1 Topic by zhangye

  • zhangye
  • Member
  • Offline
  • Registered: 2007-12-14
  • Posts: 41

Topic: Help about 'piond' has exited of its own accord

We used NION3.My project upload to nion ,Nion have " 'piond' has exited of its own accord" falt.We have two groups ,every 10 nions creat one group,one group for backup. Every day we found some nions have same error.The falt nion not all the same  one.
My project no python scrit,but every nion use two "project link" and "String Multiplexer"  module to SYN backup NION'S data,every nion about 50 button and string need syn. evey  "project link Client" poll period  100 ms.

How to slove it ? What might affect those? Thanks.

Follow is our logs:
7/30/2013 19:39:03    11033    note    piond/role_manager    role is stopped
7/30/2013 19:39:03    11032    fault    piond/role_manager    role fault: role restart failed: 3 dsps expected but only 2 found
7/30/2013 19:39:03    11031    fault    piond/fault_policy    more than one error in less than one minute; stopping engine
7/30/2013 19:39:03    11030    fault    piond/sound_engine    exception during engine start: 3 dsps expected but only 2 found
7/30/2013 19:38:48    11029    note    piond/role_manager    restarting role : Audio-V1.4.5-10Z-procing/N001/rxYvmCXftD3Ln0orgYSomZVgmFz/HvnfZvMC2YEvpcqaHVEFz5zCUuG
7/30/2013 19:38:47    11028    error    piond/fault_policy    restarting audio engine
7/30/2013 19:38:47    11027    note    piond/version    Audio network module Peavey NION CobraNet version 2.21.17(.3)CM-1(a) rev 4
7/30/2013 19:38:47    11026    note    piond/version    Firmware version release_1.6.6a
7/30/2013 19:38:47    11025    fault    piond/role_manager    role fault: role restart failed: error reading/booting kernel: failed to read/write section data: /dev/pion/sharc1: bus request timeout
7/30/2013 19:38:47    11024    fault    piond/sound_engine    exception during engine start: error reading/booting kernel: failed to read/write section data: /dev/pion/sharc1: bus request timeout
7/30/2013 19:38:27    11023    note    piond/role_manager    restarting role : Audio-V1.4.5-10Z-procing/N001/rxYvmCXftD3Ln0orgYSomZVgmFz/HvnfZvMC2YEvpcqaHVEFz5zCUuG
7/30/2013 19:38:18    11021    note    piond/time    Tue Jul 30 19:38:18 2013 CST, Tue Jul 30 11:38:18 2013 UTC
7/30/2013 19:38:10    11019    error    mcp/processes    process 324 'piond' has exited of its own accord

2 Reply by JoshM

  • JoshM
  • JoshM
  • Recording Engineer at Burnt Orange Studio
  • Offline
  • From: Meridian, MS
  • Registered: 2007-08-13
  • Posts: 1,498

Re: Help about 'piond' has exited of its own accord

It looks like something in your project file is crashing one of the DSPs.

Without at least looking at your NWare project file, I can not even begin to speculate what the problem may be.

Josh Millward
Burnt Orange Studios

JoshM's Website

3 Reply by zhangye

  • zhangye
  • Member
  • Offline
  • Registered: 2007-12-14
  • Posts: 41

Re: Help about 'piond' has exited of its own accord

Hi,JoshM
Do you have e-mail?I will send the project file to you,Please help me check the project file ,Thanks

4 Reply by zhangye

  • zhangye
  • Member
  • Offline
  • Registered: 2007-12-14
  • Posts: 41

Re: Help about 'piond' has exited of its own accord

who can help me ?

5 Reply by ivorr

  • ivorr
  • ivorr
  • DSP Geek
  • Offline
  • From: West Des Moines, IA
  • Registered: 2007-08-17
  • Posts: 56

Re: Help about 'piond' has exited of its own accord

Take a look at the compile report, and pay special attention to the memory usage for block 1. If it's anything over 93% for a chip, NION tends to become unstable and crash. The fix is dependent on your file. Sometimes changing the maximum delay time in signal delays to slightly more than what is actually needed in reality will free up enough memory to get the file to stabilize. Sometimes hand placement of algorithms to a different chip is needed.

XDAB (if used) HINTS START
Another cause of your problem could be XDAB related. Having more signals than absolutely necessary on XDAB can cause big problems. It can create XDAB loops which can be disastrous.

Make sure you don't have any unused channels in your XDAB device. NWare reserves resources for them even if they're unused. I've seen that cause weird problems, too.
XDAB (if used) HINTS END

If using any MediaMatrix Subnet Bridge devices in the file, never configure them to have two NIONs point to each other on the network. it creates a network loop which crashes piond every time. The Subnet Bridge device is for one-way traffic only.

