Author Topic: Random trash data outputted  (Read 23609 times)

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,259
  • Kudos: 182
Re: Random trash data outputted
« Reply #30 on: December 11, 2014, 01:51:37 PM »
I am not clear on what you mean by being on the fpdskip branch. Where is it and how do I get there?

My unix is more rusty than a Detroit steel mill. :)

You don't need to know Unix, :) we made it very easy to switch branches to help test or try out new features under development.

We use a source code repository called git to manage the source and released code for FPP.  git has the concept of branches which allows developers to work on multiple sets of code at the same or different times.  Each FPP release has it's own code branch (v0.3, v0.4, v1.0, etc.).  We also have a code branch called 'master' which is where new code is developed and another called 'stage' where we stage changes or bug fixes before they get pushed to a release branch.

If you normally get to FPP in your browser via http://SomeNameOrIP/  then instead go to http://SomeNameOrIP/developer.php and you will see a dropdown box labeled "Git branch".  Change that dropdown from "v1.0" to "fpdskip" and your FPP install will be switched to using the fpdskip branch which has a test patch in it.  The name is 'fpdskip' because the current test patch skips every other frame of output if the sequence player is trying to catch up to the media player and the sequence timing is adjusted to be 2ms shorter than the timing specified in the sequence file (ie, 48ms instead of 50ms or 23ms instead of 25ms).

To get back to the release v1.0 branch go back to the same developer.php page and select "v1.0" from the dropdown and you will be switched back to the release code.
-
Chris

Offline MrChristmas2000

  • Falcon Beta Team
  • **
  • Join Date: Apr 2013
  • Location:
  • Posts: 481
  • Kudos: 4
Re: Random trash data outputted
« Reply #31 on: December 11, 2014, 02:16:52 PM »
Do I need to be connected to the internet, mine is a stand alone system.

I am getting no response from the url hppt://192.168.0.128/developer.php .128 is my PI

Update, I just realized that I had not put this line back the way it was, NEEDKERNEL="3.12.29".

Am now rebooting to see if I can reach the page you are talking about.
« Last Edit: December 11, 2014, 02:23:42 PM by MrChristmas2000 »

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,259
  • Kudos: 182
Re: Random trash data outputted
« Reply #32 on: December 11, 2014, 02:28:13 PM »
The page probably hangs if the Pi can't reach the internet because we check the github.com site for the latest version and when you switch branches I think we pull in the latest version.  So, yes, it will need to be on the internet to test and get to the test 'fpdskip' branch.  Also, the branch may not show up until you run a manual update which will pull in the list of available branches.  The kernel version shouldn't matter, but you are best to switch back to 3.12.29 if the 3.10.33 didn't solve the issue.

Offline MrChristmas2000

  • Falcon Beta Team
  • **
  • Join Date: Apr 2013
  • Location:
  • Posts: 481
  • Kudos: 4
Re: Random trash data outputted
« Reply #33 on: December 11, 2014, 02:34:01 PM »
Ok, I reconnected it to the internet.

Now I am on that page following your instructions.

Let's see how this test goes.

I still say the phone is faster.

No improvement with the setup in step 1.
« Last Edit: December 11, 2014, 02:52:56 PM by MrChristmas2000 »

Offline MrChristmas2000

  • Falcon Beta Team
  • **
  • Join Date: Apr 2013
  • Location:
  • Posts: 481
  • Kudos: 4
Re: Random trash data outputted
« Reply #34 on: December 11, 2014, 04:05:19 PM »
UPDATE:

I started to run the show with the beta stuff and it was worse than before. It has even stopped running some of the elements. Goes to show you daytime testing is very hard to see everything.

 I went back to the standard release and will deal with the glitching.

Who knows my FPD could have gone bad. I do have another PI that I may try later tonight. I will try just about anything at this point. Maybe visitors will think the flashing is part of the show, who knows. On one sequence the coro trees are supposed to be red and green but they are changing multi colors during a chase.

Thanks for your attempts today capt.


Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,259
  • Kudos: 182
Random trash data outputted
« Reply #35 on: December 11, 2014, 04:31:14 PM »
By 'Worse' do you mean the flicker was worse or it 'jumped' because of the frame skipping?

I am burning in my debug patch right now and will push it in a few hours for others to test.  I will also run FPP under background tonight if I can to see if it shows anything.  The fpdskip branch should not have worse flicker, it should only skip frames when it is trying to catch up the sequence.

