latest udev (0.114-2) problem
I have an Athlon k7 box, running testing, as my gateway. Due to a new system install, I had not upgraded in a few weeks. On the 21st I decided that it was time and did the upgrade overnight. This morning I rebooted the box to upgrade my sid partition and when I rebooted testing, I had lost my my localnet connections. Dmesg shows that the new udev package had decided to change eth0 to eth1, for some reason. To use the localnet, I changed eth0 to eth1 in /etc/network/interfaces, restarted networking and killed the firewall. My localnet was back. I then rebooted and found that udev had now changed the interface to eth3. This will be a PITA to users that don't run their boxes 24/7. I have not seen this posted yet so it may be something wrong here. Has anyone else see this udev behavior? Wayne -- In English, every word can be verbed. Would that it were so in our programming languages. ___ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: latest udev (0.114-2) problem
Hi Wayne. Wayne Topa, 23.08.2007 18:21: I have an Athlon k7 box, running testing, as my gateway. Due to a new system install, I had not upgraded in a few weeks. On the 21st I decided that it was time and did the upgrade overnight. This morning I rebooted the box to upgrade my sid partition and when I rebooted testing, I had lost my my localnet connections. Dmesg shows that the new udev package had decided to change eth0 to eth1, for some reason. To use the localnet, I changed eth0 to eth1 in /etc/network/interfaces, restarted networking and killed the firewall. My localnet was back. I then rebooted and found that udev had now changed the interface to eth3. This will be a PITA to users that don't run their boxes 24/7. This definitely doesn’t sound like udev were working correctly. Is there a /etc/udev/rules.d/z25_persistent-net.rules file? This one should contain the mapping of the ethernet devices based on their MAC address. Manual reordering should be done here, the file usually never ever gets touched after initial generation. Regards, Mathias -- debian/rules signature.asc Description: OpenPGP digital signature
Re: latest udev (0.114-2) problem
Mathias Brodala([EMAIL PROTECTED]) is reported to have said: Hi Wayne. Wayne Topa, 23.08.2007 18:21: I have an Athlon k7 box, running testing, as my gateway. Due to a new system install, I had not upgraded in a few weeks. On the 21st I decided that it was time and did the upgrade overnight. This morning I rebooted the box to upgrade my sid partition and when I rebooted testing, I had lost my my localnet connections. Dmesg shows that the new udev package had decided to change eth0 to eth1, for some reason. To use the localnet, I changed eth0 to eth1 in /etc/network/interfaces, restarted networking and killed the firewall. My localnet was back. I then rebooted and found that udev had now changed the interface to eth3. This will be a PITA to users that don't run their boxes 24/7. Correction: udev stopped at eth2 NOT eth3. (I mistyped yet again) On a reboot it came up again as eth2. So it is not as bad as I first thought. This definitely doesn't sound like udev were working correctly. Is there a /etc/udev/rules.d/z25_persistent-net.rules file? This one should contain the mapping of the ethernet devices based on their MAC address. Manual reordering should be done here, the file usually never ever gets touched after initial generation. Yes there is. I've had problems with udev on systems that I was trying out various interface cards (usb, pcmcia) on, so thats the first file I checked. It is ignoring the, original, eth0 stansa that it was not having a problem with and adding new eth interfaces. Here it is, currently. - # PCI device 10ec:8139 (8139too) SUBSYSTEM==net, DRIVER==?*, SYSFS{address}==00:48:54:d1:35:1c, NAME=eth0 # USB device 2001:3c00 (rt2500usb) #SUBSYSTEM==net, DRIVERS==?*, ATTRS{address}==00:11:95:e6:19:12, ATTRS{type}==1, NAME=wlan0 # PCI device 0x10ec:0x8139 (8139too) #SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth1 # PCI device 0x10ec:0x8139 (8139too) SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth2 -- Prior to this udev version if I removed all the non existant interfaces from the /etc/udev/rules.d/z25_persistent-net.rules file, it would work as expected. Something changed in this version as that no longer works. I found a fix (?) with a google search but I can't try it until I finish downloading a new kernel. It was on an Ubuntu Forum so just in case I'll list it here. Not Tested Yet so take it with a grain of salt. - Possible Fix -- Create/Edit a file /etc/udev/rules.d/10-local.rules with the following content Code: KERNEL=eth*, SYSFS{address}==00:10:5a:33:44:55 NAME=eth0 KERNEL=eth*, SYSFS{address}==00:0c:44:55:66:77 NAME=eth1 Just change the mac addresses to your own values from ifconfig. Thanks for the reply! Wayne -- As far as we know, our computer has never had an undetected error. ___ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: latest udev (0.114-2) problem
