cyclo-manager 1.1.0.dev3__tar.gz → 1.1.0.dev5__tar.gz

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cyclo-manager
-Version: 1.1.0.dev3
+Version: 1.1.0.dev5
 Summary: cyclo_manager CLI: pip-installable launcher for cyclo_manager server and UI containers. Run 'cyclo_manager up' to start Docker stack.
 License-Expression: Apache-2.0
 Requires-Python: >=3.10
@@ -9,7 +9,6 @@ Requires-Dist: fastapi>=0.100.0
 Requires-Dist: uvicorn[standard]>=0.23.0
 Requires-Dist: psutil>=5.9.0
 Provides-Extra: dev
-Requires-Dist: pytest; extra == "dev"
 Requires-Dist: ruff; extra == "dev"
 
 # cyclo_manager CLI
@@ -38,6 +37,7 @@ For the full monorepo (API source, UI source, dev compose), see the [repository
 - [What `cyclo_manager up` does](#what-cyclo_manager-up-does)
 - [Host agent (`cyclo_host_agent`)](#host-agent-cyclo_host_agent)
 - [Compose services](#compose-services)
+- [Recording storage](#recording-storage)
 - [Configuration](#configuration)
 - [Agent sockets on the host](#agent-sockets-on-the-host)
 - [Environment variables](#environment-variables)
@@ -133,7 +133,12 @@ The API container reaches it at `/agents/host/host_agent.sock` (see bundled `con
 - Check update availability from remote tags and local `package.xml` versions.
 - Update repos on allowed branches (`main`, `jazzy`) using stash or reset workflows.
 - Stop/start a repo's `docker/container.sh` helper during an update when requested by the UI.
+- Run container starts as background jobs with a 10-minute timeout covering the helper and its output streams; timeout or cancellation terminates the helper's process group and ends the job.
 - Run Cyclo Manager package updates from the UI by delegating to `cyclo_manager update`.
+- Provide host CPU/memory/disk statistics and process information. CPU summaries
+  share a background sampler's average of the latest three one-second samples.
+- Serve the Files UI's browsing, filename search, upload, text editing, Git diff
+  and file/directory management operations under the configured workspace.
 
 `cyclo_manager up` installs the `refresh-host-agent` CLI subcommand into sudoers. It is registered as `sys.executable -m cyclo_manager_cli.cli refresh-host-agent` with `SETENV`, so UI-triggered updates can pass the current `PYTHONPATH` to sudo and do not depend on `PATH`. The command is idempotent and refreshes the socket directory, sudoers file, systemd unit, and service state after package upgrades.
 
@@ -145,10 +150,10 @@ Repository scanning and the Files UI use `CYCLO_HOST_AGENT_WORKSPACE` when set.
 
 Defined in [`cyclo_manager_cli/docker/docker-compose.yml`](cyclo_manager_cli/docker/docker-compose.yml). All services use **`network_mode: host`**.
 
-| Compose service | Container name | `cyclo_manager up` | Image (example) |
+| Compose service | Container name | `cyclo_manager up` | Image in this checkout |
 |-----------------|----------------|--------------------|-----------------|
-| `cyclo_manager` | `cyclo_manager` | **Started** | `robotis/cyclo-manager:1.0.1` |
-| `ui` | `cyclo_manager_ui` | **Started** | `robotis/cyclo-manager-ui:1.0.1` |
+| `cyclo_manager` | `cyclo_manager` | **Started** | `robotis/cyclo-manager:1.1.0` |
+| `ui` | `cyclo_manager_ui` | **Started** | `robotis/cyclo-manager-ui:1.1.0` |
 | `rmw_zenoh` | `zenoh_daemon` | **Created only** | `robotis/zenoh-daemon:latest` |
 | `novnc-server` | `novnc-server` | **Created only** | `robotis/novnc-server:latest` |
 
@@ -156,6 +161,27 @@ Start optional containers from the UI or manually, e.g. `docker start zenoh_daem
 
 ---
 
+## Recording storage
+
+The packaged stack and dev compose both mount:
+
+```yaml
+- ${HOME}/cyclo_manager_ros_bags:/cyclo_manager_ros_bags
+```
+
+Compose resolves `${HOME}` on the host at invocation time. The API container uses
+`RECORDINGS_DIR=/cyclo_manager_ros_bags`, which is also the server default when the
+variable is absent. The host directory is independent of
+`CYCLO_HOST_AGENT_WORKSPACE`; each recording stores its metadata and MCAP bag under
+its generated ID. Container recreation, `down` and package updates preserve this
+bind-mounted data. Applying a new mount requires container recreation.
+
+Recording and playback run in the API container using `rosbag2_py` and the MCAP
+storage plugin included in the manager image. They are not host-agent jobs.
+See [Record & Play](../docs/record-play.md) for capture, repeat playback and stops.
+
+---
+
 ## Configuration
 
 The CLI always mounts the **bundled** config into the API container. There is **no** `-c` / `--config` flag on the pip CLI.
@@ -168,10 +194,13 @@ The API reads it as `CONFIG_FILE=/app/config.yml` inside the container.
 
