Description of problem: After an update to initscripts-8.46-1 the following shows up in boot messages: Setting hostname localhost.localdomain: [ OK ] Missing major number for persistent device. Couldn't create ioctl argument. last_lba(): I don't know how to handle files with mode 41ed read error, sector 0 read error, sector 1 read error, sector 29 Missing major number for persistent device. Couldn't create ioctl argument. Reverting to initscripts-8.45.3-1 makes these messages to go away. Looking at differences in /etc/init.d/rc.sysinit from both versions the above is a result of difference in dmraid setup. Actually I am getting mesages like the above from running /sbin/kpartx -a -p p "/dev/mapper/$dmname" when "$dmname" is empty. If I am executing the whole 'if ... fi" corresponding block on an already running system then this works so that would mean that setting "$dmname" in a startup context fails. I am not sure right now why. OTOH there is this bit above the quoted line: [ -n "$dmname" ] && continue which does not seem to make a sense here. It looks to me that the condition is reversed and '-z' was meant instead of '-n'. Even if correct my test script that way then I am ending up running, when not in a startup, /sbin/kpartx -a -p p /dev/mapper/pdc_cjfeejidea (I have actually one disk with a raid signature) which is not causing any problems. With '[ -z "$dmname" ] && continue' in rc.sysinit I do not see qouted messages but if I am not running kpartx that is not so surprising. Version-Release number of selected component (if applicable): initscripts-8.46-1 How reproducible: always
If you change it back to dmraid from dmraid.static, does it start behaving again? (And if so, that's really wrong...)
> If you change it back to dmraid from dmraid.static, does it start > behaving again? If you can call it "behaving". :-) The catch is that /sbin/dmraid is using libz.so.1 from /usr; so if /usr is not available yet then "for x ..." loop below has nothing to do and /sbin/kpartx never runs with "/dev/mapper/" partition and the the problem is hidden. :-) If you try the later you will always get complaints like stated. As I can see it the only way to correct that is disallow "/dev/mapper/" partition, i.e reverse a test for 'continue'. Then I see no problems - static or no static. Sorry for a delay. You caught me with this question in a wrong moment.
This should be fixed in the current update initscripts.
This should be fixed in the current update initscripts, or rawhide. Please reopen if it's not.