Skip to main content

Filebrowser

Filebrowser provides a web UI to upload and download files to and from a persistent volume on the DSRI. It is useful when your application does not have a built-in file manager. Note that JupyterLab, RStudio, and Visual Studio Code already include a file browser, so you only need this for other deployments.

caution

Filebrowser can only be deployed on an existing Persistent Volume Claim. Make sure you have one before deploying. Contact the DSRI team at rcs-ub@maastrichtuniversity.nl if you need one created.

Deploy​

Find the File Browser for existing storage template in the DSRI Catalog (make sure the Templates checkbox is checked) and instantiate it with the following parameters:

  1. Application name - must be unique within your project. It will be used to generate the application URL.
  2. Password - hash the password first for security using:
   docker run filebrowser/filebrowser hash mypassword
  1. Storage name - the name of the Persistent Volume Claim (PVC) to expose. Find it under Deployments > Storage panel, or in the DSRI web UI under Administrator view > Storage > Persistent Volume Claims.
  2. Storage subpath - the subfolder within the PVC to expose. Leave empty to expose the root of the volume.
Default credentials

The default username is admin and the default password is admin. Change the password immediately after first login via the Filebrowser web UI.

Persistent storage​

Filebrowser mounts directly onto an existing PVC - it does not create new storage. To connect multiple applications to the same storage, add the existing PVC to each application via Deployments > Storage panel.

For more details on creating and connecting persistent storage, see the Storage documentation.

Particularities​

Storage access modes​

The DSRI uses OpenShift Container Storage (OCS) based on Ceph, which offers two access modes:

  • ReadWriteMany (RWX) - the volume can be mounted on multiple nodes simultaneously. Use this when possible.
  • ReadWriteOnce (RWO) - the volume can only be mounted on one node at a time. If that node fails, the volume cannot be remounted on another node until the pod is force-deleted.