Skip to content

CLI Reference#

This page provides documentation for wanna command-line tool.

wanna#

Complete MLOps framework for Vertex-AI

Usage:

 [OPTIONS] COMMAND [ARGS]...

Options:

  --install-completion  Install completion for the current shell.
  --show-completion     Show completion for the current shell, to copy it or
                        customize the installation.

components#

Managing kubeflow components templated generation.

Usage:

 components [OPTIONS] COMMAND [ARGS]...

create#

Create a kubeflow component based on a template.

Usage:

 components create [OPTIONS]

Options:

  --output-dir PATH  The output directory where wanna-ml repository will be
                     created  [required]

init#

Initiate a new wanna-ml project from template.

Usage:

 init [OPTIONS]

Options:

  --output-dir PATH               The output directory where wanna-ml
                                  repository will be created  [required]
  -t, --template [sklearn|blank]  Choose from available repository templates
                                  [default: sklearn]

job#

Plugin for building and deploying training jobs.

Usage:

 job [OPTIONS] COMMAND [ARGS]...

build#

Create a manifest based on the wanna-ml config that can be later pushed or run.

Usage:

 job build [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one job from your wanna-ml yaml
                      configuration to build. Choose 'all' to build all jobs.
                      [default: all]

push#

Build and push manifest to Cloud Storage.

Usage:

 job push [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -v, --version TEXT              job version  [default: dev]
  -n, --name TEXT                 Specify only one job from your wanna-ml yaml
                                  configuration to push. Choose 'all' to push
                                  all jobs.  [default: all]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]

report#

Displays a link to the cost report per wanna_project and optionally per job name.

Usage:

 job report [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one job from your wanna-ml yaml
                      configuration to report. Choose 'all' to report all
                      jobs.  [default: all]

run#

Run the job as specified in wanna-ml config. This command puts together build, push and run-manifest steps.

Usage:

 job run [OPTIONS]

Options:

  -f, --file PATH        Path to the wanna-ml yaml configuration  [env var:
                         WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT     Name of the GCP profile you want to use. Profiles are
                         loaded from wanna-ml yaml config and (optionally)
                         from this file too.  [env var:
                         WANNA_GCP_PROFILE_NAME; default: default]
  -v, --version TEXT     job version  [default: dev]
  -n, --name TEXT        Specify only one job from your wanna-ml yaml
                         configuration to run. Choose 'all' to run all jobs.
                         [default: all]
  -hp, --hp-params PATH  Path to the params file in yaml format
  -s, --sync             Runs the job in sync mode

run-manifest#

Run the job as specified in the wanna-ml manifest.

Usage:

 job run-manifest [OPTIONS]

Options:

  -v, --manifest TEXT    Job deployment manifest
  -hp, --hp-params PATH  Path to the params file in yaml format
  -s, --sync             Runs the pipeline in sync mode

stop#

Stop a running job.

Usage:

 job stop [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one job from your wanna-ml yaml
                      configuration to stop. Choose 'all' to stop all jobs.
                      [default: all]

managed-notebook#

Create, delete and more operations for managed Workbench (Jupyter notebook).

Usage:

 managed-notebook [OPTIONS] COMMAND [ARGS]...

build#

Validates build of notebooks as they are defined in wanna.yaml.

Usage:

 managed-notebook build [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -v, --version TEXT  managed notebook version  [default: dev]

create#

Create a Managed Workbench Notebook.

If there already is a notebook with the same name in the same location and project, you will be prompt if you want to delete the existing one and start a new one.

When the notebook instance is created, you will be given a URL link to JupyterLab.

Usage:

 managed-notebook create [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify only one managed_notebook from your
                                  wanna-ml yaml configuration to create.
                                  Choose 'all' to create all
                                  managed_notebooks.  [default: all]
  -v, --version TEXT              managed notebook version  [default: dev]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]

delete#

Delete a Managed Workbench Notebook.

Usage:

 managed-notebook delete [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one managed_notebook from your wanna-ml
                      yaml configuration to delete. Choose 'all' to delete all
                      managed_notebooks.  [default: all]

push#

Push docker containers. This command also builds the images.

Usage:

 managed-notebook push [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify only one managed notebook from your
                                  wanna-ml yaml configuration to push. Choose
                                  'all' to push all managed notebooks.
                                  [default: all]
  -v, --version TEXT              managed notebook version  [default: dev]
  -m, --mode [all|manifests|containers|quick]
                                  Managed-Notebook push mode, due to CI/CD not
                                  allowing to push to docker registry from GCP
                                  Agent, we need to split it. Notebooks
                                  currently support only containers, as we do
                                  not create manifests as of now.  [default:
                                  containers]

report#

Displays a link to the cost report per wanna_project and optionally per instance name.