 | Key | Description |
 |-----|-------------|
-| **`supported_robot_containers`** | Robot Docker container names that can open the System page (e.g. `ai_worker`, `open_manipulator`). Each must be a key in `sockets` and cannot be `host_agent`. |
+| **`supported_robot_containers`** | Containers eligible for System navigation and server-side motion profile detection (e.g. `ai_worker`, `open_manipulator`). Each must be a key in `sockets` and cannot be `host_agent`. |
 | **`sockets`** | Map of logical name → agent **Unix socket path as seen inside the API container** (under `/agents/...`). Include robot/service containers and `host_agent`. |
 
-s6 **service names** are not listed in config; each in-container agent reports them at runtime.
+s6 **service names** are defined in the manager's System/motion profiles, not in
+this config. Motion detection checks the existing individual status endpoints for
+`ai_worker_bringup` and `open_manipulator_bringup`, then reads `/run/robot_type`
+inside the selected container through Docker. This needs no new s6-agent endpoint.
 
 ### Example (bundled default)
 
@@ -198,6 +227,7 @@ On the host, sockets typically live under:
 ```text
 /var/run/robotis/agent_sockets/
 ├── ai_worker/s6_agent.sock
+├── open_manipulator/s6_agent.sock
 ├── cyclo_intelligence/s6_agent.sock
 └── host/host_agent.sock          ← created by cyclo_manager up
 ```
@@ -220,7 +250,11 @@ So `/agents/ai_worker/s6_agent.sock` in config corresponds to the host path abov
 | **`CYCLO_HOST_AGENT_WORKSPACE`** | User or CLI-generated systemd unit | Workspace scanned by `cyclo_host_agent` for managed git repositories and used as the Files UI/API root |
 | **`HOSTNAME`** | CLI (default: machine hostname) | Passed to API as `HOST_HOSTNAME` |
 | **`CONFIG_FILE`** | Compose (`/app/config.yml`) | Path inside the API container |
-| **`ROS_DOMAIN_ID`** | **Not** set by CLI | Set inside robot containers (e.g. `~/.bashrc`) so DDS matches your fleet |
+| **`HOME`** | Host environment used by Compose | Selects the host `${HOME}/cyclo_manager_ros_bags` bind-mount directory |
+| **`RECORDINGS_DIR`** | Compose (`/cyclo_manager_ros_bags`) | Server-side recording directory inside the API container; same default without the variable |
+| **`ROS_DOMAIN_ID`** | Manager image shell startup (`30`); not set by CLI | Must match the robot; the manager image configures it in `/root/.bashrc`, which the API startup sources |
+| **`RMW_IMPLEMENTATION`** | Manager image shell startup (`rmw_zenoh_cpp`) | ROS middleware used by the manager; requires a reachable Zenoh router |
+| **`NEXT_PUBLIC_API_URL`** | UI dev/build environment | Browser API address; production bundles embed it at build time, so a compose runtime value does not reconfigure a prebuilt UI |
 
 ---
 
@@ -234,6 +268,12 @@ With `network_mode: host` and default ports:
 | cyclo_manager API | http://127.0.0.1:8081 |
 | OpenAPI (Swagger) | http://127.0.0.1:8081/docs |
 
+These loopback URLs apply on the robot host. From another device, use the robot's
+hostname/IP and a browser-reachable API URL. Both compose files contain a UI
+runtime value of `http://127.0.0.1:8081`; it takes effect in development, while a
+production bundle uses its build-time value or hostname fallback. See the
+[UI API address guide](../cyclo_manager_ui/README_UI.md#api-address).
+
 ---
 
