cRIO corrupting USB drives

[icontech]

Hi,

We have started using labview 2011 for several compactRIO embedded data logging applications. Since moving to the new version we have had multiple disk failiures which were not present on the same hardware (USB drive and cRIO controller) deployed from LV2010. The problem is related to unexpected power failiures on the cRIO. With one usb stick, the drive fails to mount after 3 or 4 reboots. The stick can be refreshed in to a working state by pluging and unpluging the stick into a laptop. The second type of USB we have tried (Corsair survivor GTR 64GB) fails in a state where it no longer mounts on anything. We have currently killed 2 survivor USB drives in as many weeks. This suprised me as these stick are meant to be industrially rated and are backed by a 10 year warranty.

Are there any known issues logging to USB with LV2011 on cRIO controllers? I will continue to test to try and isolate a minimum configuration which reliably reproduces the problem.

Thanks,

Nick.

[Andrew_B]

Hello Nick,

This is Andrew Brown, an Applications Engineer from National Instruments. After researching the issue, I have not been able to locate similar cases where cRIO controllers caused USB sticks to no longer mount. To help narrow down the problem, I would appreciate if you could answer the following troubleshooting questions:

1. What cRIO controller are you using?

2. What is the first type of USB stick that you are using?

Regards,

Andrew Brown

[icontech]

Hi,

Thanks for following up.

We have had the problem on both 9014 and 9024 controllers. The stick that can be refreshed is a Kingston DataTraveler 200 64GB. The stick we have killed twice is the Corsair Survivor GTR 64GB. Our clients require a capacity of at least 64GB. The Corsair was desirable because it is significantly faster on the cRIO than any other stick we have tested.

We have setup a test rig which automatically cycles the power on the cRIO every 7 minutes and leaves it off for 2 minutes. The Kingston usually dies within 3 or 4 reboots. The Corsair survived overnight test of over 100 reboots twice before failing. We have also tested a 16GB Lexar for about 8 hours which has not yet died.

I plan to isolate the disk logging part of the code so that I can more easily compare against earlier labview versions and supply code to NI for your own testing but have been busy with other tasks so far.

I appreciate this is a difficult problem to trace as it is the interaction between the cRIO and various 3rd party USB sticks but we need to give our clients the confidence to deploy cRIO based systems on long term unattended projects.

Regards,

Nick.

[ba_0101]

Is data being written continuously to the USB drive? Was data being written when the power was cycled?

[icontech]

Data is being continuously written in blocks of 1 second or so. The file is closed between block writes. We have no control of when we lose power as it occurs when the machinery being monitored is isolated. We are aware that writing to the disk while the power is lost is likey to result in the current file being corrupted but in previous projects it hasn't left the disk in an unmountable state.

[A_Person]

Hi Nick:

Have you had a chance to look into this any further and isolate the disk logging section of your code?  Are you continuing to see the problem?  I really haven't heard of other customers experiencing this problem, especially after upgrading to LV 2011, but I would definitely like to do some more investigation.

When you upgraded to LV 2011, I assume you updated all of the software components on the controller including NI-RIO 4.0 and NI-VISA 5.1 (I mention these two as they're more relevant to your issue).  Can you confirm the exact software versions you were running under LV 2010 and what's on the controller now?

With the specific USB drives you mentioned, are they all using the same filesystem?  I would guess FAT32 but again, could you specify this?

[QFang]

We have a very similar use case, with 9014's, that will deploy shortly. Any updates or additional information on this possible problem would be hugely appreciated. 

Based on the thread openers data, it seems to indicate that lower capacity sticks are less likely to get "killed" in this manner? We would only need a 2GB stick for our deployment.

Thanks for any updates you may have!

Q 

[Matt.M]

Hi QFang,

I don't believe you should have any issues logging to a 2GB usb drive.  If you deploy, and your USB drive becomes unmountable after some amount of reboots please let me know so that I can look futher into this issue.  

If you do see this issue please include versions of the relevant software you have installed on the host computer, and the cRIO (NI-RIO, LabVIEW, Real-Time Module), and what model cRIO and modules you are using in your reply.  Also if you can please include a screenshot or vi of the code that logs data that would be very helpful in troubleshooting.

Thanks,

Matt M.

[icontech]

Hi Alex,

Thanks for reviving the thread. I haven't had a chance to test further as I have been busy on other projects - I will try and get some time in the next few weeks. The issue has only been observed on 64gb sticks so far. The disks are formatted in FAT32 with the HP USB Format Utility as Windows 7 will only format a 64gb drive as ex-fat. We are using the base NI-RIO4.0 set with the SV engine added - I can give the specific versions of modules when I get back to testing. The logging uses the TDMS palette and files are spanned at a fixed size.

Regards,

Nick

[TinoMK]

Hi,

I have a similar problem. I have a cRIO 9012 controller that is continuously logging data to USB (FAT32 formatted 32GB capacity).

After a few days I see that somehow the USB is not present at the system anymore (it is dismounted) and the cRIO continued to write data to local memory. If I unplug the USB and plug it again it is working OK but then after a few days same thing happens.

cRIO is running RT 2010SP1.

Any ideas? Thanks,