Bug 973179
| Summary: | biosdevname not applied during installation (after switchroot) for devices not used in initramfs (ie not having ifcfg file with DEVICE?) | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Fedora] Fedora | Reporter: | Radek Vykydal <rvykydal> | ||||
| Component: | anaconda | Assignee: | Anaconda Maintenance Team <anaconda-maint-list> | ||||
| Status: | CLOSED RAWHIDE | QA Contact: | Fedora Extras Quality Assurance <extras-qa> | ||||
| Severity: | unspecified | Docs Contact: | |||||
| Priority: | unspecified | ||||||
| Version: | 19 | CC: | anaconda-maint-list, dshea, g.kaviyarasu, jonathan, mkolman, mschmidt, samuel-rhbugs, sbueno, vanmeeuwen+fedora | ||||
| Target Milestone: | --- | ||||||
| Target Release: | --- | ||||||
| Hardware: | Unspecified | ||||||
| OS: | Unspecified | ||||||
| Whiteboard: | |||||||
| Fixed In Version: | Doc Type: | Bug Fix | |||||
| Doc Text: | Story Points: | --- | |||||
| Clone Of: | Environment: | ||||||
| Last Closed: | 2014-06-09 11:20:41 UTC | Type: | Bug | ||||
| Regression: | --- | Mount Type: | --- | ||||
| Documentation: | --- | CRM: | |||||
| Verified Versions: | Category: | --- | |||||
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |||||
| Cloudforms Team: | --- | Target Upstream Version: | |||||
| Embargoed: | |||||||
| Attachments: |
|
||||||
Michal would you have any ideas why 80-net-name-slot.rules and not 71-biosdevname.rules is used after switch root in installer? Or debugging tips? This got complicated (currently not reproducible in F20/master) by adding new systemd renaming in initramfs (bug 972662) which doesn't seem to quite work - bug 988766 that should be fixed first. (In reply to Radek Vykydal from comment #2) > This got complicated (currently not reproducible in F20/master) by adding > new systemd renaming in initramfs (bug 972662) which doesn't seem to quite > work - bug 988766 that should be fixed first. The 988766 seem to be fixed and I am no longer to reproduce the issue of this bz. Seems to be fixed as well. (In reply to Radek Vykydal from comment #1) > Michal would you have any ideas why 80-net-name-slot.rules and not > 71-biosdevname.rules is used after switch root in installer? Or debugging > tips? I have no idea. As for tips on debugging udev rules, all I know is the kernel command line options documented in man systemd-udevd, and udevadm info. |
Created attachment 759604 [details] journal wit log-priority=debug Netinstall, 3 network devices, only one (em1) has link. From journal: initramfs Jun 10 12:24:31 localhost systemd-udevd[315]: renamed netif to 'em1' Jun 10 12:24:31 localhost systemd-udevd[315]: renamed network interface eth0 to em1 Jun 10 12:24:31 localhost systemd-udevd[310]: renamed netif to 'p3p1' Jun 10 12:24:31 localhost systemd-udevd[310]: renamed network interface eth0 to p3p1 Jun 10 12:24:43 localhost systemd-udevd[373]: renamed network interface eth1 to p3p2 Jun 10 12:24:43 localhost systemd-udevd[373]: renamed netif to 'p3p2' switchroot Jun 10 12:24:59 gasoline.anaconda.englab.brq.redhat.com englab.brq.redhat.com br systemd-udevd[738]: '/lib/udev/rename_device' [754] exit with return code 0 Jun 10 12:24:59 gasoline.anaconda.englab.brq.redhat.com englab.brq.redhat.com br systemd-udevd[739]: '/lib/udev/rename_device'(out) 'em1' Jun 10 12:24:59 gasoline.anaconda.englab.brq.redhat.com englab.brq.redhat.com br systemd-udevd[739]: '/lib/udev/rename_device' [758] exit with return code 0 Jun 10 12:24:59 gasoline.anaconda.englab.brq.redhat.com englab.brq.redhat.com br systemd-udevd[736]: '/lib/udev/rename_device' [753] exit with return code 0 Jun 10 12:24:59 gasoline.anaconda.englab.brq.redhat.com englab.brq.redhat.com br systemd-udevd[738]: renamed network interface p3p2 to ens3f1 Jun 10 12:24:59 gasoline.anaconda.englab.brq.redhat.com englab.brq.redhat.com br systemd-udevd[736]: renamed network interface p3p1 to ens3f0 One of the devices - em1/enp63s0 is activated in initramfs and therefore ifcfg file (ifcfg-em1 containing DEVICE=em1) is created. ifcfg-em1: # Generated by dracut initrd DEVICE="em1" ONBOOT=yes NETBOOT=yes UUID="c032d129-1904-4df0-8bfe-10d3b1c1b543" BOOTPROTO=dhcp HWADDR="f4:ce:46:2c:44:7a" TYPE=Ethernet NAME="em1" For the two remaining - p3p1/ens3f0, p3p2/ens3f1 - ifcfg files (ifcfg-ens3fX, with NO DEVICE= line) are created after switch root in anaconda (the devices are not named by biosdevname) ifcfg-ens3f1: TYPE=Ethernet BOOTPROTO=dhcp DEFROUTE=yes IPV4_FAILURE_FATAL=no IPV6INIT=yes IPV6_AUTOCONF=yes IPV6_DEFROUTE=yes IPV6_PEERDNS=yes IPV6_PEERROUTES=yes IPV6_FAILURE_FATAL=no NAME=ens3f0 UUID=47b9e38e-726c-45fe-b0b9-d30bad0af4d4 ONBOOT=no HWADDR=00:15:17:96:75:0A PEERDNS=yes PEERROUTES=yes (I will attach journalctl and output of udevadm test for network devices.) So after switchroot, in anaconda, devcices are named em1, ens3f1, ens3f0. After reboot, the devices would be renamed using biosdevname (em1, p3p1, p3p2). Seems that for the em1 case only initscript's 60-net.rules (renaming device to DEVICE from ifcfg) prevents changing device name to enp63c0 later (by 80-net-name-slot.rules), which is not the case for the other two devices. Without creating ifcfg-em1 in initramfs em1 would be ranamed too during installation after switchroot: Jun 10 13:05:21 localhost systemd-udevd[322]: renamed netif to 'em1' Jun 10 13:05:21 localhost systemd-udevd[322]: renamed network interface eth0 to em1 Jun 10 13:05:21 localhost systemd-udevd[322]: renamed netif to 'p3p1' Jun 10 13:05:21 localhost systemd-udevd[322]: renamed network interface eth0 to p3p1 Jun 10 13:05:34 localhost systemd-udevd[378]: renamed netif to 'p3p2' Jun 10 13:05:34 localhost systemd-udevd[378]: renamed network interface eth1 to p3p2 Jun 10 13:06:03 localhost systemd-udevd[661]: PROGRAM '/lib/udev/rename_device' /usr/lib/udev/rules.d/60-net.rules:1 Jun 10 13:06:05 localhost systemd-udevd[673]: renamed network interface em1 to enp63s0 Jun 10 13:06:07 localhost systemd-udevd[661]: renamed network interface p3p2 to ens3f1 Jun 10 13:06:07 localhost systemd-udevd[680]: renamed network interface p3p1 to ens3f0 So if we don't take into account 60-net.rules (ie when ifcfg file with DEVICE=em1 is not created in intramfs), it seems that after switchroot 80-net-name-slot.rules is triggered and it renames devices to en* which would (?) be prevented if the devices had been named by 71-biosdevname (NAME!="", GOTO="net_name_slot_end")?