Code Hangs with Timeout Error in R when Calling Gurobi through Python API using reticulate

Hello,

I am calling Gurobi using the Python API from R using the reticulate package. In R, I have wrapped my Python call in outer functions as I perform a series of computations. However, I want to skip over certain groups of computations if they collectively take too long (say over 600 seconds). So, I am using the base::withTimeout function in R, wrapped in the try function to deal with the timeout errors and skip them.

However, occasionally, the program will not properly timeout a computation and properly handle the typical timeout error. Instead, I will get an unusual error that prints to the console (which should not happen with the try function). But unlike a typical R error, it does not terminate the code. Instead, it will hang, sometimes for a minute or two, but other times for hours (substantially longer than the typical 100-second Gurobi TimeLim parameter I typically set). I am not sure why it does this. The error is:

`Error in .Call(`_reticulate_py_call_impl`, x, args, keywords) : reached elapsed time limit`

If I set the Gurobi time limit to 1 second, it does not cause this error anymore. As far as I can gather, I only get this error if the function attempts to time out when Gurobi is in the middle of solving a model. This is likely a problem with how R interacts with calling a Gurobi-containing Python function using reticulate (rather than a problem with Gurobi itself). However, I hoped anyone would know how to stop the code from hanging. Even just skipping the current pass of the loop would be great.

Code Hangs with Timeout Error in R when Calling Gurobi through Python API using reticulate

Comments

Didn't find what you were looking for?