[mythtv-users] Storage Solution: Can ya help a brother out?

[vamythguy]

On Mon, Nov 24, 2008 at 3:24 AM, Jake Anderson <yahoo at vapourforge.com>wrote:

>  vamythguy wrote:
>
> On Sun, Nov 23, 2008 at 9:29 PM, Brad DerManouelian <
> myth at dermanouelian.com> wrote:
>
>> On Nov 23, 2008, at 9:14 PM, vamythguy wrote:
>>
>> > Ok.  Let me start with - I don't get LVM + RAID.  The idea of being
>> > able to throw differently sized disks in one side and having a
>> > failure resistant dynamically extendable disk solution come-out the
>> > other is great, but I don't get LVM + RAID.  Specifically, how it
>> > works.  Why both?
>>
>>  Read more for your answer.
>>
>> > Also, I love the idea of not being constained by the number of slots
>> > in a box, so the extent to which that can be abstracted across a
>> > protocol like iSCSI or AoE or eSATA would be great - especially
>> > since performance is only marginally important to me.  So, how
>> > would.does something like this work/get built?
>>
>>  There's your answer to the question above and the explanation above
>> answers the question here. That was easy.
>>
>> _______________________________________________
>> mythtv-users mailing list
>> mythtv-users at mythtv.org
>> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
>>
>
> Great!  All the pieces are in the box - so, how do you put them together?
> So I get how LVM let's me keep throwing disks at it, but how do I get
> fault-resistance?  RAID, right?  But I don't think RAID likes new disks,
> does it? So, maybe starting with how to get that to work.
>
> ------------------------------
>
>    Don't RAID your mythtv stuff.
> Storage groups work much better in terms of disk IO/seeking when under
> load, as a result your disks will probably last longer.
> HDD failures are so rare these days don't bother with it, worst case you
> loose some TV, you should have better things to do anyway ;-P
>
> If you have some other stuff you want to do on the machine then split some
> disks into partitions, use some partitions for software raid (MDADM) and
> some for storage groups. That is what I have now (3x 320gb drives with a
> 60gb raid partition on each and the rest as XFS parts for storage group,
> boot is off a 20gb part on a 420gb disk with the rest set as a storage
> group). If you start with 3 disks in your RAID you get RAID 5, you can add
> disks to raid5 arrays with mdadm now then grow the file system to handle the
> added space. If you pick a file system that supports online growth you can
> do that whole thing with 0 down time, pretty heavy performance hit though.
>
>
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
>
> So I guess part of my issue with the RAID stuff is how I've got 3x250 and
1x300 in a RAID5 array, so I'm missing 50G.  This is because I had a 250 go
bad and replaced it with a 300.  Ideally, that extra 50 would've just been
brought in and be usable - even if not RAID5.  Maybe I'm looking for too
much.  I've had problems before with making a decision one way and then not
being able to change it (because of the size of the filesystem), so I'm
trying to be smarter about it.

All of which skips my other issue, that being the reclamation of that
orphaned array and figuring-out how to get back at it (given I've had issues
in the past with moving the disks of an array to a new mobo, a new instance
of mdadm, whatever, and having it not be recognized).

Finally, is a multi-port eSATA solution a good way to externally house these
drives?  Is AoE a real option?

I know, I'm all over the place here...
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-users/attachments/20081124/934a8771/attachment.htm