fix: pip not available in runtime (#3306)

* try to fix pip unavailable

* update test case for pip

* force rebuild in CI

* remove extra symlink

* fix newline

* added semi-colon to line 31

* Dockerfile.j2: activate env at the end

* Revert "Dockerfile.j2: activate env at the end"

This reverts commit cf2f565102.

* cleanup Dockerfile

* switch default python image

* remove image agnostic (no longer used)

* fix tests

* switch to nikolaik/python-nodejs:python3.11-nodejs22

* fix test

* fix test

* revert docker

* update template

---------

Co-authored-by: tobitege <tobitege@gmx.de>
Co-authored-by: Graham Neubig <neubig@gmail.com>

.github/workflows/ghcr.yml

@@ -81,7 +81,7 @@ jobs:
     strategy:
       matrix:
         image: ['od_runtime']
-        base_image: ['ubuntu:22.04']
+        base_image: ['nikolaik/python-nodejs:python3.11-nodejs22']
         platform: ['amd64', 'arm64']
     steps:
       - name: Checkout
@@ -115,7 +115,7 @@ jobs:
       - name: Install Python dependencies using Poetry
         run: make install-python-dependencies
       - name: Create source distribution and Dockerfile
-        run: poetry run python3 opendevin/runtime/utils/runtime_build.py --base_image ${{ matrix.base_image }} --build_folder containers/runtime
+        run: poetry run python3 opendevin/runtime/utils/runtime_build.py --base_image ${{ matrix.base_image }} --build_folder containers/runtime --force_rebuild
       - name: Build and export image
         id: build
         run: |

config.template.toml

@@ -174,7 +174,7 @@ llm_config = 'gpt3'
 #user_id = 1000
 
 # Container image to use for the sandbox
-#container_image = "ghcr.io/opendevin/sandbox:main"
+#container_image = "nikolaik/python-nodejs:python3.11-nodejs22"
 
 # Use host network
 #use_host_network = false

evaluation/EDA/run_infer.py

@@ -62,7 +62,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=False,
             use_host_network=False,
         ),

evaluation/agent_bench/run_infer.py

@@ -44,7 +44,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
         ),

evaluation/bird/run_infer.py

@@ -75,7 +75,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
         ),

evaluation/browsing_delegation/run_infer.py

@@ -40,7 +40,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=False,
             use_host_network=False,
         ),

evaluation/gaia/run_infer.py

@@ -51,7 +51,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
         ),

evaluation/gorilla/run_infer.py

@@ -42,7 +42,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
         ),

evaluation/gpqa/run_infer.py

@@ -65,7 +65,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
         ),

evaluation/humanevalfix/run_infer.py

@@ -86,7 +86,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
         ),

evaluation/toolqa/run_infer.py

@@ -45,7 +45,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
         ),

evaluation/webarena/run_infer.py

@@ -54,7 +54,7 @@ def get_config(
         runtime='eventstream',
         max_iterations=metadata.max_iterations,
         sandbox=SandboxConfig(
-            container_image='ubuntu:22.04',
+            container_image='python:3.11-bookworm',
             enable_auto_lint=True,
             use_host_network=False,
             browsergym_eval_env=env_id,

opendevin/core/config.py

@@ -166,9 +166,7 @@ class SandboxConfig(metaclass=Singleton):
     """
 
     api_hostname: str = 'localhost'
-    container_image: str = (
-        'ubuntu:22.04'  # default to ubuntu:22.04 for eventstream runtime
-    )
+    container_image: str = 'nikolaik/python-nodejs:python3.11-nodejs22'  # default to nikolaik/python-nodejs:python3.11-nodejs22 for eventstream runtime
     user_id: int = os.getuid() if hasattr(os, 'getuid') else 1000
     timeout: int = 120
     enable_auto_lint: bool = (

