Robyn Python API - Internal Server Error 500 | Unable to track the logs

[apoorvabc]

Project Robyn

Issue

On a linux machine, I am able to connect to the Robyn API using ! Rscript <path_to_apifile>/robynapi_call.R &. I am able to access the dummy data fine. However I am unable to run InputCollect = robyn_api('robyn_inputs',payload=payload) and it throws an Internal Server Error 500. I am unable to view the logs since there is no Robynlog.txt needed to connect to the API.

This works perfectly fine on windows machine. However I have trouble running it on Linux and unable to find the cause.

Dataset used: dummy dataset

[yu-ya-tanaka]

Hi @apoorvabc, Can you share the demo jupyter you are running? You may not be executing the correct cell. Or can you try MAC option to use subprocess.Popen()?

[apoorvabc]

# Constants
API_IP = '127.0.0.1'
TIMEOUT = 1
WAIT_TIME = 5

# Logger setup
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

def is_port_open(port):
    command = "sudo netstat -tulpn | grep :{}".format(port)
    process = subprocess.Popen(command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    output, _ = process.communicate()

    return process.returncode == 0 and (":{}".format(port) in output.decode())

# Example usage:

def terminate_port(port):
    # Find the process using the specified port and terminate it
    try:
        pid = subprocess.check_output(['lsof', '-ti', f':{port}']).decode().strip()
        os.kill(int(pid), signal.SIGTERM)
        print(f"Process using port {port} terminated")
    except subprocess.CalledProcessError:
        print(f"No process found on port {port}")

def linux_run_and_check_port(r_script_path, port):
    # Run R script
    print (os.getcwd())
    rscript_command = f"Rscript {r_script_path}"
    print("rscript_command is", rscript_command)
    subprocess.Popen(rscript_command, shell=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
    time.sleep(WAIT_TIME)

    # Check if the port is open
    if is_port_open(port):
        print(f"R script is running on port {port}")
    else:
        print(f"Failed to start R script on port {port}") 

Just an FYI I am able to run the Allocator function using the same API connection by using the saved payload allocator_response = robyn_api('robyn_allocator', payload=payload)

[yu-ya-tanaka]

Is issue solved? If not, are you using customized notebook? Can you use our sample notebook?

[ebyjose]

@apoorvabc : Could be a API initiaitiation issue on linux. Can you append a message file to end of linux api start command and share that file?

[apoorvabc]

Hello @ebyjose, I was able to extract the logs and figured out there was an issue with the nevergrad installation on the docker container. Managed to fix the installation and I am able to build the model fine. This now stands resolved.