Wayne Topa wrote: Mathias Brodala([EMAIL PROTECTED]) is reported to have said: Hi Wayne. Wayne Topa, 23.08.2007 18:21: I have an Athlon k7 box, running testing, as my gateway. Due to a new system install, I had not upgraded in a few weeks. On the 21st I decided that it was time and did the upgrade overnight. This morning I rebooted the box to upgrade my sid partition and when I rebooted testing, I had lost my my localnet connections. Dmesg shows that the new udev package had decided to change eth0 to eth1, for some reason. To use the localnet, I changed eth0 to eth1 in /etc/network/interfaces, restarted networking and killed the firewall. My localnet was back. I then rebooted and found that udev had now changed the interface to eth3. This will be a PITA to users that don't run their boxes 24/7. Correction: udev stopped at eth2 NOT eth3. (I mistyped yet again) On a reboot it came up again as eth2. So it is not as bad as I first thought. This definitely doesn't sound like udev were working correctly. Is there a /etc/udev/rules.d/z25_persistent-net.rules file? This one should contain the mapping of the ethernet devices based on their MAC address. Manual reordering should be done here, the file usually never ever gets touched after initial generation. Yes there is. I've had problems with udev on systems that I was trying out various interface cards (usb, pcmcia) on, so thats the first file I checked. It is ignoring the, original, eth0 stansa that it was not having a problem with and adding new eth interfaces. Here it is, currently. - # PCI device 10ec:8139 (8139too) SUBSYSTEM==net, DRIVER==?*, SYSFS{address}==00:48:54:d1:35:1c, NAME=eth0 # USB device 2001:3c00 (rt2500usb) #SUBSYSTEM==net, DRIVERS==?*, ATTRS{address}==00:11:95:e6:19:12, ATTRS{type}==1, NAME=wlan0 # PCI device 0x10ec:0x8139 (8139too) #SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth1 # PCI device 0x10ec:0x8139 (8139too) SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth2 -- Prior to this udev version if I removed all the non existant interfaces from the /etc/udev/rules.d/z25_persistent-net.rules file, it would work as expected. Something changed in this version as that no longer works. I found a fix (?) with a google search but I can't try it until I finish downloading a new kernel. It was on an Ubuntu Forum so just in case I'll list it here. Not Tested Yet so take it with a grain of salt. - Possible Fix -- Create/Edit a file /etc/udev/rules.d/10-local.rules with the following content Code: KERNEL=eth*, SYSFS{address}==00:10:5a:33:44:55 NAME=eth0 KERNEL=eth*, SYSFS{address}==00:0c:44:55:66:77 NAME=eth1 Just change the mac addresses to your own values from ifconfig. Thanks for the reply! Wayne Wayne It looks as though udev is now detecting the card differently as the PCI device now has a '0x' before 8139 so effectively the mapping for eth0 is for a non existent card. I would try deleting all the eth* entries and restart. HTH Wackojacko -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: latest udev (0.114-2) problem (SOLVED)
Wayne Topa([EMAIL PROTECTED]) is reported to have said: Mathias Brodala([EMAIL PROTECTED]) is reported to have said: Hi Wayne. Wayne Topa, 23.08.2007 18:21: I have an Athlon k7 box, running testing, as my gateway. Due to a new system install, I had not upgraded in a few weeks. On the 21st I decided that it was time and did the upgrade overnight. This morning I rebooted the box to upgrade my sid partition and when I rebooted testing, I had lost my my localnet connections. --snip-- I found a fix (?) with a google search but I can't try it until I finish downloading a new kernel. It was on an Ubuntu Forum so just in case I'll list it here. Not Tested Yet so take it with a grain of salt. - Possible Fix -- Create/Edit a file /etc/udev/rules.d/10-local.rules with the following content Code: KERNEL=eth*, SYSFS{address}==00:10:5a:33:44:55 NAME=eth0 KERNEL=eth*, SYSFS{address}==00:0c:44:55:66:77 NAME=eth1 Just change the mac addresses to your own values from ifconfig. The fix did not work on this problem. I don't know enough about udev but I did think it looked like it would. The problem was not fixed by installing the 2.6.21-2 k7 kernel. The suggestion made by Wackojacko did work. I was able to boot with and without the above fix after I removed all rules in the /etc/udev/rules.d/z25_persistent-net.rules file. Commenting out the rules did not work, removing them did. I have to test if it still works with the old 2.6.18 kernel, but if your interfaces get changed with the new udev on the 2.6.21-2 k7 kernel, this should fix it. Thanks Wackojacko! WT -- Imagine if every Thursday your shoes exploded if you tied them the usual way. This happens to us all the time with computers, and nobody thinks of complaining. -- Jeff Raskin, interviewed in Doctor Dobb's Journal ___ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: latest udev (0.114-2) problem
