Hello, "/dev/hda1 does not exist" is a wide field but since I encountered the same symptoms and worked around them, I can hopefully narrow it down a bit.
I encountered the problem on two machines out of eight after upgrading (2.6.18-4 worked fine on both). One is an old P2-300 box with soft RAID over two IDE drives, the other a newer Xeon server with soft RAID over two SCSI drives on a Fusion-MPT controller. Relevant packages (identical on both machines) are Package: linux-image-2.6.21-2-686 Version: 2.6.21-6 Package: initramfs-tools Version: 0.89 Ok, this is what can be seen on the console: The hardware drivers are loaded and report their findings (chipsets etc.). Then the md driver attempts to assemble its RAID, and fails because it cannot find disk devices. Then the root device (which should have been md0) is not found, and the rootfs mount fails. Then the disk drivers happily announce what disks and partitions they have found. Then the boot aborts and drops to a shell. Notice the order? What looks like a race condition is easily confirmed by using the "rootdelay=nsecs" parameter of initramfs-tools. Give the drivers a few seconds to finish their initialization, and the boot process continues just fine. Stefan, it might be helpful if you tried setting a rootdelay and reported if that works around the problem for you. Now why there is this race with .21 that was not there with .18, I leave for the kernel-knowledgable people to find out. I'm not even sure it's a bug after all but it is definitely a feature that needs to be addressed, or we will be seing a lot of non-booting machines (worst case, randomly non-booting). Greetings, Ralph -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
