geom mirror now rebuilding on every reboot?

[Garrett Cooper]

On Aug 6, 2012, at 9:40 AM, Maksim Yevmenkin <maksim.yevmenkin at gmail.com> wrote:

> On Mon, Aug 6, 2012 at 9:33 AM, Kevin Oberman <kob6558 at gmail.com> wrote:
>> On Mon, Aug 6, 2012 at 9:26 AM, Maksim Yevmenkin
>> <maksim.yevmenkin at gmail.com> wrote:
>>> On Mon, Aug 6, 2012 at 9:23 AM, Garrett Cooper <yanegomi at gmail.com> wrote:
>>>> On Mon, Aug 6, 2012 at 8:34 AM, Maksim Yevmenkin
>>>> <maksim.yevmenkin at gmail.com> wrote:
>>>>> Michael,
>>>>> 
>>>>>> Something in -current and recently MFC'd to -stable is causing all of my
>>>>>> gmirror drives to rebuild on reboot :-(
>>>>>> 
>>>>>> Being remote and these being production machines, I suspect SVN r237929
>>>>>> and r237930 in -current and SVN r238500 to -stable but haven't yet been
>>>>>> able to prove it.
>>>>>> 
>>>>>> Is anyone else seeing this?
>>>>> 
>>>>> yes, i've seem something similar only much, much worse. one of our
>>>>> production systems completely kept loosing its gmirror volumes on
>>>>> every reboot. it looked like gmirror metadata were completely
>>>>> corrupted. rebuilding mirrors and reverting back to previous kernel
>>>>> seemed to work. someone else is tracking it down.
>>>> 
>>>>    Is this is in 9.1 -PRERELEASE, -RELEASE (or whatever the official
>>>> label is...)? If so, it seems like this would be a ship blocker.
>>> 
>>> sorry. its releng_9/9-stable. gmirrors are on two ssds. we use gpt and
>>> gmirror individual partitions, not entire disks.
>> 
>> I may well be confused, but I don't understand how you can use GPT for
>> a single partition. Looks to me like a disk is GPT or legacy.
>> Declaring which is the first command when setting up a new disk.
> 
> i'm sorry, to make it (hopefully) clear, we are using gpt partitioning
> scheme (as opposed to mbr partitioning scheme), and then use gmirror
> on individual partitions, i.e. we gimirror /dev/ada0p1 and /dev/ada1p1
> and not /dev/ada0 and /dev/ada1.

Can you provide more helpful info like svn revisions?