Re: failing raid on shutdown

[Patrick Welche <prlw1%cam.ac.uk@localhost>]

On Fri, May 23, 2014 at 10:23:58PM +0100, Patrick Welche wrote:
> I updated an old box to -current/amd64 (6.99.40) about a fortnight
> ago, and noticed that its first root raid0 would fail with an IO
> error on shutdown. On startup, reconstructing it to itself seemed
> to work.  I just thought the disk was old enough, and should get
> a new one.
> 
> Today, on an up-to-date -current/amd64 box (recent hardware,
> 6.99.42), I installed a brand new disk, to mirror from another
> reasonably new 2T existing disk. Same again: the new disk's first
> root raid0 partition fails with an IO error on shutdown and the
> component needs rebuilding.

On the old box, I just tried:
- update kernel to just now's -current/amd64
- boot single user
- raidctl -R /dev/wd0a raid0
- fsck -f all the paritions on raid0 (all OK)
- reboot

(copied by hand:)
Failed to write RAID component info!
Failed to write RAID component info!
wd0: detached
uvm_fault(0xfffffe8005081730, 0x0, 1) -> e
fatal page fault in supervisor mode
trap type 6 code 0 rip ffffffff80587cee cs 8 rflags 10246 cr2 b8 ilevel 0 rsp 
fffffe80050c3b90
curlwp 0xfffffe807dde3a60 pid 20.1 lowest kstack 0xfffffe80050c12c0
kernel: page fault trap, code=0
Stopped in pid 20.1 (reboot) at netbsd:wdstrategy+0x37: movq  b8(%rax),%r13
db{0}> 

Unfortunately I can't type anything in at the USB keyboard at this point...
(this one might well be "old disk" - just why is it behaving fine after
the reconstruction. The other box I find harder to explain)

Cheers,

Patrick