Merge pull request #123 from dbpunk-labs/121-feature-podman-support

feat: podman support

README.md

@@ -28,7 +28,7 @@
 Requirement
 * python 3.10 and above
 * pip
-* [docker](https://www.docker.com/products/docker-desktop/) 24.0.0 and above, docker desktop is recommended
+* [docker](https://www.docker.com/products/docker-desktop/) 24.0.0 and above, or [podman](https://podman.io/)
 
 > To deploy Octogen, the user needs permission to run Docker commands.   
 > To use codellama, your host must have at least 8 CPUs and 16 GB of RAM.
@@ -52,6 +52,8 @@ og_up
 > If you opt for CodeLlama, Octogen will automatically download it from huggingface.co.
 > In case the installation of the Octogen Terminal CLI is taking longer than expected, you might want to consider switching to a different pip mirror.
 
+The default is using docker as container engine. use podman with flag `--use_podman` 
+
 3. Open your terminal and execute the command `og`, you will see the following output
 
 ```

chat/src/og_terminal/terminal_chat.py

@@ -213,10 +213,10 @@ def handle_action_end(segments, respond, images, values):
     has_error = "✅" if not respond.on_agent_action_end.has_error else "❌"
     old_value = values.pop()
     segment = segments.pop()
-    if not images:
+    if not images and not has_error:
         images.extend(respond.on_agent_action_end.output_files)
+    values.append(old_value)
     segments.append((len(values) - 1, has_error, segment[2]))
-
     # add the next steps loading
     spinner = Spinner("dots", style="status.spinner", speed=1.0, text="")
     values.append(("text", "", []))

docker/start_all.sh

@@ -38,6 +38,7 @@ sleep 3
 AGENT_RPC_KEY=$(cat ${ROOT_DIR}/agent/.env | grep admin_key | tr -d '\r' | cut -d "=" -f 2)
 KERNEL_RPC_KEY=$(cat ${ROOT_DIR}/kernel/.env | grep rpc_key | tr -d '\r' | cut -d "=" -f 2)
 og_agent_setup --kernel_endpoint=127.0.0.1:9527 --kernel_api_key=${KERNEL_RPC_KEY} --agent_endpoint=127.0.0.1:9528 --admin_key=${AGENT_RPC_KEY}
+og_agent_setup --kernel_endpoint=127.0.0.1:9527 --kernel_api_key=${KERNEL_RPC_KEY} --agent_endpoint=127.0.0.1:9528 --admin_key=${AGENT_RPC_KEY}
 
 while true
 do

docs/source/getstarted.rst

@@ -11,7 +11,7 @@ Octopus requires the following enviroment
 
 - Python 3.10.0 and above.
 - `Pip <https://pip.pypa.io/en/stable/installation/>`_
-- `Docker Desktop 24.0.0 and above <https://www.docker.com/products/docker-desktop/>`_  
+- `Docker Desktop 24.0.0 and above <https://www.docker.com/products/docker-desktop/>`_  or `Podman <https://podman.io/docs/installation>`_
 
 To use codellama, your host must have at least 8 CPUs and 16 GB of RAM
 
@@ -28,10 +28,12 @@ the second step, use ``og_up`` to setup the octopus service and cli::
 
 You have the option to select from 
 
-- OpenAI, [apply](https://platform.openai.com/account/api-keys) the openai api key
-- Azure OpenAI, [apply](https://azure.microsoft.com/en-us/products/ai-services/openai-service) the azure api key
+- OpenAI, `apply <https://platform.openai.com/account/api-keys>`_ the openai api key
+- Azure OpenAI, `apply <https://azure.microsoft.com/en-us/products/ai-services/openai-service>`_ the azure api key
 - CodeLlama
-- Octogen(beta) agent services, [apply](https://www.octogen.dev/) the octogen agent service key
+- Octogen(beta) agent services, `apply <https://www.octogen.dev/>`_ the octogen agent service key
+
+the default is using docker as container engine, use podman with flag ``--use_podman``
 
 If you opt for CodeLlama, Octogen will automatically download it from huggingface.co. 
 In case the installation of the Octogen Terminal CLI is taking longer than expected, 

up/src/og_up/up.py

@@ -117,7 +117,28 @@ def refresh(
     live.refresh()
 
 
-def check_the_env(live, segments, need_docker=True):
+def check_container_vender(vender):
+    command = [vender, "version", "--help"]
+    all_output = ""
+    result_code = 0
+    for code, output in run_with_realtime_print(command):
+        result_code = code
+        all_output += output
+    if result_code != 0:
+        return False, f"{vender} is required"
+    if all_output.lower().find("json") < 0:
+        return False, f"Upgrade the {vender} to support json format"
+    # check alive
+    command = [vender, "ps"]
+    result_code = 0
+    for code, _ in run_with_realtime_print(command):
+        result_code = code
+    if result_code != 0:
+        return False, f"{vender} is not running"
+    return True, "ok"
+
+
+def check_the_env(live, segments, need_container=True, use_podman=False):
     # check the python version
