Red Hat Bugzilla – Bug 866584
udev properties are only shown after re-triggering
Last modified: 2016-04-26 09:32:17 EDT
Description of problem:
Udev properties are used to e.g. determin the model or vendor of a NIC. After boot, not all udev properties are determined properly.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Clean install
2. No nic configuration
3. Dro to shell and run:
4. udevadm info --query=all --path/sys/class/net/eth0
Some properties are shown, but nothing like e.g. VENDOR_INFO_FROM_DATABASE (or what it is called)
More properties are shown, including e.g. VENDOR_INFO_FROM_DATABASE (or what it is called)
An initial udevadm info after a boot without configured nics:
[root@hp-xw4550-02 admin]# udevadm --debug info --query=all --path=/sys/class/net/eth3
run_command: calling: info
udev_device_new_from_syspath: device 0x7f6f4fc5a6d0 has devpath '/devices/pci0000:00/0000:00:14.4/0000:07:04.0/net/eth3'
Thus properties are not loaded.
udev can be forced to load the properties by running:
# udevadm test /sys/class/net/eth3
The properties will now be available.
As mentioned on IRC, this setup renames the device while it runs, and it seemed
to be done several times in one and the same event. Also the rules change
during bootup by over-mounting them. Something along all these rather "exotic" setup steps causes some hickup. Sorry, I have no idea what goes wrong here.
It seems, the simplest workaround would be to 'trigger' all network interfaces
again after bootup, and 'settle' again before anything tries to access any udev
device data that is in the process of being mangled that way.
(In reply to comment #9)
> As mentioned on IRC, this setup renames the device while it runs, and it
> to be done several times in one and the same event. Also the rules change
> during bootup by over-mounting them. Something along all these rather
> "exotic" setup steps causes some hickup. Sorry, I have no idea what goes
> wrong here.
Just a note here: I could also reproduce this on a default RHEL6 machine (see comment above) which hasn't got this exotic setup.
The base issue in udev can't/won't be solved in RHEL 6. We handle this issue with the fix in bug 866585
*** This bug has been marked as a duplicate of bug 866585 ***