Usage:

 managed-notebook report [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one managed_notebook from your wanna-ml
                      yaml configuration to report. Choose 'all' to report all
                      managed_notebooks.  [default: all]

sync#

Synchronize existing Managed Notebooks with wanna.yaml

  1. Reads current notebooks where label is defined per field wanna_project.name in wanna.yaml
  2. Does a diff between what is on GCP and what is on yaml
  3. Create the ones defined in yaml and missing in GCP
  4. Delete the ones in GCP that are not in wanna.yaml

Usage:

 managed-notebook sync [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  --force                         Synchronisation without prompt
  -v, --version TEXT              managed notebook version  [default: dev]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]

notebook#

Create, delete and more operations for user-managed Workbench (Jupyter notebook).

Usage:

 notebook [OPTIONS] COMMAND [ARGS]...

build#

Validates build of notebooks as they are defined in wanna.yaml

Usage:

 notebook build [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -v, --version TEXT  notebook version  [default: dev]

create#

Create a User-Managed Workbench Notebook.

If there already is a notebook with the same name in the same location and project, you will be prompt if you want to delete the existing one and start a new one.

When the notebook instance is created, you will be given a URL link to JupyterLab.

Usage:

 notebook create [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify only one notebook from your wanna-ml
                                  yaml configuration to create. Choose 'all'
                                  to create all notebooks.  [default: all]
  -o, --owner TEXT
  -v, --version TEXT              notebook version  [default: dev]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]

delete#

Delete a User-Managed Workbench Notebook.

Usage:

 notebook delete [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one notebook from your wanna-ml yaml
                      configuration to delete. Choose 'all' to delete all
                      notebooks.  [default: all]

push#

Push docker containers. This command also builds the images.

Usage:

 notebook push [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify only one notebook from your wanna-ml
                                  yaml configuration to push. Choose 'all' to
                                  push all notebooks.  [default: all]
  -v, --version TEXT              notebook version  [default: dev]
  -m, --mode [all|manifests|containers|quick]
                                  Notebook push mode, due to CI/CD not
                                  allowing to push to docker registry from GCP
                                  Agent, we need to split it. Notebooks
                                  currently support only containers, as we do
                                  not create manifests as of now.  [default:
                                  containers]

report#

Displays a link to the cost report per wanna_project and optionally per instance name.

Usage:

 notebook report [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one notebook from your wanna-ml yaml
                      configuration to report. Choose 'all' to report all
                      notebooks.  [default: all]

ssh#

SSH connect to the Compute Engine instance that is behind the Jupyter Notebook.

This will only work if the notebook is already running.

Please note that you can connect to only one instance with one command call. If you have more notebooks defined in your YAML config, you have to select to which you want to connect to, instance_name "all" will be refused.

Usage:

 notebook ssh [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify to which notebook you want to
                                  connect via ssh. Selecting 'all' will work
                                  only if there is just one notebook defined
                                  in your configuration, an error will be
                                  thrown otherwise.  [default: all]
  -b, --background / -i, --interactive
                                  Interactive mode will start a bash directly
                                  in the Compute Engine instance backing the
                                  Jupyter notebook. Background mode serves
                                  more like a port-forwarding, you will be
                                  able to connect to the Jupyter Lab at
                                  localhost:{LOCAL_PORT}  [default:
                                  interactive]
  --port INTEGER                  Jupyter Lab will be accessible at this port
                                  at localhost.  [default: 8080]

sync#

Synchronize existing User-managed Notebooks with wanna.yaml

  1. Reads current notebooks where label is defined per field wanna_project.name in wanna.yaml
  2. Does a diff between what is on GCP and what is on yaml
  3. Create the ones defined in yaml and missing in GCP
  4. Delete the ones in GCP that are not in wanna.yaml

Usage:

 notebook sync [OPTIONS]

Options:

  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  --force                         Synchronisation without prompt
  -v, --version TEXT              notebook version  [default: dev]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]

pipeline#

Plugin for building and deploying Vertex-AI ML Pipelines.

Usage:

 pipeline [OPTIONS] COMMAND [ARGS]...

build#

Create a manifest based on the wanna-ml config that can be later pushed, deployed or run.

Usage:

 pipeline build [OPTIONS]

Options:

  -v, --version TEXT              pipeline version  [default: dev]
  --params PATH                   Path to the params file in yaml format  [env
                                  var: WANNA_ENV_PIPELINE_PARAMS]
  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify only one pipeline from your wanna-ml
                                  yaml configuration to compile. Choose 'all'
                                  to compile all pipelines.  [default: all]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]

deploy#

Deploy the pipeline. Deploying means you can set a schedule and the pipeline will not be run only once, but on regular basis.

