Rasa shell segmentation fault

Rasa Open Source
1

Hi there,

I used clean ubuntu 16 and ubuntu 18 server in wmware/virtualbox (8GB RAM) with python 3.6 and pip 20 in virtualenv to install rasa by github source. It built without errors. I tried to init a new project and speak to bot server during installation and good. After that, i enter the command with rasa shell, the segmentation fault occurred. It’s very strange that it works during the init project but fail with command rasa shell.

[Album] rasa shell seg fault

Can you help me, thanks!

2

Hey @samshaw. Looks like the issue here is related to your OS. I found the thread here which might be useful for you. Could you try some of the solutions mentioned here?

3

Thanks for help. I have tried the solution mentioned there with several times.

I did that in different computers with clean installation from ubuntu server 18 server LTS ISO file and follow the instructions: https://rasa.com/docs/rasa/user-guide/installation/ by using from git source (because using pip install rasa will cause an sanic? package version dependency issue as forum already recolonized at this time)

$ sudo apt update $ sudo apt install python3-dev python3-pip $ sudo pip3 install -U pip

$ git clone GitHub - RasaHQ/rasa: 💬 Open source machine learning framework to automate text- and voice-based conversations: NLU, dialogue management, connect to Slack, Facebook, and more - Create chatbots and voice assistants $ cd rasa $ pip install -r requirements.txt $ pip install -e .

and rasa will installed successful without any errors.

However, I try rasa init (good, no problem) rasa shell (seg fault) test example restaurant and input rasa train (seg fault)

PS. I install ubuntu OS into windows vmware and virtualbox with 3 computers. all the same errors.

4

After testing with several computers and VM, finally I got the root cause: It’s the dependency issue, please just

pip uninstall gym pip install gym==0.15.4

then everything goes well. BTW, gym version was changed to 0.15.6 from 3th Feb, 2020. So, It’s the hot issue, isn’t it? ha ha

5

@samshaw good point, that would explain why we’ve been seeing so many more of these lately!