PXI-SCXI in Linux running LV 8.0.1 consuming all available memory

[ratem]

Hi, we have the following environment here:
Software
- Mandriva 2006 PowerPack
- LabView 8.0 with update 8.0.1 for Linux
- DAQmx 8.0
- DAQmx base 2.0
Hardware
CHASSIS  PXI-1011: PXI-8176, 512Mb RAM, 6052e, SCXI-1129 (we are using 7 of them), BORNES SCXI 1129: 1335,  matrix topology 8x32
 LV program

Two analogic acquisition tasks created and cleared in a non-simultaneous way (one for rotation measurements, and the other for vibration measurements), and a switching task, in a total of 54 sensors.

Here goes the history:
After struggling (togheter with some NI folks) to configure this environment in Linux, now we have a different problem: In a matter of days our application eats all the available memory (swap and RAM) and freezes the PXI-SCXI.

We have already checked:
- (phisical) memories
- all start-clear/stop threads (including different combinations of stop AND clear and only clear)
- isolated our algorithm;
- isolated all local variables and arrays (no global variables);
- isolated all database related code.
In the end, after having only the acquisition tasks running, the LV process still eating aproximatelly 1Mb of RAM per minute, if we use a 1minute window for complete switch-loop and acquisition-loop.
We would like to know if there is some known workaround or upgrade that we are supposed to do to solve this problem.

Thanks in advance for all contributions,

Message Edited by ratem on 08-09-2007 05:34 PM

[leocavadas]

Ratem,

yu can use "Profile VIs" tool in "Tools" menu, "Advanced" tab to get a tip about that VI is "eating" memory.

Have a nice day.

[akokot]

Hi Ratem,

Memory leaks is a little difficult to debug. I can see you have already done some procedures to isolate the problem. The tests indicate that the acquisition code is consuming the memory. It is possible if you are leaving some tasks open (in memory). All tasks opened must be closed to deallocate the reserved memory.

There are others possibilities to the memory problem. The linux itself can be the problem. Are you sure the memory leaks are only observed when the LV program runs? If it's the case, I will suggest you to do a simple vi, a while loop with a timing function. Does this simple program show the same behavior?

Best Regards,
Alisson Kokot
NI BRAZIL

[ratem]

Hi Alisson,
thanks for your reply.
We don't  think its a Linux problem, anyway, we have two PXI here with us, they are not like the PXI-SCXI that we described previously, but we are going to do some tests using them, since it's not easy to have a constant access to the SCXI in question (more than 100Km from our lab, in a classified area, no external network access).
In other words, we have created a simpler version of the acquisition task and we are going to monitor the PXI to check where exactly is the problem, using Leonardo's hint.
Bye for now, if we discover it, I will post here.