Skip to content

Using envd Environment

Here we use the example at envd/examples/mnist.

Build your environment

Following previous tutorial, you'll be able to setup the project environment by envd. In this tutorial, we used mnist project. After setting up the project, you can check the environment status by envd envs list.

$ envd envs list
NAME    JUPYTER                 SSH TARGET      CONTEXT                                         IMAGE           GPU     CUDA    CUDNN   STATUS          CONTAINER ID 
mnist   http://localhost:48484   mnist.envd      /home/ubuntu/workspace/envd/examples/mnist      mnist:dev       false   <none>  <none>  Up 2 hours      f3452559ee20

TIP

envd will add a new ssh entry to your ssh config to help you connect to your environment, named <project_name>.envd.

Connect to the environment with VSCode

First, you need to install the remote development kit in VSCode. You can find here Remote - SSH to install the extension. After the extension is installed, you'll be able to see the UI below by clicking the downside left icon.

Then click the Connect to Host option and select the target you want to connect to.

After selecting the target, open the folder with your project name (here we use mnist)

Then everything is done!

Connect to the environment with Jupyter Notebook

In build.envd file, we've declared the jupyter environment, with no password.

python
def build():
    ...
    config.jupyter()

You can also get the jupyter port through envd envs list command.

bash
$ envd envs list
NAME    JUPYTER                 SSH TARGET      CONTEXT                                         IMAGE           GPU     CUDA    CUDNN   STATUS          CONTAINER ID 
mnist   http://localhost:48737   mnist.envd      /home/ubuntu/workspace/envd/examples/mnist      mnist:dev       false   <none>  <none>  Up 2 hours      f3452559ee20

Then you can directly get your jupyter notebook by using http://localhost:48737 in your browser.

Connect to the environment with RStudio server

You can set up RStudio server with config.rstudio_server.

python
def build():
    ...
    config.rstudio_server()

Endpoint can be got through envd envs list command.

$ envd envs list
NAME    ENDPOINT                        SSH TARGET      IMAGE           GPU     CUDA    CUDNN   STATUS          CONTAINER ID 
r-basic rstudio: http://localhost:34621 r-basic.envd    r-basic:dev     false   <none>  <none>  Up 6 hours      1eb7d40e5a8a

Then you can connect to the RStudio server by using http://localhost:34621 in your browser. Please use envd as the username, and the password can be any string.

Released under the Apache-2.0 License. Built with VitePress.