Skip to content

Releases: ml-tooling/ml-hub

1.0.0

03 Feb 15:59
Compare
Choose a tag to compare
  • simplified Kuberntes deployment - at least in our opinion ;) Modified Kubernetes artifacts/deployment to make it more consistent with our Docker-local deployment and JupyterHub in general.
  • documentation updates with regards to Kubernetes deployment
  • various bug fixes
  • altered mount path of the dynamic whitelist file

0.1.10

02 Dec 12:20
Compare
Choose a tag to compare
  • Add feature for dynamically whitelist usernames without the need to restart the hub container. See documentation for more information.
  • Fix issue with logging in the Jupyterhub config
  • Update base image ssh-proxy to version 0.1.11, which contains a bug fix in Kubernetes mode

0.1.9

22 Nov 10:27
Compare
Choose a tag to compare
  • update base ssh-proxy image to 0.1.10
  • add "@" character to characters that are replaced in authentication step
  • [Docker-local only] Allow setting shared memory size in the named servers creation screen

0.1.8

13 Nov 14:59
Compare
Choose a tag to compare
  • Add a JupyterHub service that allows cleaning of resources created by our custom spawners (see Readme for more information)
  • Show the "Info" and "SSH Access" buttons also in the home.html page

0.1.7

30 Oct 11:42
Compare
Choose a tag to compare
  • pre-install oauthenticator package to allow OAuth as Jupyterhub Authenticators (https://github.com/jupyterhub/oauthenticator)
  • add normalization steps for user names. This is especially useful for oauthenticators, where the returned username can contain special characters such as spaces or commas which can lead to problems when routing requests. These added normalization steps can be overriden, if needed
  • add update button to the hub home page, which users can use to update their default workspace server in case the hub container was updated with a new default workspace image
  • [Docker-local only] add namespace support that allows having multiple MLHub containers with different names running on the same machine, as created artifacts such as other containers are named accordingly. The name for the hub container must be given via the environment variable HUB_NAME now and not via the docker --name config anymore
  • [Docker-local only] add resource information to the "Create Named Server" page

0.1.6

02 Oct 11:50
Compare
Choose a tag to compare
  • Fix issue with Jupyterhub config base_url, e.g. c.JupyterHub.base_url=/mlhub works now.
  • HTTP is default now. SSL can be activated by setting the environment variable SSL_ENABLED to true, e.g. --env SSL_ENABLED=true.
    • In the helm deployment, ssl can be activated by passing the environment variable to the proxy config proxy.extraEnv.SSL_ENABLED=true
    • The ssh feature is not supported in non-ssl mode.
    • Like before, an unsigned ssl certificate is generated automatically if none is provided
  • In former versions, additional arguments could be passed to the chp-start command via container args. Now, this has to happen via the ADDITIONAL_ARGUMENTS environment variable, e.g. --env ADDITIONAL_ARGUMENTS="--ip=0.0.0.0 --api-ip=0.0.0.0"

0.1.5

23 Sep 13:48
Compare
Choose a tag to compare

0.1.4

20 Sep 12:07
Compare
Choose a tag to compare
0.1.4