[mythtv-users] Raspberry Pi as a slave backend server

[Dave MythTV]

On Sat, Mar 21, 2015 at 3:09 PM, jrh <jharbestonus at gmail.com> wrote:

>
> On Mar 21, 2015, at 12:56 PM, Dave MythTV <dave.mythtv at gmail.com> wrote:
>
>
> On Fri, Mar 6, 2015 at 8:22 PM, jrh <jharbestonus at gmail.com> wrote:
>
>> I have experimented some using the raspberry pi as a slave backend server
>> and have had generally good results with caveats.
>>
>> Configuration:
>> Raspberry Pi Slave backends running v 27.4(dated in November)
>> NFS mount to the Master backend for the mythtv directories.
>> Hauppauge USB HVR-950Q tuner(1 only ) (configured only 1 virtual tuner so
>> only 1 recording)
>> I did not configure the analog side of the tuner, only the digital part.
>>
>> Using a Raspberry Pi model B and B+:
>>
>> Able to record Standard Definition mpg video stream.
>> When attempting to record 720p or 1080i streams, there would consistently
>> be TFW taking to long errors.
>>
>> Using a Raspberry Pi 2 model B: (the quad core version)
>>
>> Able to record Standard Definition mpg video stream
>> Able to record both 720p and 1080i streams up to 15mbps!
>>
>> In summary, for 1 stream and a USB TV stick The RPi 2 is a viable
>> masterbackend with the above conditions. I didn’t try multiple streams off
>> the same real channel, but I suspicion the Pi 2 would be able to handle it.
>>
>> Note that I did not try analog recording as there are no analog NTSC Over
>> the air broadcast stations in the area and so I didn’t even try to
>> configure the analog side.
>>
>
> Thanks for sharing this information!
>
> A few quick questions for you:
> 1) Which Linux distribution are you using?
> 2) I'm assuming this testing was with a wired network (and not a USB
> wireless adapter)?
> 3) Do you have any thoughts on the source of the TFW errors on the Pi 1?
> Such as, whether the TFW errors with the Pi 1 were related to the CPU
> loading, pushing the data out across the network, or a limitation of the
> USB system?  (Since it kept up OK with standard definition streams, I'm
> wondering if the Pi 1 might be sufficient as a high-definition slave
> backend if it recorded directly to a small local USB hard drive, or even to
> its own SD card, and then did a non-realtime NFS data transfer back to the
> master backend when the tuner was idle?)
>
>
> Keep us posted on how things go if you try multiple streams or multiple
> tuners on the Pi 2!
>
> Thanks
> - Dave
>
>
> I am using raspbian off the noobs image.
>
> You are correct, I just used wired network. I didn’t even want to try
> going wireless.
>
> I didn’t even try experimenting using a local usb drive as that wouldn’t
> fit my usage scenario, so are still all good questions.
>
> I will be staying with a single USB tuner for my needs, and Since I
> started using the RPi 2 configured as above, I have had no TFW errors since
> starting to use it!
>
> I have no plans whatsoever to try to use an RPi of any kind as a frontend,
> however, they are very insteresting as a remote backend like I am using.
> The pi 2 plus the hauppauge USB tuner are only using around 7 watts while
> recording!
>
> By the Way, I noticed I goofed above when I said it was a viable master
> backend. I meant to say it is a viable SLAVE backend.
>
> An experiment I might try sometime in the future is to have an RPi as a
> masterbackend ONLY with no tuners(just define a dummy recorder). No
> immediate plans to try that tho yet.
>
> Regards
>
>

Thanks for the reply.

Remind me again what your usage scenario is?


I'm considering a possible usage for myself with a remote slave backend up
in the attic, on a wi-fi connection, to allow for separate tuners directly
connected to separate directional antennas aimed at some of our fringe
stations from nearby cities.   Being co-located up in the attic would allow
the USB tuners to be directly connected to the antennas (effectively 6
inches or less of coax), minimizing the cable loss and removing the need
for multiple additional long coax runs (and possibly pre-amps) to the
master backend/frontend in the living room.

The Raspberry Pi would seem to be a good candidate for this, but if I end
up needing two or more Pi 2s to handle the separate tuners individually, I
might as well just stick a single x86 ECS LIVA up there instead as the
overall price and power consumption would be similar, and the setup would
be much simpler.   If one of the Pi 1 models could do the task for a single
HD stream, however, then even with multiples, the price and power figures
swing into their favor.  :-)

And it's all in that interesting scenario where it would be easy to test
with experimentation, but if I try it and it doesn't work, the $$ "wasted"
on the Pis could have just bought the LIVA.    Hence the interest in the
details of what you've been testing!

Thanks
- Dave
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mythtv.org/pipermail/mythtv-users/attachments/20150322/11e9d078/attachment.html>