 ## Custom config and development
@@ -247,6 +287,11 @@ For development (local API/UI source, custom `config.yml`, hot reload):
 docker compose -f docker-compose.dev.yml up
 ```
 
+The dev compose command starts all defined services and does not install the host
+agent. API/UI source mounts do not update the Python code used by the separately
+installed `cyclo_host_agent.service`; install the updated CLI package and refresh
+that service when testing host-agent changes.
+
 See the [root README](../README.md#development).
 
 ---
@@ -259,7 +304,7 @@ Python dependencies declared in [`pyproject.toml`](pyproject.toml):
 |---------|---------|
 | **fastapi** | `cyclo_host_agent` HTTP API |
 | **uvicorn** | `cyclo_host_agent` server |
-| **psutil** | Declared package dependency for host/system utility support |
+| **psutil** | Host CPU sampler, memory/disk statistics and process inspection |
 
 The **`cyclo_manager`** CLI itself uses only the Python standard library plus **`subprocess`** calls to **`docker compose`**, **`pip`**, **`systemctl`**, and **`sudo`**.
 

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/README.md

@@ -24,6 +24,7 @@ For the full monorepo (API source, UI source, dev compose), see the [repository
 - [What `cyclo_manager up` does](#what-cyclo_manager-up-does)
 - [Host agent (`cyclo_host_agent`)](#host-agent-cyclo_host_agent)
 - [Compose services](#compose-services)
+- [Recording storage](#recording-storage)
 - [Configuration](#configuration)
 - [Agent sockets on the host](#agent-sockets-on-the-host)
 - [Environment variables](#environment-variables)
@@ -119,7 +120,12 @@ The API container reaches it at `/agents/host/host_agent.sock` (see bundled `con
 - Check update availability from remote tags and local `package.xml` versions.
 - Update repos on allowed branches (`main`, `jazzy`) using stash or reset workflows.
 - Stop/start a repo's `docker/container.sh` helper during an update when requested by the UI.
+- Run container starts as background jobs with a 10-minute timeout covering the helper and its output streams; timeout or cancellation terminates the helper's process group and ends the job.
 - Run Cyclo Manager package updates from the UI by delegating to `cyclo_manager update`.
+- Provide host CPU/memory/disk statistics and process information. CPU summaries
+  share a background sampler's average of the latest three one-second samples.
+- Serve the Files UI's browsing, filename search, upload, text editing, Git diff
+  and file/directory management operations under the configured workspace.
 
 `cyclo_manager up` installs the `refresh-host-agent` CLI subcommand into sudoers. It is registered as `sys.executable -m cyclo_manager_cli.cli refresh-host-agent` with `SETENV`, so UI-triggered updates can pass the current `PYTHONPATH` to sudo and do not depend on `PATH`. The command is idempotent and refreshes the socket directory, sudoers file, systemd unit, and service state after package upgrades.
 
@@ -131,10 +137,10 @@ Repository scanning and the Files UI use `CYCLO_HOST_AGENT_WORKSPACE` when set.
 
 Defined in [`cyclo_manager_cli/docker/docker-compose.yml`](cyclo_manager_cli/docker/docker-compose.yml). All services use **`network_mode: host`**.
 
-| Compose service | Container name | `cyclo_manager up` | Image (example) |
+| Compose service | Container name | `cyclo_manager up` | Image in this checkout |
 |-----------------|----------------|--------------------|-----------------|
-| `cyclo_manager` | `cyclo_manager` | **Started** | `robotis/cyclo-manager:1.0.1` |
-| `ui` | `cyclo_manager_ui` | **Started** | `robotis/cyclo-manager-ui:1.0.1` |
+| `cyclo_manager` | `cyclo_manager` | **Started** | `robotis/cyclo-manager:1.1.0` |
+| `ui` | `cyclo_manager_ui` | **Started** | `robotis/cyclo-manager-ui:1.1.0` |
 | `rmw_zenoh` | `zenoh_daemon` | **Created only** | `robotis/zenoh-daemon:latest` |
 | `novnc-server` | `novnc-server` | **Created only** | `robotis/novnc-server:latest` |
 
@@ -142,6 +148,27 @@ Start optional containers from the UI or manually, e.g. `docker start zenoh_daem
 
 ---
 
+## Recording storage
+
+The packaged stack and dev compose both mount:
+
+```yaml
+- ${HOME}/cyclo_manager_ros_bags:/cyclo_manager_ros_bags
+```
+
+Compose resolves `${HOME}` on the host at invocation time. The API container uses
+`RECORDINGS_DIR=/cyclo_manager_ros_bags`, which is also the server default when the
+variable is absent. The host directory is independent of
+`CYCLO_HOST_AGENT_WORKSPACE`; each recording stores its metadata and MCAP bag under
+its generated ID. Container recreation, `down` and package updates preserve this
+bind-mounted data. Applying a new mount requires container recreation.
+
+Recording and playback run in the API container using `rosbag2_py` and the MCAP
+storage plugin included in the manager image. They are not host-agent jobs.
+See [Record & Play](../docs/record-play.md) for capture, repeat playback and stops.
+
+---
+
 ## Configuration
 
 The CLI always mounts the **bundled** config into the API container. There is **no** `-c` / `--config` flag on the pip CLI.
@@ -154,10 +181,13 @@ The API reads it as `CONFIG_FILE=/app/config.yml` inside the container.
 
 | Key | Description |
 |-----|-------------|
-| **`supported_robot_containers`** | Robot Docker container names that can open the System page (e.g. `ai_worker`, `open_manipulator`). Each must be a key in `sockets` and cannot be `host_agent`. |
+| **`supported_robot_containers`** | Containers eligible for System navigation and server-side motion profile detection (e.g. `ai_worker`, `open_manipulator`). Each must be a key in `sockets` and cannot be `host_agent`. |
 | **`sockets`** | Map of logical name → agent **Unix socket path as seen inside the API container** (under `/agents/...`). Include robot/service containers and `host_agent`. |
 
-s6 **service names** are not listed in config; each in-container agent reports them at runtime.
+s6 **service names** are defined in the manager's System/motion profiles, not in
+this config. Motion detection checks the existing individual status endpoints for
+`ai_worker_bringup` and `open_manipulator_bringup`, then reads `/run/robot_type`
+inside the selected container through Docker. This needs no new s6-agent endpoint.
 
 ### Example (bundled default)
 
@@ -184,6 +214,7 @@ On the host, sockets typically live under:
 ```text
 /var/run/robotis/agent_sockets/
 ├── ai_worker/s6_agent.sock
+├── open_manipulator/s6_agent.sock
 ├── cyclo_intelligence/s6_agent.sock
 └── host/host_agent.sock          ← created by cyclo_manager up
 ```
@@ -206,7 +237,11 @@ So `/agents/ai_worker/s6_agent.sock` in config corresponds to the host path abov
 | **`CYCLO_HOST_AGENT_WORKSPACE`** | User or CLI-generated systemd unit | Workspace scanned by `cyclo_host_agent` for managed git repositories and used as the Files UI/API root |
 | **`HOSTNAME`** | CLI (default: machine hostname) | Passed to API as `HOST_HOSTNAME` |
 | **`CONFIG_FILE`** | Compose (`/app/config.yml`) | Path inside the API container |
-| **`ROS_DOMAIN_ID`** | **Not** set by CLI | Set inside robot containers (e.g. `~/.bashrc`) so DDS matches your fleet |
+| **`HOME`** | Host environment used by Compose | Selects the host `${HOME}/cyclo_manager_ros_bags` bind-mount directory |
+| **`RECORDINGS_DIR`** | Compose (`/cyclo_manager_ros_bags`) | Server-side recording directory inside the API container; same default without the variable |
+| **`ROS_DOMAIN_ID`** | Manager image shell startup (`30`); not set by CLI | Must match the robot; the manager image configures it in `/root/.bashrc`, which the API startup sources |
+| **`RMW_IMPLEMENTATION`** | Manager image shell startup (`rmw_zenoh_cpp`) | ROS middleware used by the manager; requires a reachable Zenoh router |
+| **`NEXT_PUBLIC_API_URL`** | UI dev/build environment | Browser API address; production bundles embed it at build time, so a compose runtime value does not reconfigure a prebuilt UI |
 
 ---
 
@@ -220,6 +255,12 @@ With `network_mode: host` and default ports:
 | cyclo_manager API | http://127.0.0.1:8081 |
 | OpenAPI (Swagger) | http://127.0.0.1:8081/docs |
 
+These loopback URLs apply on the robot host. From another device, use the robot's
+hostname/IP and a browser-reachable API URL. Both compose files contain a UI
+runtime value of `http://127.0.0.1:8081`; it takes effect in development, while a
+production bundle uses its build-time value or hostname fallback. See the
+[UI API address guide](../cyclo_manager_ui/README_UI.md#api-address).
+
 ---
 
 ## Custom config and development
@@ -233,6 +274,11 @@ For development (local API/UI source, custom `config.yml`, hot reload):
 docker compose -f docker-compose.dev.yml up
 ```
 
