Spider load/stress

I’m using a Web service polling scenario, which is basically this scenario: .NET Gadgeteer Relay Actuator via REST Web Service .NET Gadgeteer Relay Actuator via REST Web Service | Integral Design

The Gadgeteer application polls a web service for state information every 40 seconds. It sent a non-blocking http GET request on this interval but the system stopped working over night.

After running all night, the Spider is unresponsive. I couldn’t even re-deploy binaries from Visual Studio until I rebooted the Spider manually.

The WiFi module was still apparently working because a LED should have turned red on Network down, although the Network down event may have been missed for the same reason everything else was non-functional.

Does anybody knows what kind of limit this is hitting?

I have the application set to a 3 minute polling interval now to see if it can handle this over several hours.

the state you described generally means the system has crashed after an exception.

So it seem that a catch block, if I know where the error occurs, will solve this problem in the long running scenario.

Try the Generic Exception Logger i published earlier today…
http://www.tinyclr.com/forum/1/6896/

it will allow you to log your exception to a Text File or post them to the web or whatever destination you like…

this way you can later inspect the file…

Cheers,

Jay.

Thanks, Jay.

Mike

Have you tried running for a long time while connected to the VS debugger?

Yes, the device was plugged into USB and connected to VS when it hung. All I saw as a list of threads that stopped, not any exception output:
The thread ‘’ (0x4c) has exited with code 0 (0x0).
The thread ‘’ (0x4d) has exited with code 0 (0x0).
The thread ‘’ (0x4e) has exited with code 0 (0x0).
The thread ‘’ (0x4f) has exited with code 0 (0x0).
The thread ‘’ (0x50) has exited with code 0 (0x0).

These are the response received threads. Whatever went wrong to hang didn’t leave any tracks.

What power source are you using when the device is working standalone?

Now with the device connected to USB for power only, not debugging, it does not hang. It ran for 24 hours so far.
I encountered some very strange behavior when I tried to build in release mode! It messed things up so I could not deploy to the Spider board at all.
It took a lot of reboots to clear this up. I took out most of the debugger printouts and finally was able to rebuild in debug mode. Then I hooked the USB to another machine for power, without debugger running. It has been running ever since without problems.