Re: How to overrule bioctl "chunk already in use"

[Rachel Roch]

29 Mar 2019, 02:42 by n...@holland-consulting.net:

> On 3/28/19 10:29 AM, Rachel Roch wrote:
>
>> Hi,
>>
>> I've been following the instructions here
>> https://www.openbsd.org/faq/faq14.html 
>> 
>> <>> https://www.openbsd.org/faq/faq14.html 
>> >> > to setup softraid.
>>
>> Unfortunately I somehow messed up the original attempt through my own
>> stupidity.
>>
>
> it happens.
> And best that it happen before production than after.
>
>> So I've been trying to go through the steps again.  However nothing
>> I do can elminate the "softraid0 sd0a chunk already in use" message
>> at the "bioctl -c 1 -l sd0a,sd1a softraid0" step.
>>
>> I've tried everything !  Rebooting the server, /dev/zero to the
>> first 500MB of sd0 and sd1, changing uuid in disklabel, erasing and
>> re-writing disk label.
>>
>> I looked at the man page and thought "ah ha !" ... maybe "-C force",
>> but nope !
>>
>
> you were close with the zeroing the head of the components.  In fact,
> I'm not sure what you did wrong, but that's the solution.
>
> I'd suggest starting by zeroing the beginning of each physical disk --
> using the r device and the c partition -- i.e.,
>
>  # dd if=/dev/zero of=/dev/rsd0c
>  # dd if=/dev/zero of=/dev/rsd1c
>
> I've had enough problems, I really suggest this unless you are
> absolutely sure your disk has never even heard of OpenBSD before you
> install it. :)  (I think I had figured out at one point that zeroing the
> RAID partitions was sufficient, but when it comes to zeroing, a little
> more is never too much. :)
>
> Now, if you were going to script this, you would put a block size and a
> count in there...but since you are just typing this at the command line,
> count to three and hit CTRL-C then do the next.  You really only have to
> clear a megabyte or so, and probably a LOT less...you can't hit CTRL-C
> fast enough, I suspect. :)
>
> By using the 'r' device and the 'c' partion, you have wiped the very
> very start of the disk -- sector zero onward.
>
> I'd reboot after that.  I don't think it's needed, but either the
> disklabel or MBR partition can be held in memory and written back out to
> disk under some circumstance, I don't recall exactly what (probably
> having to do with mounted partitions), so a reboot, and then verifying
> that fdisk sd0 shows lots of zeros everywhere including the Signature.
> NOW fdisk, create your OpenBSD partition, then your RAID disklabel
> partitions, and you should be in business.
>
> If that doesn't do it, show us your exact commands and exact output you
> are seeing.
>
> Nick.
>

Thanks Nick !  Your suggestion did the trick.

Re: How to overrule bioctl "chunk already in use"

[Nick Holland]

On 3/28/19 10:29 AM, Rachel Roch wrote:
> Hi,
> 
> I've been following the instructions here
> https://www.openbsd.org/faq/faq14.html
>  to setup softraid.
> 
> Unfortunately I somehow messed up the original attempt through my own
> stupidity.

it happens.
And best that it happen before production than after.

> So I've been trying to go through the steps again.  However nothing
> I do can elminate the "softraid0 sd0a chunk already in use" message
> at the "bioctl -c 1 -l sd0a,sd1a softraid0" step.
> 
> I've tried everything !  Rebooting the server, /dev/zero to the
> first 500MB of sd0 and sd1, changing uuid in disklabel, erasing and
> re-writing disk label.
> 
> I looked at the man page and thought "ah ha !" ... maybe "-C force",
> but nope !

you were close with the zeroing the head of the components.  In fact,
I'm not sure what you did wrong, but that's the solution.

I'd suggest starting by zeroing the beginning of each physical disk --
using the r device and the c partition -- i.e.,

   # dd if=/dev/zero of=/dev/rsd0c
   # dd if=/dev/zero of=/dev/rsd1c

I've had enough problems, I really suggest this unless you are
absolutely sure your disk has never even heard of OpenBSD before you
install it. :)  (I think I had figured out at one point that zeroing the
RAID partitions was sufficient, but when it comes to zeroing, a little
more is never too much. :)

Now, if you were going to script this, you would put a block size and a
count in there...but since you are just typing this at the command line,
count to three and hit CTRL-C then do the next.  You really only have to
clear a megabyte or so, and probably a LOT less...you can't hit CTRL-C
fast enough, I suspect. :)

By using the 'r' device and the 'c' partion, you have wiped the very
very start of the disk -- sector zero onward.

I'd reboot after that.  I don't think it's needed, but either the
disklabel or MBR partition can be held in memory and written back out to
disk under some circumstance, I don't recall exactly what (probably
having to do with mounted partitions), so a reboot, and then verifying
that fdisk sd0 shows lots of zeros everywhere including the Signature.
NOW fdisk, create your OpenBSD partition, then your RAID disklabel
partitions, and you should be in business.

If that doesn't do it, show us your exact commands and exact output you
are seeing.

Nick.

How to overrule bioctl "chunk already in use"

[Rachel Roch]

Hi,

I've been following the instructions here 
https://www.openbsd.org/faq/faq14.html  
to setup softraid.

Unfortunately I somehow messed up the original attempt through my own stupidity.

So I've been trying to go through the steps again.  However nothing I do can 
elminate the "softraid0 sd0a chunk already in use" message at the "bioctl -c 1 
-l sd0a,sd1a softraid0" step.

I've tried everything !  Rebooting the server, /dev/zero to the first 500MB of 
sd0 and sd1, changing uuid in disklabel, erasing and re-writing disk label.

I looked at the man page and thought "ah ha !" ... maybe "-C force", but nope !

Any ideas 

Rachel