Offline MrChristmas2000

  • Falcon Beta Team
  • **
  • Join Date: Apr 2013
  • Location:
  • Posts: 481
  • Kudos: 4
Re: Random trash data outputted
« Reply #36 on: December 11, 2014, 04:49:07 PM »
Ok, I figured out why the show was dragging. I forgot to go turn logging off. That function just made the show go nuts.

Here is a big example of the glitching.

http://youtu.be/EZlHJowvPOU

It is easy to the trained eye to note that a lot of things that are flashing are not supposed to be, it is very obvious with the lights in the windows. The whole yard is also blinking including things that have absolutely no programming at all.


Offline drlucas

  • Developer
  • ******
  • Join Date: Sep 2013
  • Location: Whitby, Ontario - Canada
  • Posts: 1,422
  • Kudos: 3
Re: Random trash data outputted
« Reply #37 on: December 11, 2014, 05:28:57 PM »
That flashing is because of the debug settings, or that is what you see all the time in that video? Mine isn't THAT bad.

Now...on your house and setup/layout, simply beautiful!! 

I've applied the patch and will head out to the snow and see if I can see it happening. If it is, then I'm going to test with output turned off and then if that is happening after that i'm redoing the sequence that I see the problem happening the most and see if it makes any difference. I have redone about 7 of my 10 sequences with my new tune to sign, so it's work I have to do anyways.
Ryan

Offline CaptainMurdoch

  • Administrator
  • *****
  • Join Date: Sep 2013
  • Location: Washington
  • Posts: 9,259
  • Kudos: 182
Re: Random trash data outputted
« Reply #38 on: December 11, 2014, 05:32:40 PM »
High levels of debug logging can definitely cause playback issues because of the CPU usage they can require as well as all the writes to the SD card.   Which brings me to my next idea...

I am wondering if this has anything to do with I/O on the USB bus loading the sequence and media files.  Once I commit the new debug code tonight it should confirm whether the data is being corrupted inside FPP before it writes to the SPI interface or if data is somehow being corrupted during the write process.

No matter how many channels are in the sequence, for the FPD we always shuffle around 33kb of data at a time but the more channels the user has the more I/O on the USB bus at the time.

For those of you having this issue, can you describe your config a little more so I have some more data points to try to think of where the issue might be at.

A few pieces of info I would like to collect are:

- number of channels in sequence
- all output types in use (FPD, E1.31, USB, etc.)
- do the glitches occur at random times through the sequences or only at certain times
- do the glitches occur on channels all over or are they concentrated on specific channels.  If specific channel areas then can you give a few channel number ranges.
- Does it occur on long and short sequences or all sequences.
- Does the issue always occur on the same channels in a sequence or is it random every time the sequence is played?

I think that is enough for now.  I am trying to brainstorm.

Offline MrChristmas2000

  • Falcon Beta Team
  • **
  • Join Date: Apr 2013
  • Location:
  • Posts: 481
  • Kudos: 4
Re: Random trash data outputted
« Reply #39 on: December 11, 2014, 05:36:13 PM »
That video is after I turned off the logging. You should have seen it before. UGH!!!!! I was just trying to get the show backonline and forgot about the logging.

I added some effects to my MegaGrid and I am going back to take them off because it is causing the grid to get stuck on all white during the two sequences that have effects on it. Downer.

I sure hope I hit on the right combination soon. I am working on building an backup PI setup to replace it when the show ends to see if perhaps I have a bad control signals going to the FPD.







Offline drlucas

  • Developer
  • ******
  • Join Date: Sep 2013
  • Location: Whitby, Ontario - Canada
  • Posts: 1,422
  • Kudos: 3
Re: Random trash data outputted
« Reply #40 on: December 11, 2014, 05:48:36 PM »
I've attached my rgbeffects and my sequence file to give you an idea of channels i have running. my tune to sign is like a 20second very simple sequence just using the matrix. my nutrocker sequence is a longer sequence. both have the issue. BTW: with the latest patch, glitch still happening at random places. I did not turn on debug as I'm guessing until your new debug code comes out it won't show you much so right now my logging level is pretty much off - see settings below.

I am only using FPD, e1.31 not enabled.  issues seem to be limited to the lower part of my matrix...one little section - say in the 6000 to 9000 range (ish).

