Cannot run scripts (code stuck on docker container creation)

[Phan-Tom416]

Hi,

I was trying to run the example code to test out the installation on Windows 11 system: diambra run -r D:\RL python diambra_arena_gist.py

However the code get stuck after creating the container in docker, after this output: service is listening on port 0.0.0.0:xxxxx

How can I fix this issue?

Thanks

[alexpalms]

Hi @Phan-Tom416,

a few questions: 1) Which terminal are you using? 2) Are you using a virtual env? If yes, which one? 3) Can you share here the pip list output? 4) Have you downloaded and verified the roms?

It can be easier if you reach out on discord https://discord.gg/tFDS2UN5sv to iterate or schedule a call.

[Phan-Tom416]

Hi @alexpalms,

Answering each question:

1. I am using Windows PowerShell, also tried with Command Prompt.
2. Not using virtual environment.
3. Pasting the output of pip list below:

Package Version

---

anaconda 0.0.1.1 async-generator 1.10 attrs 21.2.0 certifi 2021.10.8 cffi 1.15.0 cloudpickle 2.2.1 colorama 0.4.4 cryptography 35.0.0 cycler 0.11.0 dacite 1.8.0 diambra 0.0.11 diambra-arena 2.1.0rc4 diambra-engine 2.1.0rc7 distro 1.8.0 findspark 2.0.1 fonttools 4.28.3 grpcio 1.51.1 gym 0.21.0 h11 0.12.0 idna 3.3 inputs 0.5 kiwisolver 1.3.2 lxml 4.6.4 MAMEToolkit 1.1.0 matplotlib 3.5.0 mysql 0.0.3 mysqlclient 2.1.0 numpy 1.21.4 opencv-python 4.5.4.60 outcome 1.1.0 packaging 21.3 pandas 1.3.4 Pillow 8.4.0 pip 22.0.3 protobuf 3.20.3 py4j 0.10.9.3 pycparser 2.21 pyOpenSSL 21.0.0 pyparsing 3.0.6 pyspark 3.2.1 python-dateutil 2.8.2 pytils 0.3 pytz 2021.3 PyYAML 6.0 scipy 1.7.2 screeninfo 0.8.1 selenium 4.0.0 setuptools 57.4.0 setuptools-scm 6.3.2 six 1.16.0 sniffio 1.2.0 sortedcontainers 2.4.0 spark 0.2.1 tk 0.1.0 tomli 1.2.2 tools 0.1.9 tqdm 4.62.3 trio 0.19.0 trio-websocket 0.9.2 urllib3 1.26.7 wheel 0.37.1 wsproto 1.0.0

4. I have the roms ready and checked with no issue.

Thanks

[alexpalms]

Thank you for the feedback @Phan-Tom416 All seems normal. The best thing is to jump in a call and take a look together. Please reach out to me on Discord and we will schedule a call.

[alexpalms]

A few additional questions @Phan-Tom416 : 1) How long have you waited for the script to start? It can take a while 2) Are you sure docker has the proper permissions on Win?

[discordianfish]

This seems to be a docker-for-win issue with the portmapping, we did the following already to isolate the issue:

• [x] Run docker run -p 80:80 nginx -> Reachable on localhost:80
• [x] Run docker run -p 127.0.0.1:80:80 nginx -> Unreachable
• [x] Run engine with net=host without cli as described here and...
  • [x] Connect to localhost:50051 -> failed
  • [x] Connect to host.docker.internal:50051: 192.168.101.23:50051: connectex: No connection could be made because the target machine actively refused it.
• [x] diambra run --env.host=0.0.0.0 and..
  • [x] grpcurl -plaintext 172.17.0.3:50051 list: Failed to dial target host "172.17.0.3:50051": context deadline exceeded
  • [x] grpcurl -plaintext localhost:59667 list: Failed to dial target host "localhost:59667": dial tcp [::1]:59667: connectex: No connection could be made because the target machine actively refused it."
  • [x] grpcurl -plaintext host.docker.internal:59667 list: Failed to dial target host "host.docker.internal:59667": dial tcp 192.168.101.23:59667: connectex: No connection could be made because the target machine actively refused it.
  • [x] diambra test diambra/agent-random-1:main -> works!
  • [x] Manually connect to grpc from container as described here

[alexpalms]

Do you have any update on this @Phan-Tom416 ?

[discordianfish]

Since docker isn't responding to the issue filled, the best option is to use a linux VM. Alternatively you could try using podman: https://podman.io/docs/installation but you might need to configure it to use the default docker socket location until we support that officially in the cli.

[alexpalms]

Closing the issue for inactivity