> Because the overly agresive scripts which tune the bootup time to be
> super fast don't bother to way for the SATA bus to stabilize. They have
> a check to see if the ROOT device exists (and they do sleep), but it's
> not working because my ROOT device is a RAID device... and mdadm will
> make /dev/md0 and /dev/md1 even if it can't find the sda/sdb constituents.
>
> Sloppy programming angers me.
Interestingly enough, I have a dell 2650 running edgy which doesn't
have this problem - I backported 2.6.20-13 to it awhile ago.
Today I tried that kernel on another 2650, and it wouldn't find the
LVM on boot because the scsi bus hadn't settled yet. I had to add a
sleep in there to get it to work.
Looking at the dmesg output, it looks like the one with the problem
takes longer to probe the IDE bus.
It looks like it already has a bug and a proposed solution:
https://bugs.launchpad.net/ubuntu/+source/initramfs-tools/+bug/79204
I haven't had a chance to test it yet.
Received on Tue Apr 17 15:47:07 2007
This archive was generated by hypermail 2.1.8 : Tue Apr 17 2007 - 15:47:19 CST