+The dev compose command starts all defined services and does not install the host
+agent. API/UI source mounts do not update the Python code used by the separately
+installed `cyclo_host_agent.service`; install the updated CLI package and refresh
+that service when testing host-agent changes.
+
 See the [root README](../README.md#development).
 
 ---
@@ -245,7 +291,7 @@ Python dependencies declared in [`pyproject.toml`](pyproject.toml):
 |---------|---------|
 | **fastapi** | `cyclo_host_agent` HTTP API |
 | **uvicorn** | `cyclo_host_agent` server |
-| **psutil** | Declared package dependency for host/system utility support |
+| **psutil** | Host CPU sampler, memory/disk statistics and process inspection |
 
 The **`cyclo_manager`** CLI itself uses only the Python standard library plus **`subprocess`** calls to **`docker compose`**, **`pip`**, **`systemctl`**, and **`sudo`**.
 

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/cyclo_host_agent/routers/repos.py

@@ -22,6 +22,7 @@ import asyncio
 import os
 from pathlib import Path
 import re
+import signal
 import time
 from typing import Optional
 
@@ -45,6 +46,7 @@ router = APIRouter(prefix='/repos', tags=['repos'])
 GIT_TIMEOUT = 120.0
 GIT_REMOTE_TIMEOUT = 5.0
 CONTAINER_SCRIPT_TIMEOUT = 600.0
+CONTAINER_TERMINATE_TIMEOUT = 2.0
 CONTAINER_JOB_RETENTION_SECONDS = 3600.0
 CONTAINER_JOB_OUTPUT_LIMIT = 40000
 
@@ -429,13 +431,28 @@ async def _stop_container_start_process(
     proc: asyncio.subprocess.Process | None,
     readers: list[asyncio.Task],
 ) -> None:
-    """Stop a container-start subprocess and consume its reader tasks."""
-    if proc is not None and proc.returncode is None:
+    """Stop the isolated process group, with bounded pipe and process cleanup."""
+    if proc is not None:
+        def signal_group(sig):
+            try:
+                os.killpg(proc.pid, sig)
+            except ProcessLookupError:
+                pass
+
+        signal_group(signal.SIGTERM)
         try:
-            proc.kill()
-        except ProcessLookupError:
+            await asyncio.wait_for(proc.wait(), CONTAINER_TERMINATE_TIMEOUT)
+        except asyncio.TimeoutError:
             pass
-        await proc.wait()
+        # Children may still exist even after their shell has exited.
+        signal_group(signal.SIGKILL)
+        try:
+            await asyncio.wait_for(proc.wait(), CONTAINER_TERMINATE_TIMEOUT)
+        except asyncio.TimeoutError:
+            pass
+    for reader in readers:
+        if not reader.done():
+            reader.cancel()
     await asyncio.gather(*readers, return_exceptions=True)
 
