Re: is vinum in current working for anyone

[Format recovered--see http://www.lemis.com/email/email-format.html]

On Saturday, 10 January 2004 at 23:25:16 -0500, Tim Middleton wrote:
> On Saturday 10 January 2004 06:00, Greg 'groggy' Lehey wrote:
>>> it's an interesting coincidence. And needless to say... i've removed
>>> start_vinum="YES" from rc.conf just in case.
>
> Greg, further to the email i sent you earlier, where i had resetconfig and
> re-created the raid5 config, I noticed a bit later the 3rd subdisk had gone
> away again. So this time I resetconfig, and did a clean init...

Hmm.

> By merely doing vinum start and stop a number of times you will see
> below I can lose the 3rd subdisk. Below is a transcript of all
> commands issued. They are all simply "vinum start", "vinum stop",
> and "vinum l". There's some time gaps as i was distracted a few
> times. However, nothing else touched the vinum disk in between
> times, and the disk was never even mounted. You'll see in the first
> "vinum l" i was just completing the "init" (97% on the slowest
> subdisk).

There's certainly one thing which jumps out:

> Jan 10 22:17:07 bee kernel: vinum: CONFIGURATION OBLITERATED
> Jan 10 22:17:16 bee kernel: vinum: exiting with malloc table inconsistency at 0xc6c14800 from vinumio.c:868

I haven't seen anything like that for a long time.  I should go
looking for what happened.

> Jan 10 22:29:09 bee kernel: vinum: raid5.p0.s1 is up
> Jan 10 23:02:32 bee kernel: vinum: exiting with malloc table inconsistency at 0xc6aff800 from vinumio.c:868

There are several more instances.  It's a pity you didn't include the
Vinum history file.

Unfortunately, I'm currently involved in running an important
conference, due to start in about 13 hours, and I'll be involved for a
week.  During that time, the chances of me even looking at Vinum bugs
are minimal.  But that's a smoking gun, and I'll certainly look at it
when I come back to normal.

Greg
--
When replying to this message, please take care not to mutilate the
original text.
For more information, see http://www.lemis.com/email.html
See complete headers for address and phone numbers.