[mythtv-users] Is there a distributed filesystem available?

Chris Henderson jchendo at gmail.com
Tue Jun 20 02:19:02 UTC 2006


On 6/20/06, Debabrata Banerjee <davatar at comcast.net> wrote:
>
> I already have several other raid arrays in my myth box for other
> purposes,
> I am quite sure this is not what I want. If I suppose nothing to meet
> these
> requirements exist, this could be easily implemented in mythtv, actually
> it
> 'sort of' does it right now if you use multiple backends with storage. It
> just can't manage the storage and can't have more than one per server. In
> my
> setup I could lose every disk besides one and still have some video data
> and
> an operational system. It would still record the daily show tonight. Also
> it's some protection against file system corruption (and it actually
> happened to my /myth filesystem, completely trashed)..
>
> RAID is slow, complicated, wasteful, and overkill for myth. A single disk
> can handle many streams of video. A fault-tolerant filesystem and JBOD is
> what I want.
>
> *sigh*
>
OK,

But i dont understand. Please exlpain. You say you want a fault tolerant
files system, doesnt Raid 5 give you that?

And for raid 5 being slow, umm if you cheap out and use the software raid 5,
then YEA its a bit slower, but i would not think you would see it in this
enviroment. Unless of course you are recording muiltpul HD streams (3 or
more) and streaming out at the same time.

If you go with a Hardware Raid card or board there *should* be no
differeance between what normal and RAID. Infact you *should* get better
read/write performance. (not exactly true, happens more with STRIPED
arrays.)

You could make a SAN or more likly a NAS (Network Attached Storage) out of a
slave backend.
What i am trying to build is as follows. (cash strapped. cant get the drives
ATM)

P3 or 4 or AMD (it doesnt matter for this)
2GB ram (over kill)
1 x 80GB HDD for boot and swap
6 x 200GB HDD in Raid 5 Array (this will include / )
Gbit lan

THis should make a NAS with 1TB of storage. As it is fault redundant I will
put everything on here and the Master BE will dump everything on it. I got
given a Gbit 24 port switch so i will use that to connect, that should make
it so there is no problem for the Slave Backend to stream the required data.

To that each frontend and backend will have a cron job to backup the DB at
some stage each day and each of them will have an image on the
SlaveBackend/NAS. I want to do lowpowered EpiaM 1300 frontends in the future
and would like them to net boot. These images will also be stored on the
SlaveBackend/NAS

Please let me know what you see wrong with this.

I dont know of any files system that will eliminate corruptions etc.
even in Large enterprise thisis worried over but, i dont know of anythig
they do to stop it.

Bang for Buck i dont know of a better way.

CH
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mythtv.org/pipermail/mythtv-users/attachments/20060620/ce6c3596/attachment.htm 


More information about the mythtv-users mailing list