Skip to content

YTsaurus k8s operator 0.4.0

Compare
Choose a tag to compare
@savnadya savnadya released this 26 Sep 13:09
· 246 commits to main since this release

Features

  • The operations archive will be updated when the scheduler image changes.
  • Ability to specify different images for different components.
  • Cluster update without full downtime for stateless components was supported.
  • Updating of static component configs if necessary was supported.
  • Improved SPYT controller. Added initialization status (ReleaseStatus).
  • Added CHYT controller and the ability to load several different versions on one YTsaurus cluster.
  • Added the ability to specify the log format (yson, json or plain_text), as well as the ability to enable recording of structured logs.
  • Added more diagnostics about deploying cluster components in the Ytsaurus status.
  • Added the ability to disable the launch of a full update (enableFullUpdate field in Ytsaurus spec).
  • The chyt spec field was renamed to strawberry. For backward compatibility, it remains in crd, but it is recommended to rename it.
  • The size of description in crd is now limited to 80 characters, greatly reducing the size of crd.
  • Query Tracker status tables are now automatically migrated when it is updated.
  • Added the ability to set privileged mode for exec node containers.
  • Added TCP proxy.
  • Added more spec validation: checks that the paths in the locations belong to one of the volumes, and also checks that for each specified component there are all the components necessary for its successful work.
  • strawberry controller and ui can also be updated.
  • Added the ability to deploy http-proxy with TLS.
  • Odin service address for the UI can be specified in spec.
  • Added the ability to configure tags and rack for nodes.
  • Supported OAuth service configuration in the spec.
  • Added the ability to pass additional environment variables to the UI, as well as set the theme and environment (testing, production, etc.) for the UI.
  • Data node location mediums are created automatically during the initial deployment of the cluster.