ZFS hanging at kernel boot now, but didn't before... (Re: ZFS MFC heads up)

[Kip Macy]

I have no idea what is happening. I think our best bet is having
someone with insight into ATA provide us with help in adding
diagnostics.

Sorry for the trouble. Perhaps you can just roll back to 7.2 for now.

Cheers,
Kip


On Thu, May 21, 2009 at 10:50 AM, Joe Karthauser <joe at freebsd.org> wrote:
> Hmm, I've had a bit of a miserable afternoon trying to fight my RELENG_7
> server, which now doesn't boot. :(.
>
> So, it's a ZRAID2 pool with a ufs/gmirror root partition split over 5 disks
> (gmirror on 500Mb partition on each of five disks, and zraid2 over the rest
> of each drive).
>
> What I did was to update the userland, and then reboot. I didn't upgrade the
> kernel (but I've subsequently done that and have the same problem).
>
> What happens is that the kernel hangs booting just after displaying a LABEL
> message or ZFS pool/spool message. I _can_ get it to boot if I boot single
> user with acpi switched off. When I do that I can manually start zfs, and
> mount all the partitions. However, one of the disks is missing.... more on
> that next.
>
> The machine is running a gigabyte motherboard (domestic gamer P35 board,
> similar to this
> http://www.gigabyte.com.tw/Products/Motherboard/Products_Overview.aspx?ProductID=2533,
> although it might be a DS4 variant).  I've got 5 of the 6 sata ports wired
> to a 5 unit SATA hot swap bay (5 drives vertially mounted into 3 5-1/4" bays
> kind of thing).
>
> Now, because of the gmirror I can boot the system on any disk, or
> combination of plugged in disks. I should be able to succeed with the
> kernel probe up to the attempt to mount the root filesystem irrespective of
> any zfs pool, etc. And, indeed, this has been working fine for about two
> years.
>
> But, now it hangs in the same place no matter what disk I boot on (I've
> tried every bay).
>
> But, without ACPI enabled it does appear to boot ok... what's going on here?
> Is it possible that the machine has developed a hardware fault?
>
> Ok, finally, if I boot with ACPI disabled then one of the disks is missing.
> If I unplug it I get a disconnect message from the ata device, and a
> reconnect and reinit attempt when I plug it back in, but no device appears
> on the bus. Usually I can do a 'atacontrol detach sata4; sleep 1; atacontrol
> attach sata4' and the device reappears. This happens on the other buses, but
> not on the last one. It's not the disk, because if I swap it into another
> bay, it comes up and appears on the bus. On the other hand it doesn't appear
> to be that controller or slow in the drive bay because if I unplug all the
> over disks the system will boot that disk and get as far as the hang....
> hmm.
>
> Is this a consequence of disabling the ACPI?
>
> Does anyone have a clue what might be going on?
>
> Joe
> _______________________________________________
> freebsd-stable at freebsd.org mailing list
> http://lists.freebsd.org/mailman/listinfo/freebsd-stable
> To unsubscribe, send any mail to "freebsd-stable-unsubscribe at freebsd.org"
>



-- 
When bad men combine, the good must associate; else they will fall one
by one, an unpitied sacrifice in a contemptible struggle.

    Edmund Burke