starts" checked, "Automatically obtain IP address settings with dhcp" radio button selected, and "Automatically obtain DNS information from provider" checked. That's what I have now for my working fedora 8. The device was, of course, inactive.

I then did a "tail -f /var/log/messages" in a terminal window and in another terminal window I issued "/etc/init.d/network start". I saw the "... for eth0 ..." message in the terminal window, but nothing else. However, /var/log/messages was showing messages, shown below. Text continues after those messages.

May 31 00:41:21 localhost kernel: r8169: eth0: link up
May 31 00:41:23 localhost avahi-daemon[2000]: Registering new address record for fe80::211:9ff:fe2e:a534 on eth0.*.
May 31 00:41:25 localhost dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 5
May 31 00:41:30 localhost dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 6
May 31 00:41:36 localhost dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 7
May 31 00:41:43 localhost dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 21
May 31 00:42:04 localhost dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 10
May 31 00:42:14 localhost dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 10
May 31 00:42:24 localhost dhclient: DHCPDISCOVER on eth0 to 255.255.255.255 port 67 interval 2
May 31 00:42:26 localhost dhclient: No DHCPOFFERS received.

When the "No DHCPOFFERS received" message appeared, the network start ended with "FAILED". Perhaps this is a clue.

Note that similar information is in the other fedoraforum thread I mentioned, along with the messages that appear when I issue "ifdown eth0" followed by "ifup eth0" on my working ethernet. The information is at post #39 there. For completeness, the /var/log/messages data from that thread showing a successful down and up of eth0 were

May 28 22:26:22 new-host kernel: r8169: eth0: link up
May 28 22:26:22 new-host dhclient: DHCPREQUEST on eth0 to 255.255.255.255 port 67
May 28 22:26:22 new-host dhclient: DHCPACK from 192.168.1.1
May 28 22:26:22 new-host avahi-daemon[2180]: Joining mDNS multicast group on interface eth0.IPv4 with address 192.168.1.2.
May 28 22:26:22 new-host avahi-daemon[2180]: New relevant interface eth0.IPv4 for mDNS.
May 28 22:26:22 new-host avahi-daemon[2180]: Registering new address record for 192.168.1.2 on eth0.IPv4.
May 28 22:26:22 new-host NET[3051]: /sbin/dhclient-script : updated /etc/resolv.conf
May 28 22:26:22 new-host dhclient: bound to 192.168.1.2 -- renewal in 36760 seconds.

Maybe this will help someone.

[barf]

Hi Jganci.

I have just changed my network config to get its ip info via dhcp and guess what, it didn't work. Looks like we have yet another issue to resolve.
Try setting the network address manually and boot with nohz=off.

[jganci]

Hi Jganci.

I have just changed my network config to get its ip info via dhcp and guess what, it didn't work. Looks like we have yet another issue to resolve.
Try setting the network address manually and boot with nohz=off.

I'll try it, but I'll need some instructions. I have a router from my internet service provider. I can manage it using firefox, going to http://192.168.1.1. As I'm typing this reply I see that my PC is connected via 192.168.1.2, Is all I have to do in System->Administration->Network, uncheck the auto obtain IP with dhcp, check manually set IP addresses, and enter the 192,168.1.2?

[jganci]

Hi Jganci.

I have just changed my network config to get its ip info via dhcp and guess what, it didn't work. Looks like we have yet another issue to resolve.
Try setting the network address manually and boot with nohz=off.

I set the static IP address pretty much how I stated in my last post.

I was successful with the static IP address. Both a test fedora 8 latest kernel and fedora 9 original kernel. Successfully ran yum update on fedora 9. Successfully installed kmod-nvidia on both test fedora 8 and updated fedora 9, Now I need to do this for my main fedora 8. I've really missed the kmod-nvidia!

What does the nohz=off do?

Any idea when I/we can remove it and go back to using DHCP?

Thanks for your help. I'll continue to monitor this thread.

[jganci]

I set the static IP address pretty much how I stated in my last post.

I was successful with the static IP address. Both a test fedora 8 latest kernel and fedora 9 original kernel. Successfully ran yum update on fedora 9. Successfully installed kmod-nvidia on both test fedora 8 and updated fedora 9, Now I need to do this for my main fedora 8. I've really missed the kmod-nvidia!