     spinner = Spinner("dots", style="status.spinner", speed=1.0, text="")
     step = "Check the environment"
@@ -134,32 +155,14 @@ def check_the_env(live, segments, need_docker=True):
         segments.append(("❌", "Check the environment", "Python3.10 is required"))
         refresh(live, segments)
         return False, "Python3.10 is required"
-    if need_docker:
-        command = ["docker", "version", "--help"]
-        all_output = ""
-        result_code = 0
-        for code, output in run_with_realtime_print(command):
-            result_code = code
-            all_output += output
-        if result_code != 0:
-            old_segment = segments.pop()
-            segments.append(("❌", "Check the environment", "Docker is required"))
-            refresh(live, segments)
-            return False, "Docker is required"
-        if all_output.find("json") < 0:
-            old_segment = segments.pop()
-            segments.append(("❌", "Check the environment", "Please upgrade the docker"))
-            refresh(live, segments)
-            return False, "Upgrade the docker"
-        command = ["docker", "ps"]
-        result_code = 0
-        for code, _ in run_with_realtime_print(command):
-            result_code = code
-        if result_code != 0:
+    if need_container:
+        vender = "docker" if not use_podman else "podman"
+        result, msg = check_container_vender(vender)
+        if not result:
             old_segment = segments.pop()
-            segments.append(("❌", "Check the environment", "Docker is not running"))
+            segments.append(("❌", "Check the environment", msg))
             refresh(live, segments)
-            return False, "Docker is not running"
+            return False, msg
     old_segment = segments.pop()
     segments.append(("✅", "Check the environment", ""))
     refresh(live, segments)
@@ -225,21 +228,27 @@ def download_model(
     return result_code
 
 
-def load_docker_image(version, image_name, live, segments, chunk_size=1024):
+def load_docker_image(
+    version, image_name, live, segments, chunk_size=1024, use_podman=False
+):
     """
     download the image file and load it into docker
     """
-    full_name = f"{image_name}:{version}"
+    full_name = (
+        f"{image_name}:{version}"
+        if not use_podman
+        else f"docker.io/{image_name}:{version}"
+    )
     spinner = Spinner("dots", style="status.spinner", speed=1.0, text="")
     step = "Pull octogen image"
     segments.append((spinner, step, ""))
     refresh(live, segments)
     return_code = 0
     output = ""
-    for code, msg in run_with_realtime_print(command=["docker", "pull", full_name]):
+    vender = "docker" if not use_podman else "podman"
+    for code, msg in run_with_realtime_print(command=[vender, "pull", full_name]):
         return_code = code
         output += msg
-
     old_segment = segments.pop()
     if return_code == 0:
         segments.append(("✅", old_segment[1], full_name))
@@ -250,7 +259,7 @@ def load_docker_image(version, image_name, live, segments, chunk_size=1024):
 
 
 def choose_api_service(console):
-    mk = """Choose your favourite LLM
+    mk = """Choose your favourite Large Language Model
 1. OpenAI, Kernel, Agent and Cli will be installed
 2. Azure OpenAI, Kernel, Agent and Cli will be installed
 3. Codellama, Llama.cpp Model Server, Kernel, Agent and Cli will be installed
@@ -352,22 +361,30 @@ def generate_kernel_env(live, segments, install_dir, rpc_key):
     refresh(live, segments)
 
 
-def stop_service(name):
-    command = ["docker", "ps", "-f", f"name={name}", "--format", "json"]
+def stop_service(name, use_podman=False):
+    vender = "docker" if not use_podman else "podman"
+    command = [vender, "ps", "-f", f"name={name}", "--format", "json"]
     output = ""
     for _, chunk in run_with_realtime_print(command=command):
         output += chunk
         pass
-    if output:
+    if use_podman and output:
+        rows = json.loads(output.strip())
+        for row in rows:
+            id = row["Id"]
+            command = [vender, "kill", id]
+            for _, chunk in run_with_realtime_print(command=command):
+                pass
+    elif output:
         for line in output.split(os.linesep):
             if not line:
                 break
             row = json.loads(line.strip())
             id = row["ID"]
-            command = ["docker", "kill", id]
+            command = [vender, "kill", id]
             for _, chunk in run_with_realtime_print(command=command):
