gconcat question SOLVED

[Roger Olofsson]

<-----Ursprungligt Meddelande-----> 
>From: John Nielsen [lists at jnielsen.net]
>Sent: 19/10/2008 5:52:51 PM
>To: raggen at passagen.se
>Cc: freebsd-questions at freebsd.org
>Subject: Re: gconcat question
>
>On Sunday 19 October 2008, Roger Olofsson wrote:
>> <-----Ursprungligt Meddelande----->
>>
>> >From: John Nielsen [lists at jnielsen.net]
>> >Sent: 19/10/2008 3:39:00 AM
>> >To: freebsd-questions at freebsd.org
>> >Cc: raggen at passagen.se
>> >Subject: Re: gconcat question
>> >
>> >On Saturday 18 October 2008, Roger Olofsson wrote:
>> >> What are the steps to bring back gconcatenated disks if doing an
>> >> upgrade from FreeBSD6 to FreeBSD7 like this?
>> >>
>> >> As-is situation:
>> >> FreeBSD 6.2-STABLE ad0 has FreeBSD ad1, ad2 and ad3 are 
>gconcatenated
>> >> using 'gconcat label -v data /dev/ad1 /dev/ad2 /dev/ad3'.
>> >
>> >The concat device should just appear automatically after the upgrade
as
>> >long as you (continue to) load the geom_concat kernel module. Be
aware
>> >that if
>> >the on-disk metadata format has changed then it will automatically
be
>> >upgraded. This is usually a good thing but if you need to roll back
to
>> >6.x for some reason it's something to take into consideration.
>> >
>> >> Planned upgrade:
>> >> Reboot from cdrom, install FreeBSD7 from cd to ad0
>> >
>> >Just curious, is there a reason you're going this route instead of
>> >upgrading from source?
>>
>> Hello John and thank you for your reply!
>>
>> Follow-up question - /dev contains a /dev/concat/label entry - is
this
>> entry created when loader.conf invokes the kernel module?
>
>Yes. Many of the GEOM modules (label, mirror, concat, stripe, etc)
create 
>nodes in the relevant subdirectories in /dev as soon as they "taste"
the 
>drives (or other providers) and discover metadata belonging to them.
This 
>is generally when they are loaded (if modules) or at boot time (if
compiled 
>into the kernel or preloaded by loader.conf). Any time you insert a
device 
>(such as a USB stick) the loaded modules also have an opportunity 
>to "taste" it and create nodes as appropriate.
>
>> The machine won't be rollbacked so that's not an issue.
>>
>> The reason for following this route is that it's faster than doing it
>> from source (it's an old machine). The machine has been a playground
and
>> has alot of ports installed that aren't being used anymore. The
>> concatenated drives contain data only hence the need to preserve
those.
>
>Makes sense. :)
>
>JN
>.
>

Thank you John, it worked excellent!

/Roger