TDMS write result in empty files for 20 minutes every 24h

[felixturner]

Hello everybody!

First the specs:

• LabView 2009, pretty much all available toolkits
• NI DAQ 4462 pci card
• HP Z600 Workstation
• 12 GB RAM
• 2 x Intel Xeon processors, total 12 cores
• 2 x Seagate Barracuda 1TB 7200 (RAID 1) for data storage
• 1 Seagate Barracuda 1TB 7200 for OS etc.

I have the equipment set up to constantly monitor a system for periods of 4-5 days, streaming the data (waveforms) from 2 channels collected at a sampling rate of 102400 S/s into TDMS-files. In order to have easy access to data from any given moment I have set the software to write a new TDMS-file every minute, which gives me files of almost 200 mb.

The problem

Saving the data works very well most of the time, except for the fact that every now and then i get empty files. This happens at seemingly random intervalls and then lasts for 20-40 minutes before everything goes back to normal. Typically I get 23 hours of good data and 1 hour of empty files every day. At first I figured it might be some automatic update in windows, Java etc. that messed with the system but after turning off all of these function the fault remains. I've completely stripped the OS of unecessary functions as well, with no effect.

Has anyone out there had any similar experiences? I would be extremely grateful for any help.

Regards,

Felix

[Ben]

Can't say I have seen that although I have seen some other things (that was only able to stop by getting the operators to pull the network cable durring testing).

Questions;

Are you flushing before close?

Are you handling errors?

Is indexing turned on for the disks?

Anything interesting in the system log file during the time of the failures?

Ben

[felixturner]

Thank you for your answer. Sorry for the delay of mine, I've been busy with other projects.

To answer your questions:

Are you flushing before close?

Yes. Always.

Are you handling errors?

 Yes, more on that later.

Is indexing turned on for the disks?

Yes. And there seems to be nothing wrong with the files that do get written.

Anything interesting in the system log file during the time of the failures?

Nothing strange in the system log. I've inserted an errorlog in the program and it records that for some reason the DAQmx-read function times out when the problem occurs. The program then closes all DAQmx-tasks in the producer loop and starts over but the error remains for several minutes and the suddenly everything runs smoothly again.

When I replace the TDMS write vi with the write to spreadsheet vi or just plain text files the error does not occur and all files are written correctly.