                 pass
-    command = ["docker", "container", "rm", name]
+    command = [vender, "container", "rm", name]
     for _, chunk in run_with_realtime_print(command=command):
         pass
 
@@ -380,17 +397,19 @@ def start_service(
     version,
     is_codellama="1",
     model_filename="",
+    use_podman=False,
 ):
     spinner = Spinner("dots", style="status.spinner", speed=1.0, text="")
     step = "Start octogen service"
     output = ""
+    vender = "docker" if not use_podman else "podman"
     segments.append((spinner, step, ""))
     refresh(live, segments)
-    stop_service("octogen")
+    stop_service("octogen", use_podman=use_podman)
     # TODO stop the exist service
     full_name = f"{image_name}:{version}"
     command = [
-        "docker",
+        vender,
         "run",
         "--name",
         "octogen",
@@ -441,6 +460,7 @@ def start_octogen_for_openai(
     version,
     api_key,
     model,
+    use_podman=False,
 ):
     generate_agent_openai(live, segments, install_dir, admin_key, api_key, model)
     if (
@@ -451,6 +471,7 @@ def start_octogen_for_openai(
             image_name,
             version,
             is_codellama="0",
+            use_podman=use_podman,
         )
         == 0
     ):
@@ -476,6 +497,7 @@ def start_octogen_for_azure_openai(
     api_key,
     model,
     api_base,
+    use_podman=False,
 ):
     generate_agent_azure_openai(
         live, segments, install_dir, admin_key, api_key, model, api_base
@@ -513,6 +535,7 @@ def start_octogen_for_codellama(
     image_name,
     version,
     socks_proxy="",
+    use_podman=False,
 ):
     """
     start the octogen service for codellama
@@ -567,6 +590,11 @@ def start_octogen_for_codellama(
     default="codellama-7b-instruct.Q5_K_S.gguf",
     help="the model filename in model repo",
 )
+@click.option(
+    "--use_podman",
+    is_flag=True,
+    help="use podman as the container engine",
+)
 def init_octogen(
     image_name,
     repo_name,
@@ -576,6 +604,7 @@ def init_octogen(
     socks_proxy,
     codellama_repo,
     model_filename,
+    use_podman,
 ):
     if cli_dir.find("~") == 0:
         real_cli_dir = cli_dir.replace("~", os.path.expanduser("~"))
@@ -593,7 +622,7 @@ def init_octogen(
     with Live(Group(*segments), console=console) as live:
         run_install_cli(live, segments)
         if choice == "4":
-            check_result, _ = check_the_env(live, segments, need_docker=False)
+            check_result, _ = check_the_env(live, segments, need_container=False)
             if not check_result:
                 segments.append(("❌", "Setup octogen failed", ""))
                 refresh(live, segments)
@@ -602,7 +631,9 @@ def init_octogen(
             segments.append(("👍", "Setup octogen done", ""))
             refresh(live, segments)
             return
-        check_result, _ = check_the_env(live, segments, need_docker=True)
+        check_result, _ = check_the_env(
+            live, segments, need_container=True, use_podman=use_podman
+        )
         if not check_result:
             segments.append(("❌", "Setup octogen failed", ""))
             refresh(live, segments)
@@ -612,7 +643,9 @@ def init_octogen(
         else:
             version = get_latest_release_version(repo_name, live, segments)
 
-        code = load_docker_image(version, image_name, live, segments)
+        code = load_docker_image(
+            version, image_name, live, segments, use_podman=use_podman
+        )
         if code != 0:
             return
         kernel_key = random_str(32)
@@ -632,6 +665,7 @@ def init_octogen(
                 image_name,
                 version,
                 socks_proxy,
+                use_podman=use_podman,
             )
         elif choice == "2":
             # start azure openai
@@ -647,6 +681,7 @@ def init_octogen(
                 key,
                 model,
                 api_base,
+                use_podman=use_podman,
             )
         else:
             # start for openai
@@ -661,4 +696,5 @@ def init_octogen(
                 version,
                 key,
                 model,
+                use_podman=use_podman,
             )

up/tests/up_tests.py

@@ -111,7 +111,7 @@ def test_check_the_env_win():
     console = Console()
     segments = []
     with Live(Group(*segments), console=console) as live:
-        result, msg = check_the_env(live, segments, need_docker=False)
+        result, msg = check_the_env(live, segments, need_container=False)
         assert result