Feature/#30 build the data science sandbox as docker image

.github/workflows/check_ci.yaml

@@ -18,8 +18,11 @@ jobs:
        uses: ./.github/actions/prepare_poetry_env
 
      - name: Run pytest
-       run: poetry run pytest test/ci/test_install_dependencies.py
+       run: >
+         poetry run pytest
+         test/unit
+         test/integration/test_create_dss_docker_image.py
        env: # Set the secret as an env variable
-          AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
-          AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_ACCESS_KEY_SECRET }}
-          AWS_DEFAULT_REGION: ${{ secrets.AWS_REGION }}
+         AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
+         AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_ACCESS_KEY_SECRET }}
+         AWS_DEFAULT_REGION: ${{ secrets.AWS_REGION }}

doc/changes/changes_0.1.0.md

@@ -14,6 +14,7 @@ Version: 0.1.0
 
  - #11: Created a notebook to show training with scikit-learn in the notebook
  - #15: Installed exasol-notebook-connector via ansible
+ - #30: Added script to build the Data Science Sandbox as Docker Image
 
 ## Bug Fixes
 

doc/developer_guide/developer_guide.md

@@ -39,11 +39,11 @@ A CLI command has normally a respective function in the `lib` submodule. Hence,
 
 There are generally three types of commands:
 
-| Type | Explanation |
-| ----- | --------- |
-| Release Commands | used during the release |
-| Deployment Commands | used to deploy infrastructure onto AWS cloud |
-| Development Commands | used to identify problems or for testing |
+| Type                 | Explanation                                  |
+|----------------------|----------------------------------------------|
+| Release Commands     | used during the release                      |
+| Deployment Commands  | used to deploy infrastructure onto AWS cloud |
+| Development Commands | used to identify problems or for testing     |
 
 ### Release commands
 
@@ -71,6 +71,7 @@ The following commands can be used to deploy the infrastructure onto a given AWS
 - `setup-vm-bucket` - deploys the AWS Bucket cloudformation stack which will be used to deploy the VM images
 - `setup-release-codebuild` - deploys the AWS Codebuild cloudformation stack which will be used for the release-build
 - `setup-vm-bucket-waf` - deploys the AWS Codebuild cloudformation stack which contains the WAF Acl configuration for the Cloudfront distribution of the VM Bucket
+- `create-docker-image` - creates a Docker image for data-science-sandbox and deploys it to hub.docker.com/exasol/data-science-sandbox
 
 ## Flow
 

exasol/ds/sandbox/cli/commands/__init__.py

@@ -13,3 +13,4 @@
 from .update_release import update_release
 from .make_ami_public import make_ami_public
 from .setup_vm_bucket_waf import setup_vm_bucket_waf
+from .create_docker_image import create_docker_image

exasol/ds/sandbox/cli/commands/create_docker_image.py

@@ -0,0 +1,44 @@
+import click
+
+from exasol.ds.sandbox.cli.cli import cli
+from exasol.ds.sandbox.cli.options.logging import logging_options
+from exasol.ds.sandbox.cli.common import add_options
+from exasol.ds.sandbox.lib.dss_docker import DssDockerImage
+from exasol.ds.sandbox.lib.logging import SUPPORTED_LOG_LEVELS
+from exasol.ds.sandbox.lib.logging import set_log_level
+
+
+@cli.command()
+@add_options([
+    click.option(
+        '--repository', type=str, metavar="ORG/REPO", show_default=True,
+        default="exasol/data-science-sandbox",
+        help="Organization and repository on hub.docker.com to publish the docker image to"),
+    click.option('--version', type=str, help="Docker image version tag"),
+    click.option(
+        '--publish', type=bool, is_flag=True,
+        help="Whether to publish the created Docker image"),
+    click.option(
+        '--keep-container', type=bool, is_flag=True,
+        help="""Keep the Docker Container running after creating the image.
+        Otherwise stop and remove the container."""),
+])
+@add_options(logging_options)
+def create_docker_image(
+        repository: str,
+        version: str,
+        publish: bool,
+        keep_container: bool,
+        log_level: str,
+):
+    """
+    Create a Docker image for data-science-sandbox and deploy
+    it to a Docker repository.
+    """
+    set_log_level(log_level)
+    DssDockerImage(
+        repository=repository,
+        version=version,
+        publish=publish,
+        keep_container=keep_container,
+    ).create()

exasol/ds/sandbox/lib/ansible/ansible_access.py

@@ -1,24 +1,27 @@
-from typing import Callable
-
 import ansible_runner
+import logging
+
+from typing import Callable
 
 from exasol.ds.sandbox.lib.ansible.ansible_run_context import AnsibleRunContext
+from exasol.ds.sandbox.lib.logging import get_status_logger, LogType
 
