initrd is too fast for md
I recently installed testing on a Sun V240 but couldn't set up
md/lvm during the install. I tried to submit an insallation
report, but it doesn't seem to have made it yet. I'll try and
figure out what happened to it.
Anyways, I've been trying to switch the system over to lvm on
raid1, but I'm not sure if I'm running into a bug or just missing
some minor detail or otherwise doing something dumb. The problem
is that /scripts/local-top is running and trynig to set up md
before the hard drives are detected and show up in /dev. After a
while it gives up on waiting for the root filesystem and drops to
a shell.
An earlier problem I had was that I had created md0 with /
mounted ro, so /etc/mdadm/mdadm.conf wasn't updated and when
generating the initrd, /conf/conf.d/md had MD_MODULES=''. That
took me a while to figure out, because I only needed to load
md-mod to get raid to work. It seems like it might be a good idea
to just modprobe md-mod near the top of /scripts/local-top/mdadm.
Last of all, I seem to remember some plan to have things like md
and lvm started by udev which seems like it would have avoided
this problem. Also, since when does initrd use udev? It seems
like a static /dev might avoid this probblem, at least if
something caused mdadm to wait for /dev/sda to be ready rather
than giving up because the device file didn't even exist.
So, am I doing something silly or is this a bug?
Please CC me as I'm not on debian-boot.
Thanks,
Ivan
Here are the relevant bits of output from the boot process:
[ 58.604298] SCSI subsystem initialized
[ 58.621553] sym0: <1010-66> rev 0x1 at pci 0002:00:02.0 irq 27
[ 58.623680] sym0: No NVRAM, ID 7, Fast-80, LVD, parity checking
[ 58.669694] tg3.c:v3.102 (September 1, 2009)
[ 58.676561] sym0: SCSI BUS has been reset.
[ 58.676579] scsi0 : sym-2.2.3
[ 58.677668] sym1: <1010-66> rev 0x1 at pci 0002:00:02.1 irq 28
[ 58.679907] sym1: No NVRAM, ID 7, Fast-80, LVD, parity checking
[ 58.720370] sym1: SCSI BUS has been reset.
[ 58.720383] scsi1 : sym-2.2.3
[ 58.729629] alim15x3 0001:00:0d.0: IDE controller (0x10b9:0x5229 rev 0xc4)
<skip a bunch of USB and Ethernet stuff>
[ 67.284423] usbcore: registered new interface driver usb-storage
[ 67.363345] USB Mass Storage support registered.
Begin: Loading essential drivers ... done.
Begi[ 67.467158] md: raid1 personality registered for level 1
n: Running /scripts/init-premount ... done.
Beg[ 67.581332] device-mapper: uevent: version 1.0.3
in: Mounting roo[ 67.660712] device-mapper: ioctl: 4.15.0-ioctl (2009-04-01) initialised: dm-devel@redhat.com
t file system ... Begin: Running /scripts/local-top ... Begin: LoadingSuccess: loaded module raid1.
done.
Begin: Assembling all MD arrays ... mdadm: No devices listed in conf file were found.
Failure: failed to assemble all arrays.
done.
done.
Begin: Waiting for root file system ... [ 68.384600] sd 0:0:0:0: [sda] 143374741 512-byte logical blocks: (73.4 GB/68.3 GiB)
[ 68.486672] sd 0:0:0:0: [sda] Write Protect is off
[ 68.551468] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, supports DPO and FUA
[ 68.667782] sda: sda1 sda2 sda3
[ 68.723586] sd 0:0:0:0: [sda] Attached SCSI disk
[ 72.285213] scsi 2:0:0:0: Direct-Access Seagate FreeAgent Pro 400A PQ: 0 ANSI: 4
[ 80.748636] sd 2:0:0:0: [sdb] 1465149168 512-byte logical blocks: (750 GB/698 GiB)
[ 80.857870] sd 2:0:0:0: [sdb] Write Protect is off
[ 80.920868] sd 2:0:0:0: [sdb] Assuming drive cache: write through
[ 81.028609] sd 2:0:0:0: [sdb] Assuming drive cache: write through
[ 81.108731] sdb: sdb1 sdb2 sdb3 sdb4
[ 81.269614] sd 2:0:0:0: [sdb] Assuming drive cache: write through
[ 81.349746] sd 2:0:0:0: [sdb] Attached SCSI disk
done.
Gave up waiting for root device. Common problems:
Note that this was on a serial console, and it seems kernel
messages get priority over userland ones. They also all start
with [ and end with a newline, so it's not too confusing.
Thanks again,
Ivan
Reply to: