Intrepid boot fails on 2.6.27 with software RAID1
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
linux (Ubuntu) |
Invalid
|
Undecided
|
Unassigned |
Bug Description
I can't boot into 2.6.27-2 or -3 with my software raid setup. It's an ASUS P5N-E SLI MoBo, and the disks look like this:
Disk /dev/sda: 500.1 GB, 500107862016 bytes
255 heads, 63 sectors/track, 60801 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x3a86cbf3
Device Boot Start End Blocks Id System
/dev/sda1 1 243 1951866 82 Linux swap / Solaris
/dev/sda2 244 30637 244139805 fd Linux RAID autodetect
/dev/sda3 * 30638 60800 242284297+ 7 HPFS/NTFS
Disk /dev/sdb: 500.1 GB, 500107862016 bytes
255 heads, 63 sectors/track, 60801 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Disk identifier: 0x3a86cbf3
Device Boot Start End Blocks Id System
/dev/sdb1 1 243 1951866 82 Linux swap / Solaris
/dev/sdb2 244 30637 244139805 fd Linux RAID autodetect
/dev/sdb3 * 30638 60800 242284297+ 7 HPFS/NTFS
However, on boot I'm dropped into BusyBox with messages:
ALERT! /dev/md0 does not exist
and
md0 : inactive dm3[0](S)
At this point I thought I'd try "mdadm --assemble --scan" which does, in fact, activate md0 - but only with one of the two raid partitions active. If I try to mdadm --add the other one, I get "mdadm: Cannot open /dev/sda2: Device or resource busy"
I've checked initrd and both md and raid1 modules are included.
Note: booting into 2.6.24-21 (and other earlier kernels) works absolutely fine, so that's what I'm running at the moment.
Booting into 2.6.27-3 recovery mode gives more details on the console. Both sda and sdb disks (and partitions thereof) are detected ok, but just prior to being dropped into BusyBox there's a load of messages like this:
md: md0 stopped
md: bind<dm-2>
md: md0 stopped
md: unbind<dm-2>
md: export_rdev(dm-2)
md: bind<dm-2>
md: md0 stopped
md: unbind<dm-2>
md: export_rdev(dm-2)
md: bind<dm-2>
md: md0 stopped
md: unbind<dm-2>
md: export_rdev(dm-2)
md: bind<dm-2>
Done.
** WARNING: There appears to be one or more degraded RAID devices **
(However, both drives are fine, as I can boot earlier kernels without problems).
Then, I'm offered an option to start the degraded device. If I choose "Y", md0 starts but with only one drive active (similar to the result I got from mdadm --assemble --scan)
Some kind of race condition on drive detection, maybe?
Update - 15th September
Ok - I think I may be getting to the bottom of this. I *think* kernel 2.6.27 is picking up a FakeRAID mirror set up on my NVIDIA chipset motherboard. This is a change in behavior from previous kernels which didn't detect the NVIDIA mirror and instead I could use software raid on the two softraid partitions, sda2 and sdb2. Now, however, the kernel is picking up a dm device and attempting to boot from that. Of course, as it's a RAID-1 mirror, the kernel sees it as only one device and complains as the original softraid mirror appears to have only one device rather than two.
Maybe if I disable fakeraid support until after boot then manually modprobe dm-mirror... ?
description: | updated |
i'm having similar trouble event though i'm not sure if this is related :
- i use a fakeraid 0+1 array (intel matrix)
- i installed intrepid beta using install guides to set it up on a pre-partitioned raid configuration
- my system and OS are 64-bit
my system fails to boot using a kernel that's newer than 2.6.27-5. On both 2.6.27-6 an -7 i don't get the (traditional?)
"kernel alive
Kernel really alive"
message.
And after a while i also get this "device does not exist" alert and busybox (initramfs) prompt.