What does the nohz=off do?

Any idea when I/we can remove it and go back to using DHCP?

Thanks for your help. I'll continue to monitor this thread.

Update follows. Summary: nohz=off doesn't seem to matter for my PC.

I decided to try booting without specifiying nohz=off, still using a static IP address. No problems encountered!? Did this for fedora 9 kernel 2.6.25.3-18.fc9.i686, fedora 8 with latest kernel (2.6.24.7-92.fc8), and fedora 8 with my last working kernel, namely 2.6.23.15-137.fc8. For my PC at least, it appears that the "connecting to eth0" problem is within DHCP code. My PC is an AMD Athlon with MSI K8T NEO 2 motherboard, built in ethernet RTL-8169. I'm running 32-bit fedora, not 64-bit fedora.

[barf]

Hi, sorry for not replying earlier, didn't get the notification e-mail till now. Glad to hear that your working now.
The Kernel update from 2.6.23 really stuffed up networking on a lot of systems many were quickly patched but it would seem that there are still some problems. I'll submit a bug report for this.

[jganci]

Thanks. Is there some way I can track it, or will there be updates in this thread?

[jganci]

Another piece of information.

My problem PC is my home PC. Here at work I have a spare PC, a Dell Precision laptop with a wired connection to our network running fedora 8 at the latest kernel. It has never had the problem I encountered at home, which is just as well, since I can't get a static IP address for that machine without a formal request. It's built-in network interface is 3c59x (that's what I see when I run lsmod). Isn't that 3COM? Whatever the problem is, mine seems to be related to the Realtek RTL-8169.

[barf]

Thanks. Is there some way I can track it, or will there be updates in this thread?

I've submitted as a bug. Bug number 449841. You can track the bug here https://bugzilla.redhat.com/

[jganci]

I just appled the latest fedora 8 updates. While I didn't have to reboot, I did anyway. Boot hung at "Starting sendmail". This time I had to hit the reset button. Rebooted fedora 9. Applied latest updates. Same thing. Beware!

I did manage to reboot fedora 8, entering interactive setup, and skipped sendmail startup.

FYI, I also had additional problems because of the order of events. My actual order, and additional problems follow.
1. Fedora 8. Apply updates.
2. System->Administration->Network to create name for my connection. Did this because of the change to static IP address. Okay so far.
3. System->Administration->Services to stop, restart network; however, nothing!? Huh? Tried it again; nothing.
4. Booted fedora 9. Applied updates.
5. Repeat 2 in fedora 9. Okay.
6. Repeat 3 in fedora 9. Same behavior.
7. Ran "system-config-network" from terminal window. Got error message something like "no graphical environment detected; no DISPLAY variable". That probably explains 3 and 6, too.
8. Rebooted fedora 8. Now the hang at sendmail startup. (Note: first time boot of fedora 8 after apply updates.)
9. Tried fedora 8 boot with nohz=off. Same hang.
10. Tried fedora 9 boot. Same hang.
11. Booted fedora 8; entered interactive setup; skipped sendmail startup.
12. System->Administration->Services okay now, as is System->Administration->Network. Go figure.
13. Update the thread you are now reading.

I do believe I'll stop the sendmail auto start, at least for a while.

[jganci]

I see you're online. If you have time, please read my latest post, #45.

[rclark]

I run my F9 box with no keyboard mouse or monitor. This ethernet problem was a pain as I depend on getting to the box via ssh and vnc if I need too. After upgrading I wasn't able to access the box... grrr... Anyway thanks for the following from a post above as it worked for me:

chkconfig NetworkManager off
chkconfig network on

Both of my ethernet ports now come up : one for Internet and other for local net.

[Alex-Grim]

This may help you with your wireless, in the future, as it does for many people: http://linuxintro.com/?action=ViewEntry&name=900072752405
Please do not mind the appearance at the moment, i am in the middle of remodeling right now. Also, the "Prev" and "Next" links have to be fixed, so to find the next article in the series, from the main menu, go to Distros - > Fedora and then click the next one.

Hope this helps you too :D

[srwarnernj]

I just fixed my laptop that had the same issue by adding the following command to my grub.conf:

irqpoll

Apparenetly, the irq for the network card was being shared with the video card and the network card was turned off!
Hope this helps someone,