Optimizing Zero-Downtime for Phoenix Server Inference Updates

Hi, I'm trying to run the phoenix server using px.launch_app inside a while loop to keep it alive, I'm doing this because I want the inferences(model page) to be shown on the UI with tracing and evaluation. Code which I'm using is as follows:

# create `inference_df` and `trace_dataset` before loop

while True:
    px.close_app()
    px.launch_app(inference_df, trace=trace_dataset, port=6006, host="0.0.0.0")

Is there an optimized way to run the server with updating inferences and experience zero-downtime?