 
@@ -477,6 +494,7 @@ async def _run_container_start(repo_path: Path, name: str) -> None:
             'bash', str(script), 'start',
             cwd=str(repo_path / 'docker'),
             env=env,
+            start_new_session=True,
             stdin=asyncio.subprocess.PIPE,
             stdout=asyncio.subprocess.PIPE,
             stderr=asyncio.subprocess.PIPE,
@@ -490,28 +508,31 @@ async def _run_container_start(repo_path: Path, name: str) -> None:
             asyncio.create_task(_read_container_job_stream(proc.stdout, job)),
             asyncio.create_task(_read_container_job_stream(proc.stderr, job)),
         ]
-        await asyncio.wait_for(proc.wait(), timeout=CONTAINER_SCRIPT_TIMEOUT)
-        await asyncio.gather(*readers)
+        await asyncio.wait_for(
+            asyncio.gather(proc.wait(), *readers), timeout=CONTAINER_SCRIPT_TIMEOUT)
     except asyncio.TimeoutError:
         error = 'Timeout waiting for container.sh'
-        await _stop_container_start_process(proc, readers)
-        _append_container_job_output(job, error)
-        _set_container_job_status(job, running=False, success=False, error=error)
+        try:
+            await _stop_container_start_process(proc, readers)
+        finally:
+            _append_container_job_output(job, error)
+            _set_container_job_status(job, running=False, success=False, error=error)
         return
     except asyncio.CancelledError:
-        await _stop_container_start_process(proc, readers)
-        _set_container_job_status(
-            job,
-            running=False,
-            success=False,
-            error='Container start operation was cancelled',
-        )
+        try:
+            await _stop_container_start_process(proc, readers)
+        finally:
+            _set_container_job_status(
+                job, running=False, success=False,
+                error='Container start operation was cancelled')
         raise
     except Exception as exc:
-        await _stop_container_start_process(proc, readers)
         error = str(exc) or type(exc).__name__
-        _append_container_job_output(job, error)
-        _set_container_job_status(job, running=False, success=False, error=error)
+        try:
+            await _stop_container_start_process(proc, readers)
+        finally:
+            _append_container_job_output(job, error)
+            _set_container_job_status(job, running=False, success=False, error=error)
         return
 
     assert proc is not None

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/cyclo_manager.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: cyclo-manager
-Version: 1.1.0.dev3
+Version: 1.1.0.dev5
 Summary: cyclo_manager CLI: pip-installable launcher for cyclo_manager server and UI containers. Run 'cyclo_manager up' to start Docker stack.
 License-Expression: Apache-2.0
 Requires-Python: >=3.10
@@ -9,7 +9,6 @@ Requires-Dist: fastapi>=0.100.0
 Requires-Dist: uvicorn[standard]>=0.23.0
 Requires-Dist: psutil>=5.9.0
 Provides-Extra: dev
-Requires-Dist: pytest; extra == "dev"
 Requires-Dist: ruff; extra == "dev"
 
 # cyclo_manager CLI
@@ -38,6 +37,7 @@ For the full monorepo (API source, UI source, dev compose), see the [repository
 - [What `cyclo_manager up` does](#what-cyclo_manager-up-does)
 - [Host agent (`cyclo_host_agent`)](#host-agent-cyclo_host_agent)
 - [Compose services](#compose-services)
+- [Recording storage](#recording-storage)
 - [Configuration](#configuration)
 - [Agent sockets on the host](#agent-sockets-on-the-host)
 - [Environment variables](#environment-variables)
@@ -133,7 +133,12 @@ The API container reaches it at `/agents/host/host_agent.sock` (see bundled `con
 - Check update availability from remote tags and local `package.xml` versions.
 - Update repos on allowed branches (`main`, `jazzy`) using stash or reset workflows.
 - Stop/start a repo's `docker/container.sh` helper during an update when requested by the UI.
+- Run container starts as background jobs with a 10-minute timeout covering the helper and its output streams; timeout or cancellation terminates the helper's process group and ends the job.
 - Run Cyclo Manager package updates from the UI by delegating to `cyclo_manager update`.
+- Provide host CPU/memory/disk statistics and process information. CPU summaries
+  share a background sampler's average of the latest three one-second samples.
+- Serve the Files UI's browsing, filename search, upload, text editing, Git diff
+  and file/directory management operations under the configured workspace.
 
 `cyclo_manager up` installs the `refresh-host-agent` CLI subcommand into sudoers. It is registered as `sys.executable -m cyclo_manager_cli.cli refresh-host-agent` with `SETENV`, so UI-triggered updates can pass the current `PYTHONPATH` to sudo and do not depend on `PATH`. The command is idempotent and refreshes the socket directory, sudoers file, systemd unit, and service state after package upgrades.
 
@@ -145,10 +150,10 @@ Repository scanning and the Files UI use `CYCLO_HOST_AGENT_WORKSPACE` when set.
 
 Defined in [`cyclo_manager_cli/docker/docker-compose.yml`](cyclo_manager_cli/docker/docker-compose.yml). All services use **`network_mode: host`**.
 