Last edited by ivorr (2013-08-09 17:47:18)

The only true wisdom is in knowing you know nothing. -Socrates

ivorr's Website

6 Reply by zhangye

  • zhangye
  • Member
  • Offline
  • Registered: 2007-12-14
  • Posts: 41

Re: Help about 'piond' has exited of its own accord

Hi ivorr,
I've followed you said check procedures, the memory usage for block 1 below 86% every DSP chip. So I check the DSP loop,

I found the program only the "Ambient Sensing Leveler"  module(Continuous Sensing) may generate DSP loop,i use the Gap Sensing instead of  Continuous Sensing.Running 5 days ,the nion work well done.

thank you

7 Reply by glparke

  • glparke
  • Member
  • Offline
  • Registered: 2012-10-25
  • Posts: 47

Re: Help about 'piond' has exited of its own accord

ivorr wrote:

Take a look at the compile report, and pay special attention to the memory usage for block 1. If it's anything over 93% for a chip, NION tends to become unstable and crash. The fix is dependent on your file. Sometimes changing the maximum delay time in signal delays to slightly more than what is actually needed in reality will free up enough memory to get the file to stabilize. Sometimes hand placement of algorithms to a different chip is needed.

XDAB (if used) HINTS START
Another cause of your problem could be XDAB related. Having more signals than absolutely necessary on XDAB can cause big problems. It can create XDAB loops which can be disastrous.

Make sure you don't have any unused channels in your XDAB device. NWare reserves resources for them even if they're unused. I've seen that cause weird problems, too.
XDAB (if used) HINTS END

If using any MediaMatrix Subnet Bridge devices in the file, never configure them to have two NIONs point to each other on the network. it creates a network loop which crashes piond every time. The Subnet Bridge device is for one-way traffic only.

Hi Ivor - I was working on a file and taking a look at my Block 1 usage and there, and there are a few chips on each Nion (using N6's) that are at 99%. While chip F is at 5% and C,D,E are looking good. I tried pulling about 10 delays from one of my A chips to try and get that block 1 number  lower, and Nware just puts other devices back on that chip. So Nware, is filling those first chips up as full as it can. if 93% is a top limit for stability, shouldn't Nware compiler stop at that limit??
  I have not had any issues though with my file,

Thanks
Gary

8 Reply by JoshM

  • JoshM
  • JoshM
  • Recording Engineer at Burnt Orange Studio
  • Offline
  • From: Meridian, MS
  • Registered: 2007-08-13
  • Posts: 1,498

Re: Help about 'piond' has exited of its own accord

Hi Gary,

This is NOT an accurate statement:

Take a look at the compile report, and pay special attention to the memory usage for block 1. If it's anything over 93% for a chip, NION tends to become unstable and crash.

What is an accurate statement is that the values shown in the compile report are NOT MEASURED. They are estimates. As with any estimate, there is a little guesswork going on and this leads to inherent inaccuracies.

IF you are having problems with your project file (which you said you are not so this doesn't apply anyway) wherein your NION is crashing on occasion and the Remote Log for the NION is pointing at a specific DSP chip (this usually will happen immediately when the role is attempted to start, it is never something that "develops with time", when looking at the remote log you will generally see a few comments in a row that say something about a "DMA error"); then I would review the Resource Report for that chip and if I see any memory or DSP allocations that are almost 100% (something like 99.99% or 99.98%) AND I look at the "details" for the Resource Report and see a LOT of individual algorithms on that DSP, then I would consider that the compiler is inaccurately stacking too much stuff in that DSP and the aggregate error in the resource usage estimation is pushing the actual requirements to be more than 100%. Generally, taking a single large algorithm and placing it on that chip so there isn't as much room left over for the little algorithms will fix the issue.

The compiler will ALWAYS stack as much stuff in the first DSP chip as possible, then fill up additional DSP chips. This is specifically to reduce the number of times the signal has to be moved between DSP chips. The less it has to be shuttled around, the less latency there is in the whole system.

There is no inherent issue with using more than 93% of the resources in a DSP. The inherent problem is with using lots and lots and lots of tiny algorithms. Why do you have 50 different individual meters in your project? You could have one meter device with 50 meters in it.

The important thing to remember, however, is that if you are not having problems with your project file, you don't need to worry about it.

Josh Millward
Burnt Orange Studios

JoshM's Website

9 Reply by glparke

  • glparke
  • Member
  • Offline
  • Registered: 2012-10-25
  • Posts: 47

Re: Help about 'piond' has exited of its own accord

Awesome. Thanks Josh. That clarifies things and makes sense.

Gary