Skip to content

Storage usage

Yurii Rebryk edited this page Jan 13, 2019 · 3 revisions

The Neuromation Platform provides you with personal storage, which allows you to:

  • Persist data generated by and used by jobs. For example, datasets, trained models etc
  • Share data among multiple jobs

1. Manage your data

You can use neuro store command to:

  • List directory contents
  • Make directories
  • Copy files and directories
  • Remove files or directories
  • Move or rename files and directories

Let's copy dataset.zip from our computer to the storage:

$ neuro store cp dataset.zip storage://<USERNAME>/

Use neuro store ls to list storage contents:

$ neuro store ls storage://<USERNAME>/

Example output:

directory      0              kaggle
file           1,183,714      dataset.zip

Be wise take a tip
Instead of storage://<USERNAME>/ you can use just storage:///

2. Mount the storage

Add --volume flag when you submit a job to mount the storage.

The following options will mount your personal storage to the data directory inside the container:

$ neuro job submit --volume storage://<USERNAME>:/data:rw registry.staging.neuromation.io/<USERNAME>/platform-api-hello-world:latest
Clone this wiki locally