I didn't answer the other questions because I'd be guessing but I'm guessing it happening at the same place (location wise in my display, not all over the place) and those channels are fed off a F16 that spans multiple universes. all contiguous channels. 

Code: [Select]
PiFaceDetected = 0
FalconHardwareDetected = 1
HostName = FPP
piRTC = 1
fppMode = master
restartFlag = 0
MultiSyncRemotes = 192.168.15.136
FPDEnabled = 1
screensaver = 1
forceLocalAudio = 1
rebootFlag = 0
PI_LCD_Enabled = 1
alwaysTransmit = 1
LogLevel = warn
volume = 89
LogMask =
Repeat =
E131Enabled = 0
Frequency = 103.7
Station = Merry Christmas
[code]

[attachment deleted by admin]

Offline drlucas

  • Developer
  • ******
  • Join Date: Sep 2013
  • Location: Whitby, Ontario - Canada
  • Posts: 1,422
  • Kudos: 3
Re: Random trash data outputted
« Reply #41 on: December 11, 2014, 06:07:18 PM »
This just in.....

Row 28 for my matrix was a frequent place...channel range 8073-8342. happened a lot during the tune-to radio sequence.

I turned off the setting to always transmit channel data and it made no difference on the glitching. Now I did have a heck of a problem with fppd restarting from the web page - only way I could get it to restart was reboot PI. Maybe could of tried from CLI, but didn't want to waste any more time with the show running. it's a crazy night so unlikely a lot of cars out after the brutal driving conditions around the city today (over 500 accidents on the highways today - yeah, Toronto has a lot of bad drivers - but thats a different discussion!).

Offline MrChristmas2000

  • Falcon Beta Team
  • **
  • Join Date: Apr 2013
  • Location:
  • Posts: 481
  • Kudos: 4
Re: Random trash data outputted
« Reply #42 on: December 11, 2014, 06:19:23 PM »
Pixelnet channel count is 22011
DMX channel count is 441

Using an FPD
Jack 1
Pixelnet universe 1, 2, 3, 4
Jack 2
Pixelnet universe 5, 6, 2
Jack 3
DMX Universe 1 - Used in the house inside lights

All other DMX is from Falcon 16 Hubs DMX Jacks

Glitches occur at random times through all my sequences.

Glitches occur across all channels at one time or another

Sequence length does not seem to matter.

Once it was thought to be happening in just the my Mega Grid in the PN universe 5 and 6 but now there are glitches all over the whole show as the video shows.

Here is a video of the same sequence from last year when I was running a FPPD driven from xLights.

https://www.youtube.com/watch?v=Xk8oyiFa_Nc


Offline JonB256

  • Supporting Member
  • ******
  • Join Date: Mar 2013
  • Location: Granbury, Texas
  • Posts: 4,915
  • Kudos: 111
    • Granbury Christmas Lights
Re: Random trash data outputted
« Reply #43 on: December 11, 2014, 08:13:39 PM »
Pixelnet channel count is 22011
DMX channel count is 441

Using an FPD
Jack 1
Pixelnet universe 1, 2, 3, 4
Jack 2
Pixelnet universe 5, 6, 2
Jack 3
DMX Universe 1 - Used in the house inside lights

All other DMX is from Falcon 16 Hubs DMX Jacks

https://www.youtube.com/watch?v=Xk8oyiFa_Nc

Try putting everything that is on Jack 2 on Jack 3 and vice versa.
You have virtually 100% of the load on a single PIC (1 2 3 4 5 and 6)
and very little on the second (7 8 9 10 11 and 12)

Offline Steve Gase

  • Supporting Member
  • ******
  • Join Date: Mar 2013
  • Location: Georgetown, TX (near Austin)
  • Posts: 1,037
  • Kudos: 5
    • WinterLightShow in Georgetown, TX
Re: Random trash data outputted
« Reply #44 on: December 11, 2014, 08:41:04 PM »
Aren't all channels transmitted?  if there are is no defined pixels, then 0x00 is still transmitted for a channel. 
The decision and work about the color is done by nutcracker of LSP.
FPP and FPD is just moving data and the work should be consistent across all universes.  right?
http://WinterLightShow.com  |  110K channels, 50K lights  |  Nutcracker, Falcon, DLA, HolidayCoro

 

Back to top