|
Bugzilla – Full Text Bug Listing |
| Summary: | in /etc/sysconfig/network the config ifcfg-eth0 is not converted in ifcfg-MAC | ||
|---|---|---|---|
| Product: | [openSUSE] SUSE LINUX 10.0 | Reporter: | Tim Ehlers <tim> |
| Component: | Network | Assignee: | Peter Poeml <poeml> |
| Status: | RESOLVED WONTFIX | QA Contact: | E-mail List <qa-bugs> |
| Severity: | Normal | ||
| Priority: | P5 - None | ||
| Version: | Beta 3 | ||
| Target Milestone: | --- | ||
| Hardware: | i386 | ||
| OS: | SUSE Other | ||
| Whiteboard: | |||
| Found By: | Other | Services Priority: | |
| Business Priority: | Blocker: | --- | |
| Marketing QA Status: | --- | IT Deployment: | --- |
|
Description
Tim Ehlers
2005-08-30 10:57:07 UTC
Update from what version? There are cases where this is not possible to do that automatically. For 10.0 you can keep ifcfg-ethN if you like. See my commnt in bug 114088. Peter, we still should convert them, even if we have now persistent names, because we don't know if names changed during update. At the moment I don't understand what the problem is; could you provide logs please? There are two log files in /etc/sysconfig/network/ after update. Ok, ist was an update from 9.0 to 10.0 beta3. The problem has to do with bug 114088. The update Skript created hwcfg-static-0 to save the current order of ethX-interfaces. But this file seems no longer to be used. So in combination with the error, that the config ifcfg-eth0 etc is not converted into ifcfg-eth-id-MAC it happend, that the previos order: eth0 -> 3com (3c59x) eth1, eth2(unused) -> dual-NIC Gigabit (e1000) changed after install-reboot into: eth0, eth1 -> e1000 eth2 -> 3c59x and previos config for eth0 is now set up for Gigabit-NIC. So Network was misconfigured after update and the computer wasn't reachable. cat update-log Updating etc/sysconfig/network Backup copies of the original ifcfg files are saved in /var/adm/backup/sysconfig-backup processing modules.conf... creating /etc/sysconfig/hardware/hwcfg-static-0 processing ifcfg-eth0... found matching entry in hardware database, but card is currently unavailable processing ifcfg-eth1... found matching entry in hardware database, but card is currently unavailable pass #1 done. pass #2 done. pass #3 done. pass #4 done. processing routes... pass #5 done. cat update-log.debug Mon Aug 29 18:04:49 CEST 2005 -rw-r--r-- 1 root root 141 Aug 22 19:24 ifcfg-lo -rw-r--r-- 1 root root 180 Dec 10 2003 ifcfg-eth0~ -rw-r--r-- 1 root root 180 Jan 16 2004 ifcfg-eth0 -rw-r--r-- 1 root root 183 Feb 11 2004 ifcfg-eth1 -rw-r--r-- 1 root root 5779 Aug 22 19:24 ifcfg.template valid files: ifcfg-eth0 ifcfg-eth1 -rw-r--r-- 1 root root 239 Aug 22 19:24 ifroute-lo valid files: routes ls: /sys/class/net/eth0/driver: No such file or directory lrwxrwxrwx 1 root root 0 Aug 29 18:04 /sys/class/net/eth0/device -> ../../../devices/pci0000:00/0000:00:1e.0/0000:02:08.0 ls: /sys/class/net/eth1/driver: No such file or directory lrwxrwxrwx 1 root root 0 Aug 29 18:04 /sys/class/net/eth1/device -> ../../../devices/pci0000:00/0000:00:1e.0/0000:02:08.1 count of plain eth interfaces found: 2 found ../../modules.conf configured network driver modules: >>> begin modconf.Vx7861 alias eth0 3c59x alias eth1 e1000 <<< searching module options for 3c59x searching module options for e1000 >>> begin modconf.Vx7861 alias eth0 3c59x alias eth1 e1000 <<< found 2 aliases found 2 ethernet aliases found 0 tr aliases creating /etc/sysconfig/hardware/hwcfg-static-0 >>> begin ../hardware/hwcfg-static-0 # This file has been created during sysconfig update for reasons of backward # compatibility. It is not necessary in the new sysconfig concept, and could # theoretically be deleted. # # It ensures that the drivers below are loaded in the same order as # before the system update, to make sure that the network interface names # are not changed on your system. STARTMODE='auto' MODULE_1='3c59x' MODULE_OPTIONS_1='' MODULE_2='e1000' MODULE_OPTIONS_2='' <<< processing ifcfg-eth0... >>> begin ifcfg-eth0 BOOTPROTO='static' BROADCAST='134.76.7.255' IPADDR='134.76.7.101' MTU='' NETMASK='255.255.254.0' NETWORK='134.76.6.0' REMOTE_IPADDR='' STARTMODE='onboot' UNIQUE='6Ogo.rbKoCKmo5xF' # Note: a fixed alias name association has been found during update in # your modules.conf file. To make sure that the interface names are # the same as before the update, a file # /etc/sysconfig/hardware/hwcfg-static-0 # has been created, which loads the network