[mythtv-users] Fwd: Pixelation/Bad Recordings HDHR Prime -- I am at my wits end

Stephen P. Villano stephen.p.villano at gmail.com
Thu Sep 12 04:51:37 UTC 2013


On 9/12/13 12:38 AM, Joseph Fry wrote:
>>>>>>>>> I have rewired everything again as I am not comfortable using an
>>>>>>>>> amplifier
>>>>>>>>> to amplify my signal, only to use multiple splitters to reduce it
>>>>>>>>> back
>>>>>>>>> down.
>>>>>>>>> It doesn't seem to make sense.  Instead I have connected the Monster
>>>>>>>>> cable
>>>>>>>>> 4-way splitter (-7.5dBmV/output) to the cable coming in from Comcast
>>>>>>>>> and
>>>>>>>>> connected my HDHR Prime directly to one of the outputs.  Another
>>>>>>>>> output
>>>>>>>>> of
>>>>>>>>> the Monster Cable splitter is connected to another splitter
>>>>>>>>> (-3.5dBmV/output) which goes to my TVs.  Another output of the
>>>>>>>>> Monster
>>>>>>>>> Cable
>>>>>>>>> splitter then goes to my existing long run to my cable modem.
>>>>>> I don't know if it's relevant, but you only mention 3 feeds off of the
>>>>>> splitter, is the other output terminated properly? Or did you mean a
>>>>>> four
>>>>>> outlet splitter, one in and 3 out? And are all outputs on the second
>>>>>> splitter also terminated properly?
>>>>>>
>>>>>> _______________________________________________
>>>>>> mythtv-users mailing list
>>>>>> mythtv-users at mythtv.org
>>>>>> http://www.mythtv.org/mailman/listinfo/mythtv-users
>>>>>>
>>>>> Regarding interference, I am seeing this on what seems like every
>>>>> single channel at least some of the time, so I don't think that it is
>>>>> interference, unless I am seeing interference on everything on the
>>>>> HDHR Prime and this isn't affecting my cable modem or either Comcast
>>>>> set top box (including one which is a DVR).
>>>>>
>>>>> My mistake on the wiring; I probably confused myself I rewired it so
>>>>> many times.  I just looked at it and I have the following:
>>>>>
>>>>> Comcast signal in to MoCA filter to a two output splitter.  One end
>>>>> goes to my HDHR Prime, the other end goes to another two output
>>>>> splitter.  This two output splitter is then connected to my cable
>>>>> modem and then the other end goes into yet another two output splitter
>>>>> that goes to my two televisions.  So between my HDHR Prime and the
>>>>> Comcast "feed", I only have one splitter.  Yet on my TVs, I have three
>>>>> splitters in front of them and no issues; and my cable modem has two
>>>>> splitters in front of it and no issues.
>>>> I believe I have narrowed this down to an issue with my hardware, or my
>>>> MythTV setup and have been wasting my time chasing signal issues.  I have a
>>>> thread on Silicon Dust's site (much is duplicated here, but here is the link
>>>> if you are bored:
>>>> http://www.silicondust.com/forum2/viewtopic.php?f=44&t=15543) and a member
>>>> of their team was assisting me.  It appears that the HDHR Prime is getting
>>>> perfect signal, and the video in both my desktop (Windows MCE) and VLC (HDHR
>>>> GUI in Linux on my frontend) are pristine with zero audio dropouts or video
>>>> pixelation when watched simultaneously.
>>>>
>>>> Now I am not sure where to go from here. In looking at the performance data
>>>> in vSphere, I am not seeing anything that may cause this. This is a pretty
>>>> decent machine, i5 3570K (overclocked to 4 ghz, though I don't know that
>>>> ESXi is respecting this), 16GB of RAM. I have 4 CPU cores and 12GB of RAM
>>>> provisioned to the Myth Backend server and I am not doing anything else with
>>>> it, so I can't imagine it is a performance issue with anything other than
>>>> the storage, which is all in the same box and carved up as VMDKs by ESXi.
>>>>
>>>> I have a few routes I can go here; first, I could use a separate spindle for
>>>> just my recordings (right now it is being shared with a Windows Server data
>>>> share that sees fairly little i/o). Second, I could pop in a 60gb SSD I have
>>>> laying around and write some recordings to that to see if it makes a
>>>> difference.
>>>>
>>>> Any suggestions?  I really don't want to build another box and turn it into
