Backup PVCs created by jupyterhub to S3(the only backend supported right now).
This project will launch a kubernetes job which determines all of the jupyterhub
user PVCs in the given namespace. It does this by looking for PVCs which have
the claim- prefix. This was found to be reliable enough for our installation
of zero-to-jupyterhub.
Once it knows which PVCs should be backed up, it creates a kubernetes job for
each of these which backs up the user's home directory to the backend of
choice(only S3 for now).
make test
make install
The easiest way to deploy this right now is with the helm chart,located in
chart/jupyterhub-kubernetes-backup. At this time, no docker image is published
for this project, so you will need to publish your own after running: make docker-image. You'll then need to push that image to your docker registry of
choice and reference it below in the values.yaml override. A minimal
installation will override the following in a custom values.yaml:
image:
repository: <where you pushed the docker image>
tag: <the tag you pushed to>
backend:
type: s3
s3:
bucket: <bucket you want the files backed up to>
prefix: <the prefix in the bucket you want to backup to>
accessKey: <an accessKey that has access to write to the bucket>
secret: <the secret for the accessKey>
region: <the region the bucket is located in>
cronJob:
schedule: <the schedule that you want to run it on>