Maximum waveform size for a networked shared variable

[Rick_F_Foster]

I am using a network shared variable to hold an large waveform (66,560 doubles) on a RT system. The RT system is very stable actualling this waveform from a multifunction board, but the XP system that is reading and analyzing the waveform will hard freeze my XP after some random time (sometimes 10 minutes, sometimes 1 hour). I am acquiring these waveforms at about 5 waveforms/sec.

As a sanity check, I just used the Variable Manager on the XP  to monitor that particular variable and it too would hard freeze my XP after some random time.

There are no errors in the XP event log after I reboot, so it is difficult to track what is going wrong.

Is there any contraints on the physical size of a Network Shared Variable?

[Jon_S.]

Hello Rick,

I think the problem you are seeing is the total LabVIEW memory reaching its limits.  Open the task manager when you run this program.  Does the memory of LabVIEW.exe constantly increase until you see a crash?   What are you doing on the PC side of this application.  Are you storing all of these waveforms in memory (e.g. ploting them on a chart)?

[Rick_F_Foster]

The memory usage is very stable and at a reasonably low level as I am not accummulating these waveforms. I am either doing a simple quicklook plot/analyses or writing them to disk.

Just to make sure it wasn't something I was doing in my VI on the PC side, I just ran the NI Variable Manager and added the network variable with the large waveform to the watchlist. It also hard frooze the PC after some appearantly random amount of time (in that case it was ~1/2 hour).

[Rick_F_Foster]

One thing that I just did was to apply "Show Timeout" to the Network Shared Variable containing the large waveform.

I then added a 50ms timeout constant and put an indicator on the timeout flag output.

This would crash LV8.6.1 almost immediatley.

[Ben_S]

Rick_F_Foster,

Can you give us some more information about your system?

1.) What Hardware are you using?(NI Products, PC, etc)

2.) What software versions are you using (LabVIEW, OS, Real-Time Module, etc)

3.) What version of LabVIEW was this code written in?

4.) Are you running this as an executable or in the development environment?

Thanks!

[Rick_F_Foster]

The RT target is a Dell Dimesion 4400 with a NI PCI-6259 Multifunction Card (RT OS 8.6.1)

The XP Workstation is a Dell T7400

They talk to each other via a 1Gbit Ethernet switch

Labview is 8.6.1

DACmx is 8.9

I am executing from within the development environment.

[Rick_F_Foster]

Another observation is the when I run, I am getting a HUGE number of page faults assocaited with the host running LabView.exe and a LARGE number of page faults for tagsrv.eve.

[Jon_S.]

Hello Rick,

You said that you are connected via a switch.  Is there any other traffic on the switch (are you connected to the internet as well)?  If so, try connecting the RT target directly to the host via a crossover cable.

[Rick_F_Foster]

Only the Host and the RT System are on the lan side of the switch.

I can try a crossover cable, but I wonder if a intermittant problem with the switch can lead to a sudden irrecoverable kernal crash.

Rick

[Rick_F_Foster]

I tried running with just a cross over cable in place of the switch last night  and the host system still did a hard crash 20 minutes into the run. RT system is still chugging away this morning.

I have tried running  the SVE only on the RT side and only on the Host side. It is always the host side that hard crashes after some amount of time (tens of minutes to several hours).