-| Compose service | Container name | `cyclo_manager up` | Image (example) |
+| Compose service | Container name | `cyclo_manager up` | Image in this checkout |
 |-----------------|----------------|--------------------|-----------------|
-| `cyclo_manager` | `cyclo_manager` | **Started** | `robotis/cyclo-manager:1.0.1` |
-| `ui` | `cyclo_manager_ui` | **Started** | `robotis/cyclo-manager-ui:1.0.1` |
+| `cyclo_manager` | `cyclo_manager` | **Started** | `robotis/cyclo-manager:1.1.0` |
+| `ui` | `cyclo_manager_ui` | **Started** | `robotis/cyclo-manager-ui:1.1.0` |
 | `rmw_zenoh` | `zenoh_daemon` | **Created only** | `robotis/zenoh-daemon:latest` |
 | `novnc-server` | `novnc-server` | **Created only** | `robotis/novnc-server:latest` |
 
@@ -156,6 +161,27 @@ Start optional containers from the UI or manually, e.g. `docker start zenoh_daem
 
 ---
 
+## Recording storage
+
+The packaged stack and dev compose both mount:
+
+```yaml
+- ${HOME}/cyclo_manager_ros_bags:/cyclo_manager_ros_bags
+```
+
+Compose resolves `${HOME}` on the host at invocation time. The API container uses
+`RECORDINGS_DIR=/cyclo_manager_ros_bags`, which is also the server default when the
+variable is absent. The host directory is independent of
+`CYCLO_HOST_AGENT_WORKSPACE`; each recording stores its metadata and MCAP bag under
+its generated ID. Container recreation, `down` and package updates preserve this
+bind-mounted data. Applying a new mount requires container recreation.
+
+Recording and playback run in the API container using `rosbag2_py` and the MCAP
+storage plugin included in the manager image. They are not host-agent jobs.
+See [Record & Play](../docs/record-play.md) for capture, repeat playback and stops.
+
+---
+
 ## Configuration
 
 The CLI always mounts the **bundled** config into the API container. There is **no** `-c` / `--config` flag on the pip CLI.
@@ -168,10 +194,13 @@ The API reads it as `CONFIG_FILE=/app/config.yml` inside the container.
 
 | Key | Description |
 |-----|-------------|
-| **`supported_robot_containers`** | Robot Docker container names that can open the System page (e.g. `ai_worker`, `open_manipulator`). Each must be a key in `sockets` and cannot be `host_agent`. |
+| **`supported_robot_containers`** | Containers eligible for System navigation and server-side motion profile detection (e.g. `ai_worker`, `open_manipulator`). Each must be a key in `sockets` and cannot be `host_agent`. |
 | **`sockets`** | Map of logical name → agent **Unix socket path as seen inside the API container** (under `/agents/...`). Include robot/service containers and `host_agent`. |
 
-s6 **service names** are not listed in config; each in-container agent reports them at runtime.
+s6 **service names** are defined in the manager's System/motion profiles, not in
+this config. Motion detection checks the existing individual status endpoints for
+`ai_worker_bringup` and `open_manipulator_bringup`, then reads `/run/robot_type`
+inside the selected container through Docker. This needs no new s6-agent endpoint.
 
 ### Example (bundled default)
 
@@ -198,6 +227,7 @@ On the host, sockets typically live under:
 ```text
 /var/run/robotis/agent_sockets/
 ├── ai_worker/s6_agent.sock
+├── open_manipulator/s6_agent.sock
 ├── cyclo_intelligence/s6_agent.sock
 └── host/host_agent.sock          ← created by cyclo_manager up
 ```
@@ -220,7 +250,11 @@ So `/agents/ai_worker/s6_agent.sock` in config corresponds to the host path abov
 | **`CYCLO_HOST_AGENT_WORKSPACE`** | User or CLI-generated systemd unit | Workspace scanned by `cyclo_host_agent` for managed git repositories and used as the Files UI/API root |
 | **`HOSTNAME`** | CLI (default: machine hostname) | Passed to API as `HOST_HOSTNAME` |
 | **`CONFIG_FILE`** | Compose (`/app/config.yml`) | Path inside the API container |
-| **`ROS_DOMAIN_ID`** | **Not** set by CLI | Set inside robot containers (e.g. `~/.bashrc`) so DDS matches your fleet |
+| **`HOME`** | Host environment used by Compose | Selects the host `${HOME}/cyclo_manager_ros_bags` bind-mount directory |
+| **`RECORDINGS_DIR`** | Compose (`/cyclo_manager_ros_bags`) | Server-side recording directory inside the API container; same default without the variable |
+| **`ROS_DOMAIN_ID`** | Manager image shell startup (`30`); not set by CLI | Must match the robot; the manager image configures it in `/root/.bashrc`, which the API startup sources |
+| **`RMW_IMPLEMENTATION`** | Manager image shell startup (`rmw_zenoh_cpp`) | ROS middleware used by the manager; requires a reachable Zenoh router |
+| **`NEXT_PUBLIC_API_URL`** | UI dev/build environment | Browser API address; production bundles embed it at build time, so a compose runtime value does not reconfigure a prebuilt UI |
 
 ---
 
@@ -234,6 +268,12 @@ With `network_mode: host` and default ports:
 | cyclo_manager API | http://127.0.0.1:8081 |
 | OpenAPI (Swagger) | http://127.0.0.1:8081/docs |
 