Wackojacko([EMAIL PROTECTED]) is reported to have said: Wayne Topa wrote: Mathias Brodala([EMAIL PROTECTED]) is reported to have said: Hi Wayne. Wayne Topa, 23.08.2007 18:21: I have an Athlon k7 box, running testing, as my gateway. Due to a new system install, I had not upgraded in a few weeks. On the 21st I decided that it was time and did the upgrade overnight. This morning I rebooted the box to upgrade my sid partition and when I rebooted testing, I had lost my my localnet connections. --snip-- s Here it is, currently. - # PCI device 10ec:8139 (8139too) SUBSYSTEM==net, DRIVER==?*, SYSFS{address}==00:48:54:d1:35:1c, NAME=eth0 # USB device 2001:3c00 (rt2500usb) #SUBSYSTEM==net, DRIVERS==?*, ATTRS{address}==00:11:95:e6:19:12, ATTRS{type}==1, NAME=wlan0 # PCI device 0x10ec:0x8139 (8139too) #SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth1 # PCI device 0x10ec:0x8139 (8139too) SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth2 -- --snip-- It looks as though udev is now detecting the card differently as the PCI device now has a '0x' before 8139 so effectively the mapping for eth0 is for a non existent card. I would try deleting all the eth* entries and restart. The 0x was added by the new version, the non 0x was inserted by the previous version. I have just rebooted with the 2.6.21-2 kernel, on the k7, and it is still showing this odd stuff in dmesg. dmesg |grep eth eth0: RealTek RTL8139 at 0xdc00, 00:48:54:d1:35:1c, IRQ 11 eth0: Identified 8139 chip type 'RTL-8139C' udev: renamed network interface eth0 to eth2 eth2: link up, 10Mbps, half-duplex, lpa 0x eth2: no IPv6 routers present I will try your suggestion next. Thanks wt -- Daddy, what does FORMATTING DRIVE C mean? ___ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: latest udev (0.114-2) problem
Wackojacko([EMAIL PROTECTED]) is reported to have said: Wayne Topa wrote: Mathias Brodala([EMAIL PROTECTED]) is reported to have said: Hi Wayne. --snip-- - # PCI device 10ec:8139 (8139too) SUBSYSTEM==net, DRIVER==?*, SYSFS{address}==00:48:54:d1:35:1c, NAME=eth0 # USB device 2001:3c00 (rt2500usb) #SUBSYSTEM==net, DRIVERS==?*, ATTRS{address}==00:11:95:e6:19:12, ATTRS{type}==1, NAME=wlan0 # PCI device 0x10ec:0x8139 (8139too) #SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth1 # PCI device 0x10ec:0x8139 (8139too) SUBSYSTEM==net, DRIVERS==?*, ATTR{address}==00:48:54:d1:35:1c, NAME=eth2 --snip-- It looks as though udev is now detecting the card differently as the PCI device now has a '0x' before 8139 so effectively the mapping for eth0 is for a non existent card. I would try deleting all the eth* entries and restart. I forget to mention that I had commented all but the first eth out and had added the 0x in the commented line. ie # PCI device 10ec:8139 (8139too) - # PCI device 0x10ec:8139 (8139too) but it didn't work, but the other lines were just commented out, not removed as you suggested. Thanks again. Wayne -- Daddy, what does FORMATTING DRIVE C mean? ___ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: latest udev (0.114-2) problem (SOLVED)
Wayne Topa([EMAIL PROTECTED]) is reported to have said: Wayne Topa([EMAIL PROTECTED]) is reported to have said: Mathias Brodala([EMAIL PROTECTED]) is reported to have said: Hi Wayne. Wayne Topa, 23.08.2007 18:21: I have an Athlon k7 box, running testing, as my gateway. Due to a new system install, I had not upgraded in a few weeks. On the 21st I decided that it was time and did the upgrade overnight. This morning I rebooted the box to upgrade my sid partition and when I rebooted testing, I had lost my my localnet connections. --snip-- I found a fix (?) with a google search but I can't try it until I finish downloading a new kernel. It was on an Ubuntu Forum so just in case I'll list it here. Not Tested Yet so take it with a grain of salt. - Possible Fix -- Create/Edit a file /etc/udev/rules.d/10-local.rules with the following content Code: KERNEL=eth*, SYSFS{address}==00:10:5a:33:44:55 NAME=eth0 KERNEL=eth*, SYSFS{address}==00:0c:44:55:66:77 NAME=eth1 Just change the mac addresses to your own values from ifconfig. The fix did not work on this problem. I don't know enough about udev but I did think it looked like it would. The problem was not fixed by installing the 2.6.21-2 k7 kernel. The suggestion made by Wackojacko did work. I was able to boot with and without the above fix after I removed all rules in the /etc/udev/rules.d/z25_persistent-net.rules file. Commenting out the rules did not work, removing them did. I have to test if it still works with the old 2.6.18 kernel, but if your interfaces get changed with the new udev on the 2.6.21-2 k7 kernel, this should fix it. Confirmed. removing all of the entries in the /etc/udev/rules.d/z25_persistent-net.rules file works in the 2.6.18 kernels as well. The question is, why would commented rules get read by udev and why would it think a rule, made by a previous version of udev, wasn't correct? Well I guess thats what _testing_ is for. It sure didn't get caught in Sid. Thanks to all that replied! Wayne -- I am a computer, dumber than any human and smarter than any administrator. ___ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]