I am running a networked LabVIEW 6.0.2 system using datasocket server (version 4 (build 367)) to communicate between the computers. I am running the DSS under Windows 2000. Until recently, I was running 4 LabVIEW DS writers under Windows 98 and several LabVIEW DS readers under Windows 2000. Everything worked fine. I have now upgraded one LabVIEW DS writer to Windows 2000 and left the other three as Windows 98 machines. Periodically, the LabVIEW DS Readers crash with the message 'LabVIEW.exe has generated an error and will be closed'. I believe this to be an incompatibility between the different operating systems since the LabVIEW code has not been changed. Does any one have any information about this? For various project management reasons, I cannot rush through an upgrade of the 3 remaining DS writer computers to Windows 2000 (even though I believe this would solve the problem).
(1) We originally ran with the DSS on Win 98 but this proved very unstable (regular DS writer and reader system hangs or fatal crashes) so we upgraded to Win 2K and haven't eperienced that problem since.
(2) No problem when I do this. Problem appears to happen when I introduce a second (or more) Win 98 writer in to the system. No problem when all four writers are win 98. Also no problem if I replace the win 98 machine with a win 95 machine. Don't understand that!
(3) Problem is that the readers are also writers (about 5% of the time and to a different datastream) and that these tasks have to execute in parallel because the 'important' activity is reading and cannot wait for a write execution which may not take place for many minutes at a time.
(4) This was an amusing exercise! I predefined the data items and set all the appropriate IP numbers and just about everything failed to work after that so I went back to dynamically creating items. This failure manifests itself by the readers by returning empty data.
(5) I am sending a 1D array of about 900 I32s although I had been flattening this to a single string in a previous version of the program. Why should this make any differnce to whether or not the readers crash as a result of introducing a writer on Win 2K (given that it didn't before)?
I get the following error in the Win 2K (the reader) system's event log after the crash:
"The application, , generated an epplication error. The error occurred on 03/13/2002 @ 18:01:23.539 The exception was c0000005 at address 008B511A (inp)"
I believe that the space between the two commas should say 'Labview.exe' and the pid is 968 which appears to be an 'access violation'. (The Win 2K readers are _not_ trying to access hardware. )