+These loopback URLs apply on the robot host. From another device, use the robot's
+hostname/IP and a browser-reachable API URL. Both compose files contain a UI
+runtime value of `http://127.0.0.1:8081`; it takes effect in development, while a
+production bundle uses its build-time value or hostname fallback. See the
+[UI API address guide](../cyclo_manager_ui/README_UI.md#api-address).
+
 ---
 
 ## Custom config and development
@@ -247,6 +287,11 @@ For development (local API/UI source, custom `config.yml`, hot reload):
 docker compose -f docker-compose.dev.yml up
 ```
 
+The dev compose command starts all defined services and does not install the host
+agent. API/UI source mounts do not update the Python code used by the separately
+installed `cyclo_host_agent.service`; install the updated CLI package and refresh
+that service when testing host-agent changes.
+
 See the [root README](../README.md#development).
 
 ---
@@ -259,7 +304,7 @@ Python dependencies declared in [`pyproject.toml`](pyproject.toml):
 |---------|---------|
 | **fastapi** | `cyclo_host_agent` HTTP API |
 | **uvicorn** | `cyclo_host_agent` server |
-| **psutil** | Declared package dependency for host/system utility support |
+| **psutil** | Host CPU sampler, memory/disk statistics and process inspection |
 
 The **`cyclo_manager`** CLI itself uses only the Python standard library plus **`subprocess`** calls to **`docker compose`**, **`pip`**, **`systemctl`**, and **`sudo`**.
 

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/cyclo_manager.egg-info/SOURCES.txt

@@ -19,5 +19,4 @@ cyclo_manager.egg-info/top_level.txt
 cyclo_manager_cli/__init__.py
 cyclo_manager_cli/cli.py
 cyclo_manager_cli/config/config.yml
-cyclo_manager_cli/docker/docker-compose.yml
-tests/test_cpu_usage.py
+cyclo_manager_cli/docker/docker-compose.yml

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/cyclo_manager.egg-info/requires.txt

@@ -3,5 +3,4 @@ uvicorn[standard]>=0.23.0
 psutil>=5.9.0
 
 [dev]
-pytest
 ruff

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/cyclo_manager_cli/docker/docker-compose.yml

@@ -21,8 +21,10 @@ services:
       - ${CYCLO_MANAGER_CONFIG_FILE}:/app/config.yml # For hot-reload on config changes
       - /var/run/docker.sock:/var/run/docker.sock # For Docker API access
       - /:/host_root:ro # For host disk usage and OS info
+      - ${HOME}/cyclo_manager_ros_bags:/cyclo_manager_ros_bags
     environment:
       - CONFIG_FILE=/app/config.yml
+      - RECORDINGS_DIR=/cyclo_manager_ros_bags
       - HOST_HOSTNAME=${HOSTNAME}
 
   ui:

{cyclo_manager-1.1.0.dev3 → cyclo_manager-1.1.0.dev5}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "cyclo-manager"
-version = "1.1.0.dev3"
+version = "1.1.0.dev5"
 description = "cyclo_manager CLI: pip-installable launcher for cyclo_manager server and UI containers. Run 'cyclo_manager up' to start Docker stack."
 readme = "README.md"
 license = "Apache-2.0"
@@ -17,7 +17,6 @@ dependencies = [
 
 [project.optional-dependencies]
 dev = [
-    "pytest",
     "ruff",
 ]
 

cyclo_manager-1.1.0.dev3/tests/test_cpu_usage.py

@@ -1,169 +0,0 @@
-#!/usr/bin/env python3
-#
-# Copyright 2026 ROBOTIS CO., LTD.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-#
-# Author: Hyungyu Kim
-
-"""Regression tests for continuous CPU sampling and shared API summaries."""
-
-from types import SimpleNamespace
-import threading
-import time
-import unittest
-from unittest.mock import Mock, call, patch
-
-from cyclo_host_agent.cpu_usage import CpuUsageSampler
-from cyclo_host_agent.main import app, lifespan
-from cyclo_host_agent.routers import system_stats
-from fastapi import HTTPException
-
-
-class CpuUsageTests(unittest.TestCase):
-    """Check measurement windows, startup behavior, and sampling ownership."""
-
-    def test_rolling_window_discards_prime_and_oldest_sample(self):
-        """Average available samples at startup and only the last three afterward."""
-        sampler = CpuUsageSampler()
-        observed = []
-
-        def next_interval(interval):
-            self.assertEqual(interval, 1.0)
-            observed.append(sampler.average())
-            if len(observed) == 5:
-                sampler._stop.set()
-                return True
-            return False
-
-        with patch('cyclo_host_agent.cpu_usage.psutil.cpu_percent',
-                   side_effect=[99.0, 10.0, 20.0, 60.0, 90.0]) as cpu:
-            with patch.object(sampler._stop, 'wait', side_effect=next_interval):
-                sampler._run()
-        self.assertEqual(observed, [None, 10.0, 15.0, 30.0, 56.7])
-        self.assertEqual(cpu.call_args_list, [call(interval=None)] * 5)
-
-    def test_background_sampling_is_independent_of_readers(self):
-        """Prime and sample on one thread, even without API requests."""
-        sampler = CpuUsageSampler(interval=0.01)
-        thread_ids = []
-
-        def cpu_percent(interval):
-            thread_ids.append(threading.get_ident())
-            return 30.0
-
-        with patch('cyclo_host_agent.cpu_usage.psutil.cpu_percent',
