search

ml-tooling / ml-workspace

🛠 All-in-one web-based IDE specialized for machine learning and data science.
https://mltooling.org/ml-workspace
Apache License 2.0
3.42k stars 446 forks source link

Data Loss when closing docker #82

Closed minhson-kaist closed 3 years ago

minhson-kaist commented 3 years ago

I closed the jupyter notebook GUI and start it again, and all of my data disappeared. Are there any reasons for that?

raethlein commented 3 years ago

Hey @detuvoldo, could you show the Docker command that you used to start the workspace, please? And could you also elaborate a little bit more on the "closed the jupyter notebook GUI" part? Do you mean removing the container or killing the process or closing the browser window or something like that?

minhson-kaist commented 3 years ago

I simply use the command for GPU flavor docker run -p 8080:8080 --gpus all mltooling/ml-workspace-gpu:0.12.1. I use Ctrl+C to kill the process and start the image again. Then, all the data I uploaded to this image disappear.

raethlein commented 3 years ago

... and start the image again

Do you mean executing the command docker run -p 8080:8080 --gpus all mltooling/ml-workspace-gpu:0.12.1 again or to start the exited container again via docker start $container_id (where $container_id refers to the id of the exited container which you get, for example, via docker ps -a? When restarting the same container, all data should actually be persisted. Though, maybe our persist-data section helps.

minhson-kaist commented 3 years ago

Oh thanks for pointing it out. I found my problems.