opendevin/runtime/utils/image_agnostic.py

@@ -1,113 +0,0 @@
-"""This module contains functions for building and managing the agnostic sandbox image.
-
-This WILL BE DEPRECATED when EventStreamRuntime is fully implemented and adopted.
-"""
-
-import tempfile
-
-import docker
-
-from opendevin.core.logger import opendevin_logger as logger
-
-
-def generate_dockerfile(base_image: str) -> str:
-    """Generate the Dockerfile content for the agnostic sandbox image based on user-provided base image.
-
-    NOTE: This is only tested on debian yet.
-    """
-    # FIXME: Remove the requirement of ssh in future version
-    dockerfile_content = (
-        f'FROM {base_image}\n'
-        'RUN apt update && apt install -y openssh-server wget sudo\n'
-        'RUN mkdir -p -m0755 /var/run/sshd\n'
-        'RUN mkdir -p /opendevin && mkdir -p /opendevin/logs && chmod 777 /opendevin/logs\n'
-        'RUN echo "" > /opendevin/bash.bashrc\n'
-        'RUN if [ ! -d /opendevin/miniforge3 ]; then \\\n'
-        '        wget --progress=bar:force -O Miniforge3.sh "https://github.com/conda-forge/miniforge/releases/latest/download/Miniforge3-$(uname)-$(uname -m).sh" && \\\n'
-        '        bash Miniforge3.sh -b -p /opendevin/miniforge3 && \\\n'
-        '        rm Miniforge3.sh && \\\n'
-        '        chmod -R g+w /opendevin/miniforge3 && \\\n'
-        '        bash -c ". /opendevin/miniforge3/etc/profile.d/conda.sh && conda config --set changeps1 False && conda config --append channels conda-forge"; \\\n'
-        '    fi\n'
-        'RUN /opendevin/miniforge3/bin/pip install --upgrade pip\n'
-        'RUN /opendevin/miniforge3/bin/pip install jupyterlab notebook jupyter_kernel_gateway flake8\n'
-        'RUN /opendevin/miniforge3/bin/pip install python-docx PyPDF2 python-pptx pylatexenc openai\n'
-    ).strip()
-    return dockerfile_content
-
-
-def _build_sandbox_image(
-    base_image: str, target_image_name: str, docker_client: docker.DockerClient
-):
-    try:
-        with tempfile.TemporaryDirectory() as temp_dir:
-            dockerfile_content = generate_dockerfile(base_image)
-
-            logger.info(f'Building agnostic sandbox image: {target_image_name}')
-            logger.info(
-                (
-                    f'===== Dockerfile content =====\n'
-                    f'{dockerfile_content}\n'
-                    f'==============================='
-                )
-            )
-            with open(f'{temp_dir}/Dockerfile', 'w') as file:
-                file.write(dockerfile_content)
-
-            api_client = docker_client.api
-            build_logs = api_client.build(
-                path=temp_dir, tag=target_image_name, rm=True, decode=True
-            )
-
-            for log in build_logs:
-                if 'stream' in log:
-                    print(log['stream'].strip())
-                elif 'error' in log:
-                    logger.error(log['error'].strip())
-                else:
-                    logger.info(str(log))
-
-        logger.info(f'Image {target_image_name} built successfully')
-    except docker.errors.BuildError as e:
-        logger.error(f'Sandbox image build failed: {e}')
-        raise e
-    except Exception as e:
-        logger.error(f'An error occurred during sandbox image build: {e}')
-        raise e
-
-
-def _get_new_image_name(base_image: str) -> str:
-    prefix = 'od_sandbox'
-    if ':' not in base_image:
-        base_image = base_image + ':latest'
-
-    [repo, tag] = base_image.split(':')
-    repo = repo.replace('/', '___')
-    return f'{prefix}:{repo}__{tag}'
-
-
-def get_od_sandbox_image(base_image: str, docker_client: docker.DockerClient) -> str:
-    """Return the sandbox image name based on user-provided base image.
-
-    The returned sandbox image is assumed to contains all the required dependencies for OpenDevin.
-    If the sandbox image is not found, it will be built.
-    """
-    # OpenDevin's offcial sandbox already contains the required dependencies for OpenDevin.
-    if 'ghcr.io/opendevin/sandbox' in base_image:
-        return base_image
-
-    new_image_name = _get_new_image_name(base_image)
-
-    # Detect if the sandbox image is built
-    images = docker_client.images.list()
-    for image in images:
-        if new_image_name in image.tags:
-            logger.info('Found existing od_sandbox image, reuse:' + new_image_name)
-            return new_image_name
-
-    # If the sandbox image is not found, build it
-    logger.info(
-        f'od_sandbox image is not found for {base_image}, will build: {new_image_name}'
-    )
-    _build_sandbox_image(base_image, new_image_name, docker_client)
-    return new_image_name

opendevin/runtime/utils/runtime_build.py

@@ -239,6 +239,7 @@ def build_runtime_image(
     extra_deps: str | None = None,
     docker_build_folder: str | None = None,
     dry_run: bool = False,
+    force_rebuild: bool = False,
 ) -> str:
     """Build the runtime image for the OpenDevin runtime.
 
