Flush from cache to disk failed - data discarded

[KaBooOoom]

Hi,

I have a problem with one of our bench. At the end of a test (during the recording) we sometimes get the following error:

WARNING: flush from cache to disk failed - data discarded. (see enclosed picture)

Because of this error, our RT application lost the connection with the software on the PC and the data we wanted to record.

- Is this error due to the disk ?

- Is there a way to scan the disk ?

- Any other idea ?

OS : PharLap

Software: Labview 8.5.

[Bob_Schor]

It looks like this error was generated by the RT OS.  I presume that you are running a LabVIEW RT program on a target such as a PXI Controller.  Do you have Error Handling enabled in your RT Code?  That is to say, are you "trapping" errors and, if they occur, are you writing Error Messages to a dedicated file on your RT system's disk?

I'm not sure about the examples that shipped with LabVIEW 8.5, but recent versions of LabVIEW have Project Templates that include Real-Time Data Acquisition and Logging that shows a method for doing this.  The virtue of this, of course, is that you can then pinpoint the place in your RT Code where the Error arose, and potentially get a better idea of the nature of the root cause.  In addition, once you have an idea of where the Error can arise, you can put in Debugging Code that either adds additional information once the Error has occurred (such as writing out values of State Variables) or pro-actively writes a Debug file tracking incoming parameters to the Problem Section of code to allow you to see "Oops, how did a negative number get in here?" types of situations.

We've successfully used this to find some subtle timing "mistakes" in some of our routines.

Bob Schor

[KaBooOoom]

Hi Bob,

Thanks for your answer.

I also think this error is generated from the OS (Pharlap), and i suspect the hard drive or something like that, that's why i asked if someone knows how to check the disk.

As i wrote, this error only happen sometimes. I cannot access the code in order to add debugging informations, even if i'm not thinking this come from there (the code hasn't change since the installation 8 years ago).

[Bob_Schor]

If you are working with a PXI Controller, many (all?) of them are based on Intel PC technology -- indeed, they can be configured to run Windows (but then that kills the RTOS possibility).  In particular, this means that the hard drive on the Controller is accessible to a PC.

If you really suspect the hard drive on your controller, I'd recommend the following (relatively-safe) steps, particularly if you have ever poked around inside a PC (say, to replace a disk drive, add memory, etc.):

1. Turn off the PXI and remove the Controller (in Slot 1).
2. Find the Hard Drive, usually a 2.5" drive.
3. Remove it (carefully), being sure to save the mounting screws.
4. Mount it using a suitable "external disk" interface (or kit) on another Windows PC.
5. Run non-destructive test routines on the disk.  At a minimum, run ChkDsk in its "Check Disk Integrity, don't try to fix" mode.  Note that your Disk should show up as a FAT32-formatted disk on your PC with a letter like E: (since C: and 😧 are probably already in use).

If the disk is bad, simply replace it.  2.5" hard drives are not that expensive.  You will, of course, have to rebuild your RTOS system, but there are procedures for that (somewhere I wrote myself a Set of Instructions, if you get stuck ...).

Bob Schor

[KaBooOoom]

Thanks for the procedure to remove the disk.

I also found a tool to check the disk (i will try this at first).

http://digital.ni.com/public.nsf/allkb/E7C79C598DDF40D1862573460026FDDA

But if someone have another explanation/ idea i will really appreciate.

[Bob_Schor]

My recommendation is that you do not attempt a method that (a) is at least 5 years old, (b) NI has announced they are no longer updating/maintaining it, and (c) involves unspecified software for Seagate and Fujitsu disks (which you may or may not have in your Controller).  It also involves testing the Disk "in place" in the Controller, so what does a failure mean?  Bad disk?  Bad controller?  Bad memory in the Controller?  But suit yourself.

Bob Schor

[KaBooOoom]

I understand what you mean, but is it nor better to test the disk before removing it ?

I suspect the hardware of being responsable of the mesage, but i'm not really sure.

Removing the hard disk, replace it, reinstall the OS and the software will take more time than checking the disk with the tool... Moreover, the error happens sometimes, if i can prove the error came from the disk i will replace it. 

Thanks for your help ...