On Sun, 2013-04-28 at 15:49 +0100, peter green wrote: > I decided to upgrade a machine to a 3.8 kernel. At the time the machine > was running a mostly squeeze system. The machine failed to mount the > root filesystem. > > At first I thought this was related to the use of uuid for the root > filesystem, so I changed that to directly specifying /dev/md0 and it > made no difference. I also tried a rootdelay but that also made no > difference. Everything seems to happen long before the system drops me > at a shell. > > I then tried upgrading the system to wheezy, this installed a wheezy 3.2 > kernel which booted fine, however the experimental 3.8 kernel still > fails to boot. > > Begin: Loading essential drivers ... done. > Begin: Running /scripts/init-premoun[ 3.531519] device-mapper: > uevent: version 1.0.3 > t ... done. > Beg[ 3.536742] device-mapper: ioctl: 4.23.1-ioctl (2012-12-18) > initialised: dm-devel@redhat.com > in: Mounting root file system ... Begin: Running /scripts/local-top ... > Begin: Assembling all MD arrays ... mdadm: No devices listed in conf > file were found. > Failure: failed to assemble all arrays. [...] scsi_wait_scan no longer exists, so rootdelay is the only thing to stop mdadm and other such hooks from running too early. rootdelay really is the only way to work around this at present, and you'll have to increase it until it works. It looks like about 5 seconds should do, though the late messages from uhci_hcd are odd. I think we're going to have to reintroduce scsi_wait_scan in the kernel as a temporary measure, but this is really unsustainable. mdadm (and lvm2) should be triggered by udev hooks, and initramfs-tools needs to support this rather than trying force a synchronous process. Ben. -- Ben Hutchings Sturgeon's Law: Ninety percent of everything is crap.
Attachment:
signature.asc
Description: This is a digitally signed message part