From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040217 Description of problem: I recently updated to fedora core development from the fc2t1 release. Since that upgrade my eth0 connection no-longer starts on boot. /etc/sysconfig/network-scripts/ifcfg-eth0 contains ONBOOT=yes and /etc/sysconfig/networking/profiles/default/ has a file called ifcfg-eth0. Version-Release number of selected component (if applicable): initscripts-7.46-1.1 How reproducible: Always Steps to Reproduce: 1. Boot the computer 2. Try to use the network Actual Results: eth0 didn't start Expected Results: eth0 should have started. Additional info:
Let me know any files which should be attached.
Do you have a firewire adapter?
Yes. Only used for my iPod though. There's also another old&unused network card in there.
Does 'rmmod eth1394 ; ifup eth0' work?
I've just arrived at work, I'll try that this evening (NZDT) when I get home.
ifup eth0 works irrespective of whether or not the eth1394 module is loaded (it isn't on boot). The only problem is that the ONBOOT=yes flag seems to be ignored.
Can you attach the full ifcfg-eth0?
Created attachment 98435 [details] ifcfg-eth0 Sure, Here's the file. Any others you'd like?
Are there any error messages?
There are two errors on boot. one is something about insmod failing for a toshiba acpi module or something. eth0 starts just fine according to the output. Then when sshd is starting there's an ifup error about sit0 not existing. Now, I just noticed that ifup eth0 didn't work this time, I needed to use system-config-network. I ran an lsmod before and after, here's the diff: Module Size Used by +8139too 22016 0 +mii 3712 1 8139too +r8169 9732 0 +crc32 3968 2 8139too,r8169 +sg 28064 0 +scsi_mod 101564 1 sg Also, when upgrading from fc1 to the test release, my onboard ethernet 'became' eth0. FC1 didn't even detect it, but FC2t1 did and replaced the 8139 card with it.
So, it *says* that it's bringing up eth0, but it's not actually up after the boot process finishes?
Yes, that's correct. I don't know if it's up halfway through or something, but the output clearly says: "Bringing up interface eth0 [ OK ]" ifup eth0 on the command line doesn't show any errors either, it just doesn't do anything.
that is. It doesn't do anything after the boot. Once I use system-config-network to activate the interface ifup and ifdown work as expected. I assume it's some kind of module related thing?
After booting, what does 'ifconfig eth0' say?
I'll test that when I get home from work. Any other files you think may help? modules.conf or whatever?
modprobe.conf, ifcfg-*, and /etc/sysconfig/hwconf
Created attachment 98450 [details] The requested files I'd never heard of /etc/modprobe.conf, I can't help but noticing that there's no alias eth0 8169
Added alias eth0 r8169 and it starts on boot now. However, for the sake of consistency, here is the ifconfig output you wanted. Kind of weird really. [root@gandalf root]# ifconfig eth0 eth0: error fetching interface information: Device not found [root@gandalf root]# ifconfig eth1 eth1: error fetching interface information: Device not found [root@gandalf root]# ifconfig eth0 eth0 Link encap:Ethernet HWaddr 00:05:1C:1A:1F:C2 inet addr:10.0.2.1 Bcast:10.0.2.255 Mask:255.255.255.0 inet6 addr: fe80::205:1cff:fe1a:1fc2/64 Scope:Link UP BROADCAST MULTICAST MTU:1500 Metric:1 RX packets:0 errors:0 dropped:0 overruns:0 frame:0 TX packets:6 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:0 (0.0 b) TX bytes:418 (418.0 b) Interrupt:11 Base address:0xa000
How did you upgrade?
Downloaded the ISOs, burned them. booted them and chose upgrade. The card it's marked as eth1 in the modprobe.conf used to be eth0, then upon installing FC2 it had finally detected my on board ethernet, but marked it as eth0. Maybe it was actually just an off by one error somewhere?
Did kudzu ever pop up any warning dialogs about hardware changing for you? (The problem is that the initial change to modprobe.conf probably happened long enough ago that's its going to be tricky to track down where it went wrong. :/ )
I don't think it ever came up with warnings. I haven't seen a kudzu screen in a while. Except when I had plugged in my ipod. Truth be told, I always choose ignore there anyways.
Given the quirky nature of this one how about this. I have a test PC with two NICs, I'll format it and install FC1, then update to the test release. Then yum update. If I can reproduce this I'll have all the information we need. Otherwise, I'll mark this as not a bug and assume that I screwed something up.
Didn't mean to do that. Javascript.
