Distributed Control & Automation Framework (DCAF)

cancel
Showing results for 
Search instead for 
Did you mean: 

TDMS Datalogger Memory leak on 9068

Solved!
Go to solution

I am using  an old version of the TDMS Datalogger that has been modified slightly. I can email you our version if need be. However, we are seeing a memory leak of about 1MB/Minute when executing on the 9068 controller. When we run in Windows, memory is stable down to the KB. We are using the same project config, simply dragging the project to PC and setting the log path to a windows path.

 

Have you all seen this? Are you able to reproduce this on 9068 targets?

0 Kudos
Message 1 of 8
(7,376 Views)

There's a known issue with memory reporting on LinuxRT targets where it can seem like you have a memory leak but do not actually have one.  This occurs specifically when doing large amounts of file IO.

 

http://digital.ni.com/public.nsf/allkb/AC6200D19D23C61586257C8D006E6DC2

 

As a fun experiment, let the app keep running until memory should be zero.

Cheers,

Matt Pollock
National Instruments
0 Kudos
Message 2 of 8
(7,364 Views)

We have done that unfortunately. When the memory drops to around 4MB, we start having massive control lag, outputs firing out of time, and commands taking several seconds to execute. 

0 Kudos
Message 3 of 8
(7,362 Views)

Hmm, that is consistent with a real leak (paging disk when out of physical memory).

Cheers,

Matt Pollock
National Instruments
0 Kudos
Message 4 of 8
(7,360 Views)

Can you send over your package?  I'll look into it on Monday and see if we can reproduce, and whether the current version shows same behavior.

Cheers,

Matt Pollock
National Instruments
0 Kudos
Message 5 of 8
(7,357 Views)

I'll get you my package by monday.

If you see this before then, I'm going to attempt to use the latest DCAF TDMS logger and see if I get the same issue.

 

Can you think of a good way to monitor weather or not I'm leaking memory besides trying to set outputs and waiting for them to get erroneous?

 

I am concerned that I may be chasing a red herring with this linux memory reporting issue.

0 Kudos
Message 6 of 8
(7,326 Views)
Solution
Accepted by MPCC

We didn't actually have a memory leak. The issue was that as the bounded buffer filled, the chassis ran out of memory. It actually couldn't fit the bounded buffer. This wasn't an issue with the module as much as it was an issue of hardware specification on our end.

Thanks for all of the support.

Message 7 of 8
(7,147 Views)

Thanks for posting status here for posterity!  

Cheers,

Matt Pollock
National Instruments
0 Kudos
Message 8 of 8
(7,145 Views)