 
 class AnsibleException(RuntimeError):
     pass
 
 
 class AnsibleAccess:
-
     """
     Provides access to ansible runner.
     @raises: AnsibleException if ansible execution fails
     """
     @staticmethod
     def run(private_data_dir: str, run_ctx: AnsibleRunContext, printer: Callable[[str], None]):
+        quiet = not get_status_logger(LogType.ANSIBLE).isEnabledFor(logging.INFO)
         r = ansible_runner.run(private_data_dir=private_data_dir,
                                playbook=run_ctx.playbook,
+                               quiet=quiet,
                                extravars=run_ctx.extra_vars)
         for e in r.events:
             printer(e)

exasol/ds/sandbox/lib/ansible/ansible_runner.py

@@ -20,7 +20,7 @@ def __init__(self, ansible_access: AnsibleAccess, work_dir: Path):
 
     @staticmethod
     def printer(msg: str):
-        LOG.info(msg)
+        LOG.debug(msg)
 
     def run(self, ansible_run_context: AnsibleRunContext, host_infos: Tuple[HostInfo]):
         inventory_content = render_template("inventory.jinja", host_infos=host_infos)

exasol/ds/sandbox/lib/dss_docker/Dockerfile

@@ -0,0 +1,4 @@
+FROM ubuntu:20.04
+ENV DEBIAN_FRONTEND noninteractive
+RUN apt-get update && apt-get install --no-install-recommends --assume-yes python3 python3-pexpect
+EXPOSE 8888/tcp

exasol/ds/sandbox/lib/dss_docker/__init__.py

@@ -0,0 +1 @@
+from .create_image import DssDockerImage

exasol/ds/sandbox/lib/dss_docker/create_image.py

@@ -0,0 +1,102 @@
+import docker
+import humanfriendly
+import importlib_resources
+
+from datetime import datetime
+from docker.types import Mount
+from exasol.ds.sandbox.lib import pretty_print
+from importlib_metadata import version
+from pathlib import Path
+
+from exasol.ds.sandbox.lib.config import ConfigObject, SLC_VERSION
+from exasol.ds.sandbox.lib.logging import get_status_logger, LogType
+from exasol.ds.sandbox.lib.ansible import ansible_repository
+from exasol.ds.sandbox.lib.ansible.ansible_run_context import AnsibleRunContext
+from exasol.ds.sandbox.lib.ansible.ansible_access import AnsibleAccess
+from exasol.ds.sandbox.lib.setup_ec2.run_install_dependencies import run_install_dependencies
+
+
+DSS_VERSION = version("exasol-data-science-sandbox")
+
+
+class DssDockerImage:
+    @classmethod
+    def timestamp(cls) -> str:
+        return f'{datetime.now().timestamp():.0f}'
+
+    def __init__(
+            self,
+            repository: str,
+            version: str = None,
+            publish: bool = False,
+            keep_container: bool = False,
+    ):
+        version = version if version else DSS_VERSION
+        self.container_name = f"ds-sandbox-{DssDockerImage.timestamp()}"
+        self.image_name = f"{repository}:{version}"
+        self.publish = publish
+        self.keep_container = keep_container
+
+    def _ansible_run_context(self) -> AnsibleRunContext:
+        extra_vars = {
+            "docker_container": self.container_name,
+        }
+        return AnsibleRunContext(
+            playbook="dss_docker_playbook.yml",
+            extra_vars=extra_vars,
+        )
+
+    def _ansible_config(self) -> ConfigObject:
+        return ConfigObject(
+            time_to_wait_for_polling=0.1,
+            slc_version=SLC_VERSION,
+        )
+
+    def _docker_file(self) -> Path:
+        return (
+            importlib_resources
+            .files("exasol.ds.sandbox.lib.dss_docker")
+            .joinpath("Dockerfile")
+        )
+
+    def create(self):
+        logger = get_status_logger(LogType.DOCKER_IMAGE)
+        docker_file = self._docker_file()
+        try:
+            start = datetime.now()
+            docker_client = docker.from_env()
+            logger.info(f"Creating docker image {self.image_name} from {docker_file}")
+            docker_client.images.build(path=str(docker_file.parent), tag=self.image_name)
+            container = docker_client.containers.create(
+                image=self.image_name,
+                name=self.container_name,
+                command="sleep infinity",
+                detach=True,
+            )
+            logger.info("Starting container")
+            container.start()
+            logger.info("Installing dependencies")
+            run_install_dependencies(
+                AnsibleAccess(),
+                configuration=self._ansible_config(),
+                host_infos=tuple(),
+                ansible_run_context=self._ansible_run_context(),
+                ansible_repositories=ansible_repository.default_repositories,
+            )
+            logger.info("Committing changes to docker container")
+            image = container.commit(
+                repository=self.image_name,
+            )
+        except Exception as ex:
+            raise ex
+        finally:
+            if self.keep_container:
+                logger.info("Keeping container running")
+            else:
+                logger.info("Stopping container")
+                container.stop()
+                logger.info("Removing container")
+                container.remove()
+        size = humanfriendly.format_size(image.attrs["Size"])
+        elapsed = pretty_print.elapsed(start)
+        logger.info(f"Built Docker image {self.image_name} size {size} in {elapsed}.")