I have had a similar experience here. Upgraded from FC2 T2 to T3 yesterday using the iso's. Nothing errored out on upgrade. Then upon reboot, kernel messages shows the following: Apr 28 20:51:33 maxer kernel: tg3: Unknown parameter `irq' Apr 28 20:51:33 maxer kernel: ip_tables: (C) 2000-2002 Netfilter core team Apr 28 20:51:33 maxer kernel: ip_conntrack version 2.1 (8191 buckets, 65528 max) - 296 bytes per conntrack Apr 28 20:51:33 maxer kernel: tg3: Unknown parameter `irq' Apr 28 20:51:33 maxer kernel: tg3: Unknown parameter `irq' Apr 28 20:55:38 maxer kernel: tg3: Unknown parameter `irq' Apr 28 20:55:38 maxer modprobe: FATAL: Error inserting tg3 (/lib/modules/2.6.5-1.339custom/kernel/drivers/net/tg3.ko): Unknown symbol in module, or unknown parameter (see dmesg) Kudzu removes my NetXtreme BCM570X Gigabit Ethernet (tg3.ko), then re- adds it. I reconfigure for my ADSL connection and then eth0 still is unresponsive. The entry in hwconf : class: NETWORK bus: PCI detached: 0 device: eth0 driver: tg3 desc: "Broadcom Corporation|NetXtreme BCM5702X Gigabit Ethernet" vendorId: 14e4 deviceId: 16a6 subVendorId: 1043 subDeviceId: 80a9 pciType: 1 pcidom: 0 pcibus: 2 pcidev: 5 pcifn: 0
Resolved! Apparently the upgrade takes your modprobe.conf and puts an alias in for eth0 and sets the IRQ for my tg3 to 9. Found a backup that anaconda made back on March 7 and voila my network is just fine. GO FIGURE?
Raxet, care to explain the resolution in a bit more detail?
I went ... -> FC2pre1 -> FC2, and after the last stage I was left with these files in /etc: modprobe.conf.anacbak modprobe.conf.dist modules.conf modules.conf.DIST where modprobe.conf stands out by its absence. Is my failure the same as reporters?
No, but it *is* the same as #121737, which is marked as fixed.
Hi, I have a similar problem with Fedora Core 1. But in my case the line "Bringing up interface eth0" doesn't show up during the boot process. Everything was just fine before I ran a mass update using up2date. Several packages have been downloaded and installed; I cannot remember all the names. Some comments that might be usefull. 1. There's a line: alias eth0 8139too in /etc/modprobe.conf. Apparently the module is being loaded correctly. 2. There's no other file named "modprobe.***" besides ".conf" and ".conf.dist" in the /etc directory. 3. The command "ifup eth0" works perfectly after boot. Then I can access the internet, mount remote filesystems, and so on. Should I update to Fedora 2 now? Is there any log file that I should check to figure out what went wrong during last boot?
William: I doubt your problem is related at all. If I had to guess, perhaps the network driver broke in the update you did?
Ok, But if it is broken, how can I fix it? If it is really broken, shouldn't it prevent the interface of loading after the boot process? How can I make sure the driver is not loading since I see no error messages. The line "Bringing up interface eth0" simply doesn't appear. Sorry if this message is not directly related to this bug!
What does your ifcfg-eth0 look like?
Created attachment 100985 [details] network configuration file
I'm not experiencing this bug anymore. I'm runnig FC2 now. The upgrade corrected the problem, although some other bugs have shown up. :).
I happened to wander by this bug while searching for some other data and noticed an alarming number of people doing unsupported upgrades, and then being bitten by weird upgrade bugs. As a note to the many people doing unsupported upgrades: When you perform an upgrade, there are only a few routes that are valid for testing. Upgrades from a final version to a test version are OK. Upgrades from a test version to ANYTHING is NOT OK. If you want to use a test version, the only valid ways to test its upgrade is to upgrade from a FINAL version. So if you want to test FC3testX, you have to either install it fresh, or upgrade from FC2 FINAL. You can't upgrade from FC3test1 to FC3test2 and expect things to work right; that's not the way they're intended to work. If you want to test FC3test2, and you already have FC3test1 installed, you will need to return to a FC2 installation by starting fresh (probably a new install), and then upgrade straight to FC3test2. This was discussed many times on fedora-test-list, but in case folks missed it.... Obviously having a separate /home and backing up /etc from time to time is highly recommended. Good luck all!
While it's certainly true that the upgrades aren't supported or tested, Stating that they are *NOT OK* is a dangerous position. The anaconda upgrade process needs testing just like the rest of FC. Test n to n+1 upgrades work fine most of the time, and when they don't work it's something that should be reported. I know that such upgrades aren't supported, so when FC2 final was released I reinstalled.
Fedora Core 2 is now maintained by the Fedora Legacy project for security updates only. If this problem is a security issue, please reopen and reassign to the Fedora Legacy product. If it is not a security issue and hasn't been resolved in the current FC3 updates or in the FC4 test release, reopen and change the version to match.
Closing bugs on older, no longer supported, releases. Apologies for any lack of response. Please reopen if this persists on a current release, such as Fedora Core 3.