Possible TCP/IP bug in LV 6.02

[charles@pentek.com]

I've a vi written in LV 6.02 that connects to a socket on a remote server. When the connection is established, the host streams data until the connection is broken. The server is one or more processors on a Quad PowerPC VMEBus board, connected via ETHERNET.

The client is running on a 2GHz Pentium with 512MB of RAM.

Each new connection spawns a template vi that displays the streamed data on a waveform graph.

If each instance of the client vi connects to a different remote processor, I am able successfully to stream data from each remote connection.

If, however, a single host cpu opens multiple sockets, the second client instance times out on its first read, after which the original instance also times out.

We've made multipl
e connections to sockets on a single processor of the remote board before using normal Winsock calls from VC++ or Cygwin, and we've a ton of experience with the OS on the other end of the connection (vxWorks)

I'm using LV 6.02 Professional Development System. I get the same result whether I run two instances of my template vi or if I run two distinct client vis. Running the executable as opposed to running under the development environment doesn't seem to make any difference.

Any thoughts?


Charles Krug

[JRA]

I'm not completely clear on how your application is set up, but you certainly could have found something. Try writing a small application which exhibits the behavior and posting it here. I'll take a look at it and see if I see the same behavior on my machine in more recent versions.

[Ben]

Does using unique ports for each help?

Ben

[Spaz]

A wild stab here:

If you are using an array to hold references to your multiple sockets and you are accessing them by auto indexing at a for loop, try using a while loop with a shift register input for the references and indexing the references individually within the loop.

I've had similar problems with dynamic datasockets disconnecting at a for loop. NI said that's the way it is.