[mythtv-users] .25 mythtranscode copy over in .26 failing after upgrade

[Rich Freeman]

On Fri, Mar 15, 2013 at 1:33 PM, Michael T. Dean
<mtdean at thirdcontact.com> wrote:
> Do you really need (or even want?) RAID for DVR recordings?  IMHO, it's just
> TV, so it's not worth the cost of RAID.

Already running on RAID.  Makes it easy to add/replace/etc drives, and
I have had them fail.  Sure, it isn't the end of the world if I lose
some TV, but I'd prefer not to.  I have 735G allocated to my mythtv
video lvm storage group.  The server stores more than just mythtv
video.


>  You can get a 2TB for $70 - $100
> (such as http://www.newegg.com/Product/Product.aspx?Item=N82E16822148834 for
> $90 after promo code) or a 3TB HDD for ~ $110 - $130 (such as
> http://www.newegg.com/Product/Product.aspx?Item=22-136-874 for $120), to
> either double or triple your current storage.

Well, unless I start getting creative with SATA interface cards or NAS
adding a 2TB drive will only get me 1TB of actual storage since I'll
have to remove a 1TB drive to do it.  I'd prefer to wait until
technology marches along and capacities increase.

>
> IMHO, transcoding isn't worth the time (*your* time) that it consumes.

Transcoding doesn't take any of my time.  It is triggered automatically.

> Commercial detection is likely to fail on any other-than-normal video format
> (as it's not tested, nor intended for that), so you'd either have to hope
> mythcommflag works and/or make it work by fixing any issues or try to remap
> old cuts to new video.

That's exactly what mythtranscode does already.  The cutlists are
mapped to frame numbers, and those don't change unless you're doing
something crazy like changing framerates.

>  You'd also have to clear out the old seek table
> and--depending on the new format--possibly want to create a new seek table.
> And that doesn't even go into the processor time (read, expenditure of real
> money on electricity) involved, either.

Rebuilding the seektable just requires invoking mythcommflag - that is
pretty easy (again, assuming it accepts the video format).  I have
thought a little about money for transcoding vs storage.  The issue
with that is that you have to burn through a lot of CPU to equal the
cost of a hard drive.  1TB of disk space costs about $80 give or take,
and it would hold less than 200 hours of video.  Transcoding 200 hours
of video uses about 200 hours of CPU give or take, and even at 200W
that only costs maybe $5.  Granted, that isn't strictly apples to
apples - you have to pay that $5 every time you turn over all the
space on the drive where the drive is a one-time expense.  Still, even
if I record and keep 10h/day of video it would take a year or two to
rack up the cost of an extra hard drive.  Factor in net-present-value
and all that I'm not sure it is obvious that a drive makes sense.

RAID isn't that expensive, either, if you're talking about RAID5 with
a reasonably large array.  Sure, if you want to do RAID1 you're
doubling your costs, but if you have 5-6 drives in an array it is only
20% overhead.

Another reason I'm holding off on yet more drives is that I'd prefer
to make my next drive upgrade a large one so that I can also move all
my data to btrfs.  I wouldn't go near it except for testing today, but
by the time I could fit my entire RAID on one or two drives I'm hoping
it will be stable, in which case it won't be that hard to add a drive,
migrate my data, and then extend the new array.  You can't really do
that easily when you're talking about moving from 1TB to 2TB drives.

But, I do appreciate the perspective - it is worth thinking about
these things.  I've seen enough drives die that I'd prefer to not run
without RAID.  I really can't afford to do offsite backups of MythTV
video like I do with anything valuable on the server.  Worst case I
can extend the video lvm - I have a few hundred GB of unallocated
space and around 750G I'm using to do clonezilla backups of PCs around
the house which isn't really space-efficient.  I'd probably tap into
that before adding drives.

Rich