It is too slow too reply when I run in location pc

Rasa Open Source
1

This is my first question maybe I say not good , but it will be better .ok now I will say my question. when I run my project on my notebook computer (CPU I7 8750H GTX 1060 RAM 32G) if the answer need to predict it will be slow in action . I found the same [question](http://forum.rasa.com/t/rasa-nlu- slow-response-on-localhost-machine/10254) but I don’t think it can help me . I found some is slowly such as

processor.py 91

      processor.py 91
       wupj_start_time = t.time()
        await self._predict_and_execute_next_action(message, tracker)
        print("processor_91 cot:", (t.time() - wupj_start_time))

action.py 402

 wupj_start_time = t.time()
            response = await self.action_endpoint.request(
                json=json_body, method="post", timeout=DEFAULT_REQUEST_TIMEOUT
            )
            print("self.action_endpoint.request(action402) cost:", (t.time() - wupj_start_time))

            wupj_start_time = t.time()
            should_predict_another_action = await self._run_action(
                action, tracker, message.output_channel, self.nlg, policy, confidence
            )
            print("368-->", action)
            print("processor_362:_run_action:", (t.time() - wupj_start_time))
           but I also print time in my actions but  my action not need much time

can any one give me some suggests. what is wrong with me and how can I do .

2

Hi @BigDuck. It’s a great question :slight_smile: To try and help you I would need to know which Rasa version are you running and how much training data do you have for NLU and dialogue models?

3

thank you for your answer .my rasa version is 1.1.8 and my model(core+nlu) only 2.1M .

4

Hmm, that shouldn’t cause too much trouble. Do you have any custom actions that might take the long time to execute - some backend integrations (api calls, etc)?

5

I have my actions but I print the time and it 's only cost less than 0.5s.

6

Do you think you could share the project with me so I can replicate? If it something you don’t want to share here, you can shoot me an email at juste@rasa.com

7

ok no problem ，I will send my project to you in my free .if there are something wrong or unreasonable please help me point out. It’s my 'hello world ’ with rasa . thank you very much

8

Hey, did you guys figure out what the problem was? I am facing pretty much the same problem and would love to know what the solution was

9

I figured it out for my case. I am running it on a Debian server and all I had to do was change the endpoint configuration in the endpoints.yml from “http://localhost:5055/webhook” to the server’s local ip address. Still a bit baffled though because it worked on the local environment and also on the Debian virtual machine without having to change it.

1 Like