Sometimes RC hangs in a certain state, we haven’t been able to figure out exactly why, but it’s usually on alignment or meshing. And it could be related to the network (dropping connection to the server) or the machine it’s running on. I can stall the automated processing severely, especially when it runs overnight. Our scans never take more then 1h30m to process because of the nature of our inputs and settings. So I was wondering, what is the proper way to deal with “time outs”. I would like to kill the process when it takes over 2 hours to run, kill the process and return any other exit code than 0.
Is there anything we can do in RC itself, if not what would be a good way to set it up with python?
It does seem to be a license issue. When using the PPI mode, or the non-floating Enterprise license, RealityCapture can be used offline and without an internet connection. Only for floating is the internet connection a requirement.
Stalling processing may be caused by various factors. For example: complicated geometry, issues with finding common features during the alignment, insufficient memory, etc.
We currently don’t have an error handling workflow (or in this case stalling handling workflow).
Thanks, I now have -stdConsole flag set and I’m reading the stdout stream for ‘failed’ keywords, using python process. With failed I cancel the process immediately and the process itself is run with a timeout set. That should capture most time-out and processes with failing parts.