modules in a fixed order. # It is recommended to use the new ifcfg-eth-id-00:d0:b7:b2:9c:a3 style # configuration names and remove the file hwcfg-static-0 file. <<< wireless=false, interface_type=eth found unique string: 6Ogo.rbKoCKmo5xF + 169: local i + 170: for i in '$*' + 171: test -e ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF + 172: echo + 173: echo '>>> begin ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF' >>> begin ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF + 174: sed 's/^/ /' ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF [General] UniqueID=6Ogo.rbKoCKmo5xF ParentID=6NW+.eDbWAYN5SdB HWClass=network Model=Dell 3c905C-TX/TX-M [Tornado] [Status] Configured=yes Available=no Needed=no Active=unknown [Hardware] Bus=0x4 Slot=0x20c BaseClass=0x2 VendorID=10b7 DeviceID=9200 SubVendorID=1028 SubDeviceID=010d RevisionID=0x78 DeviceName=3c905C-TX/TX-M [Tornado] VendorName=3Com Corporation SubVendorName=Dell Computer Corporation HWClassList=000002200000 Res.IO=0xdc80,0x80,1,3 Res.Memory=0xff6ffc00,0x0,1,3,1 Res.Memory=0xff700000,0x0,0,1,0 Res.Interrupts=18,1494,1 + 175: echo '<<<' <<< + 176: echo processing ifcfg-eth1... >>> begin ifcfg-eth1 BOOTPROTO='static' BROADCAST='192.168.0.255' IPADDR='192.168.0.101' MTU='' NETMASK='255.255.254.0' NETWORK='192.168.0.0' REMOTE_IPADDR='' STARTMODE='onboot' UNIQUE='6Ogo.rbKoCKmo5xF' # Note: a fixed alias name association has been found during update in # your modules.conf file. To make sure that the interface names are # the same as before the update, a file # /etc/sysconfig/hardware/hwcfg-static-0 # has been created, which loads the network modules in a fixed order. # It is recommended to use the new ifcfg-eth-id-00:d0:b7:b2:9c:a3 style # configuration names and remove the file hwcfg-static-0 file. <<< wireless=false, interface_type=eth found unique string: 6Ogo.rbKoCKmo5xF + 169: local i + 170: for i in '$*' + 171: test -e ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF + 172: echo + 173: echo '>>> begin ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF' >>> begin ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF + 174: sed 's/^/ /' ../../../var/lib/hardware/unique-keys/6Ogo.rbKoCKmo5xF [General] UniqueID=6Ogo.rbKoCKmo5xF ParentID=6NW+.eDbWAYN5SdB HWClass=network Model=Dell 3c905C-TX/TX-M [Tornado] [Status] Configured=yes Available=no Needed=no Active=unknown [Hardware] Bus=0x4 Slot=0x20c BaseClass=0x2 VendorID=10b7 DeviceID=9200 SubVendorID=1028 SubDeviceID=010d RevisionID=0x78 DeviceName=3c905C-TX/TX-M [Tornado] VendorName=3Com Corporation SubVendorName=Dell Computer Corporation HWClassList=000002200000 Res.IO=0xdc80,0x80,1,3 Res.Memory=0xff6ffc00,0x0,1,3,1 Res.Memory=0xff700000,0x0,0,1,0 Res.Interrupts=18,1494,1 + 175: echo '<<<' <<< + 176: echo pass #1 done. pass #2 done. pass #3 done. pass #4 done. processing routes... >>> begin routes default 134.76.7.254 - - <<< + 63: test -x /usr/bin/diff + 64: /usr/bin/diff -u /var/adm/backup/sysconfig-backup/routes ifcfg.tO8025 pass #5 done. Hrm, I overlooked this bug. Tim, thanks for the information. FYI: after providing the needed information you can set the bug back to 'assigned', so it doesn't stay in 'needinfo'. Hm, what happens? A description file about each hardware device is stored in /var/lib/hardware/unique-keys. There are respective files for the NICs as well. However, both files show that the devices where "Unavailable" during the last scan of hwinfo/hwbootscan. The update script does look for that in the function get_data_from_uuid(), unavailable devices are not handled in the way that "available" devices are. Unavailable devices were seen as no longer existing iirc. Consequently, the devpath lookup for the MAC ID is not done for them. When the system is configured with yast, without manual intervention, this doesn't happen to my knowledge. I'm sorry, but there is not much that we can do about such special cases I think. (I am one of the people myself who edit ifcfg files by hand, so that the unique keys tend to become wrong, so I know those problems) There is a suspicious error message in the log... that's because the update script does: ls -l /sys/class/net/$dev/device /sys/class/net/$dev/driver 2>&1 | debuglog This path is no longer valid, it seems to have changed to /sys/class/net/$dev/device/driver with current kernels. Fixed that in svnBut this info is not used anyway. (Could it be used?) So this is going to be a wontfix. |