@@ -277,7 +278,10 @@ def build_runtime_image(
     # 2. If the exact hash is not found, we will FIRST try to re-build it
     # by leveraging the non-hash `generic_runtime_image_name` to save some time
     # from re-building the dependencies (e.g., poetry install, apt install)
-    elif _check_image_exists(generic_runtime_image_name, docker_client):
+    elif (
+        _check_image_exists(generic_runtime_image_name, docker_client)
+        and not force_rebuild
+    ):
         logger.info(
             f'Cannot find matched hash for image [{hash_runtime_image_name}]\n'
             f'Will try to re-build it from latest [{generic_runtime_image_name}] image to potentially save '
@@ -319,6 +323,10 @@ def build_runtime_image(
     # 3. If the image is not found AND we cannot re-use the non-hash latest relavant image,
     # we will build it completely from scratch
     else:
+        if force_rebuild:
+            logger.info(
+                f'Force re-build: Will try to re-build image [{generic_runtime_image_name}] from scratch.\n'
+            )
         cur_docker_build_folder = docker_build_folder or tempfile.mkdtemp()
         _new_from_scratch_hash = prep_docker_build_folder(
             cur_docker_build_folder,
@@ -352,8 +360,11 @@ def build_runtime_image(
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
-    parser.add_argument('--base_image', type=str, default='ubuntu:22.04')
+    parser.add_argument(
+        '--base_image', type=str, default='nikolaik/python-nodejs:python3.11-nodejs22'
+    )
     parser.add_argument('--build_folder', type=str, default=None)
+    parser.add_argument('--force_rebuild', action='store_true', default=False)
     args = parser.parse_args()
 
     if args.build_folder is not None:
@@ -373,6 +384,7 @@ if __name__ == '__main__':
                 docker_client=docker.from_env(),
                 docker_build_folder=temp_dir,
                 dry_run=True,
+                force_rebuild=args.force_rebuild,
             )
             _runtime_image_repo, runtime_image_hash_tag = runtime_image_hash_name.split(
                 ':'

opendevin/runtime/utils/runtime_templates/Dockerfile.j2

@@ -18,13 +18,6 @@ RUN apt-get update && \
     apt-get clean && \
     rm -rf /var/lib/apt/lists/*
 
-# Install Python if not already installed
-RUN if [ ! -e /usr/bin/python ]; then \
-    apt-get update && \
-    apt-get install -y python3 && \
-    ln -s /usr/bin/python3 /usr/bin/python; \
-fi
-
 # Create necessary directories
 RUN mkdir -p /opendevin && \
     mkdir -p /opendevin/logs && \

tests/integration/regenerate.sh

@@ -59,7 +59,7 @@ mkdir -p $WORKSPACE_BASE
 TEST_RUNTIME="${TEST_RUNTIME:-eventstream}"  # can be server or eventstream
 # TODO: set this as default after ServerRuntime is deprecated
 if [ "$TEST_RUNTIME" == "eventstream" ] && [ -z "$SANDBOX_CONTAINER_IMAGE" ]; then
-  SANDBOX_CONTAINER_IMAGE="ubuntu:22.04"
+  SANDBOX_CONTAINER_IMAGE="nikolaik/python-nodejs:python3.11-nodejs22"
 fi
 
 MAX_ITERATIONS=15

tests/unit/test_config.py

@@ -358,7 +358,7 @@ def test_defaults_dict_after_updates(default_config):
     assert defaults_after_updates['sandbox']['timeout']['default'] == 120
     assert (
         defaults_after_updates['sandbox']['container_image']['default']
-        == 'ubuntu:22.04'
+        == 'nikolaik/python-nodejs:python3.11-nodejs22'
     )
     assert defaults_after_updates == initial_defaults
 

tests/unit/test_image_agnostic_util.py

@@ -1,51 +0,0 @@
-from unittest.mock import MagicMock, patch
-
-from opendevin.runtime.utils.image_agnostic import (
-    _get_new_image_name,
-    generate_dockerfile,
-    get_od_sandbox_image,
-)
-
-
-def test_generate_dockerfile():
-    base_image = 'debian:11'
-    dockerfile_content = generate_dockerfile(base_image)
-    assert base_image in dockerfile_content
-    assert (
-        'RUN apt update && apt install -y openssh-server wget sudo'
-        in dockerfile_content
-    )
-
-
-def test_get_new_image_name_legacy():
-    # test non-eventstream runtime (sandbox-based)
-    base_image = 'debian:11'
-    new_image_name = _get_new_image_name(base_image)
-    assert new_image_name == 'od_sandbox:debian__11'
-
-    base_image = 'ubuntu:22.04'
-    new_image_name = _get_new_image_name(base_image)
-    assert new_image_name == 'od_sandbox:ubuntu__22.04'
-
-    base_image = 'ubuntu'
-    new_image_name = _get_new_image_name(base_image)
-    assert new_image_name == 'od_sandbox:ubuntu__latest'
-
-
-@patch('opendevin.runtime.utils.image_agnostic._build_sandbox_image')
-@patch('opendevin.runtime.utils.image_agnostic.docker.DockerClient')
-def test_get_od_sandbox_image(mock_docker_client, mock_build_sandbox_image):
-    base_image = 'debian:11'
-    mock_docker_client.images.list.return_value = [
-        MagicMock(tags=['od_sandbox:debian__11'])
-    ]
-
-    image_name = get_od_sandbox_image(base_image, mock_docker_client)
-    assert image_name == 'od_sandbox:debian__11'
-
-    mock_docker_client.images.list.return_value = []
-    image_name = get_od_sandbox_image(base_image, mock_docker_client)
-    assert image_name == 'od_sandbox:debian__11'
-    mock_build_sandbox_image.assert_called_once_with(
-        base_image, 'od_sandbox:debian__11', mock_docker_client
-    )

