Skip to main content

​
Mounting Storage Volumes

Mount a storage volume to your instance during deployment to access persistent data inside your container.

​
Mount During Deployment

1

Start a Deployment

Go to Dashboard → Instances and click Deploy Instance.
2

Configure Your Instance

Select your GPU, image, and other settings as usual.
3

Select a Storage Volume

In the storage section of the deployment form, select an existing volume from the dropdown. The volume will be mounted inside your container.
4

Deploy

Complete the deployment. Your volume will be available once the instance is running.
The volume must be in the same project as the instance you are deploying.

​
Accessing Mounted Storage via SSH

Once your instance is running with a mounted volume, connect via SSH and access your data: 
# Connect to your instance
ssh root@<instance-ip> -p <port>

# Your storage volume is mounted and accessible
ls /storage
The exact mount path will be shown in the instance details on the dashboard.

​
Example Workflows

Save Checkpoints

Write training checkpoints to your mounted volume so they persist after the instance stops.
# In your training script
model.save("/storage/checkpoints/epoch_10.pt")

Download Datasets

Download large datasets once to your volume and reuse them across multiple instance deployments.
# Download dataset to persistent storage
wget -P /storage/datasets/ https://example.com/dataset.tar.gz
tar -xzf /storage/datasets/dataset.tar.gz -C /storage/datasets/

Clone Repos

Clone your project repositories to the volume so your code is ready on every new instance.
cd /storage
git clone https://github.com/your-org/your-repo.git

Share Data Across Instances

Use the same volume across different instances to share model weights or processed data.

​
Best Practices

Organize with Directories

Create a clear folder structure — /storage/datasets/, /storage/checkpoints/, /storage/models/ — to keep your volume organized.

Use Symlinks

Symlink paths from your project to the storage mount to keep training scripts portable.
ln -s /storage/datasets /workspace/data

Clean Up Regularly

Remove outdated checkpoints and temporary files to avoid unnecessary storage costs.

Choose Nearby Regions

Select a storage region close to your typical instance region for faster read/write performance.

​
Troubleshooting

Make sure the volume is in the same project as the instance you are deploying. Switch projects using the sidebar selector if needed.
Verify that the volume was selected during deployment. Check the instance details page on the dashboard to confirm the volume is mounted. If not, you will need to redeploy with the volume attached.
This is usually caused by a region mismatch between your storage volume and instance. For best performance, choose the same or nearest region for both.
The volume is still mounted to a running instance. Terminate the instance first, then delete the volume.
Confirm you mounted the correct volume. If you created a new volume instead of selecting the existing one, your data will be in the original volume.