-                   side_effect=cpu_percent):
-            sampler.start()
-            worker = sampler._thread
-            try:
-                sampler.start()
-                self.assertIs(sampler._thread, worker)
-                deadline = time.monotonic() + 2
-                while sampler.average() is None and time.monotonic() < deadline:
-                    time.sleep(0.005)
-                self.assertEqual(sampler.average(), 30.0)
-            finally:
-                sampler.stop()
-            self.assertFalse(worker.is_alive())
-            count = len(thread_ids)
-            for _ in range(50):
-                self.assertEqual(sampler.average(), 30.0)
-            self.assertEqual(len(thread_ids), count)
-            self.assertGreaterEqual(count, 2)
-            self.assertEqual(len(set(thread_ids)), 1)
-            self.assertNotEqual(thread_ids[0], threading.get_ident())
-            sampler._interval = 1.0
-            sampler.start()
-            try:
-                self.assertIsNone(sampler.average())
-            finally:
-                sampler.stop()
-
-    def test_warmup_does_not_report_a_fake_zero(self):
-        """Return a retryable response before a complete sample exists."""
-        with patch.object(system_stats, 'cpu_sampler', CpuUsageSampler()):
-            with self.assertRaises(HTTPException) as result:
-                system_stats.get_system_stats()
-        self.assertEqual(result.exception.status_code, 503)
-
-    def test_sampling_failure_clears_history_and_primes_again(self):
-        """Do not keep old CPU values when a measurement fails."""
-        sampler = CpuUsageSampler()
-        observed = []
-
-        def next_interval(interval):
-            observed.append(sampler.average())
-            if len(observed) == 5:
-                sampler._stop.set()
-                return True
-            return False
-
-        with patch('cyclo_host_agent.cpu_usage.psutil.cpu_percent',
-                   side_effect=[99.0, 10.0, OSError('CPU read failed'), 99.0, 60.0]), \
-                patch('cyclo_host_agent.cpu_usage.logger.exception') as log, \
-                patch.object(sampler._stop, 'wait', side_effect=next_interval):
-            sampler._run()
-        self.assertEqual(observed, [None, 10.0, None, None, 60.0])
-        log.assert_called_once()
-
-    def test_status_and_process_summary_read_the_same_average(self):
-        """Total CPU is shared while process rows retain their own measurements."""
-        process = Mock()
-        process.as_dict.return_value = {
-            'pid': 42, 'username': 'robot', 'cpu_percent': 80.0,
-            'memory_percent': 1.0, 'memory_info': SimpleNamespace(rss=1024),
-            'cmdline': ['robot-node'], 'name': 'robot-node',
-        }
-        with patch.object(system_stats.cpu_sampler, 'average', return_value=12.3), \
-                patch.object(system_stats.psutil, 'cpu_percent',
-                             side_effect=AssertionError('API must not resample total CPU')), \
-                patch.object(system_stats.psutil, 'virtual_memory',
-                             return_value=SimpleNamespace(used=1024**3, total=8*1024**3)), \
-                patch.object(system_stats.psutil, 'disk_usage',
-                             return_value=SimpleNamespace(used=10*1024**3, total=100*1024**3)), \
-                patch.object(system_stats, '_extra_storage_disk_usage', return_value=(None, None)), \
-                patch.object(system_stats, '_temperature', return_value=None), \
-                patch.object(system_stats.psutil, 'process_iter', return_value=[process]), \
-                patch.object(system_stats.psutil, 'cpu_count', return_value=4), \
-                patch.object(system_stats.time, 'sleep') as sleep:
-            status = system_stats.get_system_stats()
-            processes = system_stats._sample_processes()
-        self.assertEqual(status.cpu_percent, 12.3)
-        self.assertEqual(processes.cpu_percent, status.cpu_percent)
-        self.assertEqual(processes.processes[0].cpu_percent, 20.0)
-        process.cpu_percent.assert_called_once_with(None)
-        sleep.assert_called_once_with(0.2)
-
-
-class CpuLifespanTests(unittest.IsolatedAsyncioTestCase):
-    """Check sampler cleanup when the API exits."""
-
-    async def test_lifespan_stops_sampler_even_on_failure(self):
-        """Stop the background worker on both normal and exceptional shutdown."""
-        sampler = Mock()
-        with patch.object(system_stats, 'cpu_sampler', sampler):
-            async with lifespan(app):
-                sampler.start.assert_called_once_with()
-                sampler.stop.assert_not_called()
-            sampler.stop.assert_called_once_with()
-            sampler.reset_mock()
-            with self.assertRaises(RuntimeError):
-                async with lifespan(app):
-                    raise RuntimeError('shutdown')
-            sampler.start.assert_called_once_with()
-            sampler.stop.assert_called_once_with()
-
-
-if __name__ == '__main__':
-    unittest.main()