tests/unit/test_runtime.py

@@ -83,7 +83,9 @@ def enable_auto_lint(request):
     return request.param
 
 
-@pytest.fixture(scope='module', params=['ubuntu:22.04', 'debian:11'])
+@pytest.fixture(
+    scope='module', params=['nikolaik/python-nodejs:python3.11-nodejs22', 'debian:11']
+)
 def container_image(request):
     time.sleep(1)
     return request.param
@@ -127,7 +129,7 @@ async def _load_runtime(
         if 'od_runtime' not in cur_container_image and cur_container_image not in {
             'xingyaoww/od-eval-miniwob:v1.0'
         }:  # a special exception list
-            cur_container_image = 'ubuntu:22.04'
+            cur_container_image = 'nikolaik/python-nodejs:python3.11-nodejs22'
             logger.warning(
                 f'`{config.sandbox.container_image}` is not an od_runtime image. Will use `{cur_container_image}` as the container image for testing.'
             )
@@ -1033,6 +1035,13 @@ async def test_bash_python_version(temp_dir, box_class):
     assert obs.exit_code == 0
     # Should not error out
 
+    action = CmdRunAction(command='pip --version')
+    logger.info(action, extra={'msg_type': 'ACTION'})
+    obs = await runtime.run_action(action)
+    logger.info(obs, extra={'msg_type': 'OBSERVATION'})
+    assert obs.exit_code == 0
+    # Should not error out
+
     await runtime.close()
     await asyncio.sleep(1)
 

tests/unit/test_runtime_build.py

@@ -62,7 +62,7 @@ def test_put_source_code_to_dir(temp_dir):
 def test_docker_build_folder(temp_dir):
     prep_docker_build_folder(
         temp_dir,
-        base_image='ubuntu:22.04',
+        base_image='nikolaik/python-nodejs:python3.11-nodejs22',
         skip_init=False,
     )
 
@@ -81,14 +81,14 @@ def test_docker_build_folder(temp_dir):
 def test_hash_folder_same(temp_dir):
     dir_hash_1 = prep_docker_build_folder(
         temp_dir,
-        base_image='ubuntu:22.04',
+        base_image='nikolaik/python-nodejs:python3.11-nodejs22',
         skip_init=False,
     )
 
     with tempfile.TemporaryDirectory() as temp_dir_2:
         dir_hash_2 = prep_docker_build_folder(
             temp_dir_2,
-            base_image='ubuntu:22.04',
+            base_image='nikolaik/python-nodejs:python3.11-nodejs22',
             skip_init=False,
         )
     assert dir_hash_1 == dir_hash_2
@@ -97,14 +97,14 @@ def test_hash_folder_same(temp_dir):
 def test_hash_folder_diff_init(temp_dir):
     dir_hash_1 = prep_docker_build_folder(
         temp_dir,
-        base_image='ubuntu:22.04',
+        base_image='nikolaik/python-nodejs:python3.11-nodejs22',
         skip_init=False,
     )
 
     with tempfile.TemporaryDirectory() as temp_dir_2:
         dir_hash_2 = prep_docker_build_folder(
             temp_dir_2,
-            base_image='ubuntu:22.04',
+            base_image='nikolaik/python-nodejs:python3.11-nodejs22',
             skip_init=True,
         )
     assert dir_hash_1 != dir_hash_2
@@ -113,7 +113,7 @@ def test_hash_folder_diff_init(temp_dir):
 def test_hash_folder_diff_image(temp_dir):
     dir_hash_1 = prep_docker_build_folder(
         temp_dir,
-        base_image='ubuntu:22.04',
+        base_image='nikolaik/python-nodejs:python3.11-nodejs22',
         skip_init=False,
     )
 
@@ -178,11 +178,12 @@ def test_get_runtime_image_repo_and_tag_eventstream():
         and img_tag == f'{OD_VERSION}_image_debian_tag_11'
     )
 
-    base_image = 'ubuntu:22.04'
+    base_image = 'nikolaik/python-nodejs:python3.11-nodejs22'
     img_repo, img_tag = get_runtime_image_repo_and_tag(base_image)
     assert (
         img_repo == f'{RUNTIME_IMAGE_REPO}'
-        and img_tag == f'{OD_VERSION}_image_ubuntu_tag_22.04'
+        and img_tag
+        == f'{OD_VERSION}_image_nikolaik___python-nodejs_tag_python3.11-nodejs22'
     )
 
     base_image = 'ubuntu'