Usage:

 pipeline deploy [OPTIONS]

Options:

  -v, --version TEXT  pipeline version  [default: dev]
  -e, --env TEXT      Pipeline env  [default: local]
  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one pipeline from your wanna-ml yaml
                      configuration to deploy. Choose 'all' to deploy all
                      pipelines.  [default: all]

push#

Build and push manifest to Cloud Storage.

Usage:

 pipeline push [OPTIONS]

Options:

  -v, --version TEXT              pipeline version  [default: dev]
  --params PATH                   Path to the params file in yaml format  [env
                                  var: WANNA_ENV_PIPELINE_PARAMS]
  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify only one pipeline from your wanna-ml
                                  yaml configuration to push. Choose 'all' to
                                  push all pipelines.  [default: all]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]

report#

Displays a link to the cost report per wanna_project and optionally per instance name.

Usage:

 pipeline report [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one pipeline from your wanna-ml yaml
                      configuration to report. Choose 'all' to report all
                      pipelines.  [default: all]

run#

Run the pipeline as specified in wanna-ml config. This command puts together build, push and run-manifest steps.

Usage:

 pipeline run [OPTIONS]

Options:

  -v, --version TEXT              pipeline notebook version  [default: dev]
  --params PATH                   Path to the params file in yaml format  [env
                                  var: WANNA_ENV_PIPELINE_PARAMS]
  -s, --sync                      Runs the pipeline in sync mode
  -f, --file PATH                 Path to the wanna-ml yaml configuration
                                  [env var: WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT              Name of the GCP profile you want to use.
                                  Profiles are loaded from wanna-ml yaml
                                  config and (optionally) from this file too.
                                  [env var: WANNA_GCP_PROFILE_NAME; default:
                                  default]
  -n, --name TEXT                 Specify only one pipeline from your wanna-ml
                                  yaml configuration to run. Choose 'all' to
                                  run all pipelines.  [default: all]
  -m, --mode [all|manifests|containers|quick]
                                  Push mode, this is useful if you want to
                                  push containers in one step and deploy
                                  instances in other.Use all for dev
                                  [default: all]
  --skip-execution-cache          Kubeflow pipeline cache configuration

run-manifest#

Run the pipeline as specified in the wanna-ml manifest.

Usage:

 pipeline run-manifest [OPTIONS]

Options:

  -v, --manifest TEXT  Job deployment manifest
  --params PATH        Path to the params file in yaml format  [env var:
                       WANNA_ENV_PIPELINE_PARAMS]
  -s, --sync           Runs the pipeline in sync mode

tensorboard#

Create, delete or list Tensorboard instances.

Usage:

 tensorboard [OPTIONS] COMMAND [ARGS]...

create#

Create Tensorboard Instance in GCP Vertex AI Experiments.

If there already is a tensorboard with the same name in the same location and project, you will be prompt if you want to delete the existing one and start a new one.

When the tensorboard instance is created, you will be given a full resource name.

Usage:

 tensorboard create [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one tensorboard from your wanna-ml yaml
                      configuration to create. Choose 'all' to create all
                      tensorboards.  [default: all]

delete#

Delete Tensorboard Instance in GCP Vertex AI Experiments.

Usage:

 tensorboard delete [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  -n, --name TEXT     Specify only one tensorboard from your wanna-ml yaml
                      configuration to delete. Choose 'all' to delete all
                      tensorboards.  [default: all]

list#

List Tensorboard Instances in GCP Vertex AI Experiments.

We also show Tensorboard Experiments and Tensorboard Runs for each Instance in the tree format.

Usage:

 tensorboard list [OPTIONS]

Options:

  -f, --file PATH     Path to the wanna-ml yaml configuration  [env var:
                      WANNA_FILE; default: wanna.yaml]
  -p, --profile TEXT  Name of the GCP profile you want to use. Profiles are
                      loaded from wanna-ml yaml config and (optionally) from
                      this file too.  [env var: WANNA_GCP_PROFILE_NAME;
                      default: default]
  --region TEXT       Overwrites the region from wanna-ml yaml configuration
  --filter TEXT       GCP filter expression for tensorboard instances. Read
                      more on GCP filters on https://cloud.google.com/sdk/gclo
                      ud/reference/topic/filters  Example: display_name=my-
                      tensorboard.  Example: labels.wanna_project:* - to show
                      all tensorboard created by wanna-ml. Example:
                      labels.wanna_project:sushi-ssl.
  --url / --no-url    Weather to show URL link to experiments  [default: url]

version#

Print your current and latest available version

Usage:

 version [OPTIONS]