>>>> a NAS, though that's probably my best bet...  I have enough money sunk into
>>>> this at this point.
>>> I hate to say this... but this is the exact reason why running the
>>> backend on a VM is only recommended by those that have done it
>>> successfully for themselves, and not by anyone who actually has had to
>>> try and resolve an issue with one.
>>>
>>> You should have plenty of horsepower to actually play the stream on
>>> your backend in VLC, at least then you could determine if it is an
>>> issue with the data coming into the backend, or the the IO with the
>>> disk.
>>>
>>> Next, you might try shutting down the Windows VM (and any others),
>>> just to remove them from the equation.
>>>
>>> If you suspect a network IO issue, rather than disk (seeing pixelation
>>> playing the stream in vlc), you might add a network card and dedicate
>>> it to the backend VM.
>>>
>>> If you suspect disk IO (stream plays fine in vlc),  I would first
>>> ensure that your recording drives are formatted using advanced format
>>> (4k sectors) and that they are properly aligned.  Misalignment can
>>> cause a VERY significant impact to write speeds (
>>> http://hothardware.com/Articles/WDs-1TB-Caviar-Green-w-Advanced-Format-Windows-XP-Users-Pay-Attention/?page=2).
>>>  Next, you might try changing the filesystem your using, EXT4 is a
>>> pretty safe bet; just to see if it makes any difference.  Finally, you
>>> might need to see if there is any way to give the backend direct
>>> access to the SATA controller (assuming you have more than one)... it
>>> may be an issue with ESX's drivers, or the virtual controller driver.
>>> Oh, and did you select the LSI virtual SCSI controller when you
>>> configured the VM, the other one isn't supposed to be as fast?
>>>
>>> There are just so many places thing can go wrong with a VM... I wish you luck.
>> You should also warn against fake raid, which can suck the performance
>> out of a system so horrifically that one could end up swearing that the
>> thing needs to be wound up again.
> Indeed.  I'm not even sure that ESX would support fake raid
> controllers, but if it does it shouldn't.... and you definitely
> shouldn't be using it.  In fact, I would eliminate any RAID
> whatsoever, if that's part of your build.
>
> One other thing I didn't think of... your not using dynamic virtual
> disks as recording drives are you... you did assign each of your
> recording drives, as entire drives, to the backend VM?
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://www.mythtv.org/mailman/listinfo/mythtv-users

I don't know about eliminating RAID *entirely*, hardware RAID is quite
quick and doesn't impact throughput significantly. But then, I'm talking
about "big iron" RAID and servers, which are designed to throw data
about quickly, while serving up video up anemically at best.
Each device to its purpose.  :)
I'm seriously considering putting one of my old Dell 2850 servers into
service as a myth backend, let faster performing workstations handle the
frontend exclusively.
The processors and board are designed to toss data about and process it,
but aren't worth writing home about displaying it to one's screen,
they're servers.
Meanwhile, my ancient workstation can handle zoneminder exclusively,
rather than a fitful mix of zoneminder and myth backend.

But, tuning it to actually, if barely, work was quite fun!
Hard problems are fun. Insolvable problems, such as making fake raid
work are masochism. Hard problems can be solved.

OT question for any who know about cablecard crypto, anyone have the
tech info on the crypto utilized?
I have a theory that could explain why it works when not authenticated
on some providers, but not on others.
>From the large amount of reading I've done before I finally got my HR
Prime today and what I also experienced, I have some suspicions on what
is going on at various providers in how it's implemented as far as fail
safe vs fail dead at different service tiers.


More information about the mythtv-users mailing list