On 11/16/2006 19:29, Chacko Cherackal wrote: > I have a debian box with an IDE boot drive and 4 SATA drives used in a > software raid 5 configuration with mdadm. This was been working completely > fine. A few days ago I rebooted after a kernel update and now the > partitions on the drives aren't recognized. [...] > fdisk seems to indicate that the partition tables are still there, but I > can't work out why they are no longer recognized as /dev/sd[abcd]1. I guess most newer distributions are using udev (or possibly hotplug?) to dynamically create device nodes. Given that it is something tied pretty tightly to the kernel, is under very active development, and the fact that you just upgraded the kernel makes me suspect it's some problem with that. I'm not real familiar with udev, but, like most things, there are scripts that go along with the kernel software. > Is there any way to force the system to recognize the partitioned drives? Maybe try booting your old kernel as a stop-gap? You could probably create the needed device nodes in /dev by hand using the mknod command. man 1 mknod and See /Documentation/devices.txt in the kernel source tree for the major, minor number combos you need to create those. I wouldn't use this as a permanent solution; better fix the root cause. But this route may allow you to move data off in order to lower the heart rate... ~Jason --
More information about the ECLUG mailing list