exasol/ds/sandbox/lib/logging.py

@@ -13,6 +13,7 @@ class LogType(Enum):
     SETUP_CI_CODEBUILD = "setup_ci_codebuild"
     AWS_ACCESS = "aws_access"
     ANSIBLE = "ansible"
+    DOCKER_IMAGE = "docker_image"
     CREATE_VM = "create_vm"
     SETUP_RELEASE_BUILD = "setup_release_build"
     RELEASE_BUILD = "release_build"

exasol/ds/sandbox/lib/pretty_print.py

@@ -0,0 +1,8 @@
+from datetime import datetime, timedelta
+
+
+def elapsed(start: datetime, round_to_seconds=True) -> str:
+    d = datetime.now() - start
+    if round_to_seconds:
+        d = d - timedelta(microseconds=d.microseconds)
+    return str(d)

exasol/ds/sandbox/runtime/ansible/dss_docker_playbook.yml

@@ -0,0 +1,23 @@
+- name: Prepare environment
+  hosts: localhost
+  gather_facts: false
+  vars:
+      ansible_python_interpreter: python3
+  tasks:
+    - name: Add docker container to inventory
+      add_host:
+        name: "{{docker_container}}"
+        groups: docker_container_group
+        ansible_connection: docker
+
+- name: Setup DSS Docker Container
+  hosts: docker_container_group
+  gather_facts: false
+  vars:
+      ansible_python_interpreter: python3
+      user_name: root
+      user_home: /root
+      need_sudo: false
+      docker_integration_test: true
+  tasks:
+    - import_tasks: general_setup_tasks.yml

exasol/ds/sandbox/runtime/ansible/ec2_setup_tasks.yml

@@ -0,0 +1,6 @@
+- name: Install Script_languages
+  include_role:
+    name: script_languages
+- name: Update netplan
+  include_role:
+    name: netplan

exasol/ds/sandbox/runtime/ansible/slc_setup_tasks.yml → exasol/ds/sandbox/runtime/ansible/general_setup_tasks.yml

@@ -5,17 +5,14 @@
   become: "{{need_sudo}}"
 - name: Install Poetry
   include_role:
-      name: poetry
+    name: poetry
 - name: Install Jupyter
   include_role:
-        name: jupyter
+    name: jupyter
+- name: Clear pip cache
+  ansible.builtin.file:
+    path: /root/.cache/pip
+    state: absent
 - name: Install Docker
   include_role:
-        name: docker
-- name: Install Script_languages
-  include_role:
-      name: script_languages
-- name: Update netplan
-  include_role:
-      name: netplan
-
+    name: docker

exasol/ds/sandbox/runtime/ansible/roles/jupyter/files/requirements_dependencies.txt

@@ -1,4 +1,4 @@
 uncertainties==3.1.7
 numpy==1.23.1
 pandas==1.4.3
-exasol-notebook-connector==0.1.0
+exasol-notebook-connector==0.2.0

exasol/ds/sandbox/runtime/ansible/roles/jupyter/tasks/main.yml

@@ -3,7 +3,7 @@
 
 - name: Setup Jupyter
   block:
-    - name: Install dependant apt packages
+    - name: Install dependent apt packages
       apt:
         name: "{{apt_dependencies}}"
         state: present

exasol/ds/sandbox/runtime/ansible/roles/poetry/tasks/main.yml

@@ -1,6 +1,6 @@
 ---
 
-- name: Install dependant apt packages
+- name: Install dependent apt packages
   apt:
     name: "{{apt_dependencies}}"
     state: present

exasol/ds/sandbox/runtime/ansible/roles/script_languages/tasks/main.yml

@@ -1,6 +1,6 @@
 ---
 
-- name: Install dependant apt packages
+- name: Install dependent apt packages
   apt:
     name: "{{apt_dependencies}}"
     state: present

exasol/ds/sandbox/runtime/ansible/slc_setup.yml

@@ -7,4 +7,5 @@
       need_sudo: yes
   remote_user: ubuntu
   tasks:
-  - import_tasks: slc_setup_tasks.yml
+  - import_tasks: general_setup_tasks.yml
+  - import_tasks: ec2_setup_tasks.yml