> For the complete documentation index, see [llms.txt](https://docs.roboflow.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.roboflow.com/deployment/self-hosted/enterprise/deployment-manager.md). # Deployment Manager Manage and monitor computer vision models deployed on edge hardware. ## About Roboflow Deployment Manager lets you easily set up, deploy, and manage computer vision models on edge devices. After training your model and building a workflow, it provides an all-in-one solution for scaling deployments. You can use Deployment Manager to: 1. Set up new devices with the [Roboflow Inference server](/deployment/self-hosted/self-hosted.md). 2. Configure camera streams for use in a deployment. 3. Deploy [Workflows](https://docs.roboflow.com/workflows) on the edge. 4. Monitor logs, stream status, and telemetry for deployed edge devices. {% hint style="warning" %} Deployment Manager is available exclusively for Enterprise customers. [Contact the Roboflow sales team](https://roboflow.com/sales) to learn more about this feature and how it can be used to manage your deployments at scale. {% endhint %}
### Key Requirements: Deployment Manager is built for Roboflow-supported hardware: * Supported: Hardware purchased via Roboflow * Unsupported (use at your own risk): NVIDIA Jetson, x86 machines running Debian-based Linux with NVIDIA GPUs * Not yet supported: Mac or Windows-based systems Devices must be connected to the internet for setup and ongoing operation, with continuous access to Roboflow for remote management and monitoring. ### Edge Services Alongside the inference server, a device can run additional Roboflow [Services](/deployment/self-hosted/enterprise/deployment-manager/services.md). Add and configure them from the device's Configuration tab, described in [Update Device Configuration](/deployment/self-hosted/enterprise/deployment-manager/making-changes/update-device-configuration.md). * [Event Store](/deployment/self-hosted/enterprise/deployment-manager/services/event-store.md) stores inference events on the device, with automatic retention and optional cloud backup. * [OPC UA Server](/deployment/self-hosted/enterprise/deployment-manager/services/opc-ua-server.md) publishes data as OPC UA tags for PLCs and SCADA systems. * [PLC Relay](/deployment/self-hosted/enterprise/deployment-manager/services/plc-relay.md) reads and writes PLC tags over Allen-Bradley, Modbus TCP, or Siemens S7. * [RTSP Simulator](/deployment/self-hosted/enterprise/deployment-manager/services/rtsp-simulator.md) streams an uploaded video file as an RTSP source for testing. Each service serves its own HTTP API on the device's address. See [Services](/deployment/self-hosted/enterprise/deployment-manager/services.md#using-the-apis) for the base URL, authentication, and error-shape rules they share. ## Guides ### Setting Up Follow the [Setting Up](/deployment/self-hosted/enterprise/deployment-manager/setting-up.md) guides in order to get a device online and running Workflows. | Guide | Description | | -------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------ | | [Hardware Requirements](/deployment/self-hosted/enterprise/deployment-manager/setting-up/hardware-requirements.md) | Supported hardware, cameras, network, and outbound traffic requirements. | | [Add a Device](/deployment/self-hosted/enterprise/deployment-manager/setting-up/add-a-device.md) | Provision a new edge device and register it with your account. | | [Add a Stream](/deployment/self-hosted/enterprise/deployment-manager/setting-up/add-a-stream.md) | Configure a camera stream that runs a Workflow. | | [Set up Device Alerts](/deployment/self-hosted/enterprise/deployment-manager/setting-up/set-up-device-alerts.md) | Enable email alerts for connectivity, disk, and FPS issues. | | [Setup Maintenance Windows](/deployment/self-hosted/enterprise/deployment-manager/setting-up/setup-maintenance-windows.md) | Schedule when deploy-affecting changes take effect. | ### Making Changes The [Making Changes](/deployment/self-hosted/enterprise/deployment-manager/making-changes.md) guides cover updating, reconfiguring, and removing devices and streams after setup. | Guide | Description | | -------------------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------ | | [Update Device Configuration](/deployment/self-hosted/enterprise/deployment-manager/making-changes/update-device-configuration.md) | Manage device settings, service versions, and additional services. | | [Configure Device Network](/deployment/self-hosted/enterprise/deployment-manager/making-changes/configure-device-network.md) | Set IP addressing, gateway, DNS, and hostname from the cloud. | | [Configure AI1 Camera Settings](/deployment/self-hosted/enterprise/deployment-manager/making-changes/configure-ai1-camera-settings.md) | Tune exposure, gain, focus, and other AI1 camera settings live. | | [Set a Static IP for a Camera](/deployment/self-hosted/enterprise/deployment-manager/making-changes/set-camera-static-ip.md) | Assign a persistent static IP to a GigE Basler or Lucid camera. | | [Soft Reset a PoE Port](/deployment/self-hosted/enterprise/deployment-manager/making-changes/soft-reset-poe-port.md) | Power-cycle a PoE port to recover stuck camera links. | | [Pause and Resume a Stream](/deployment/self-hosted/enterprise/deployment-manager/making-changes/stop-a-stream.md) | Temporarily pause a running stream and resume it later. | | [Trigger a Stream](/deployment/self-hosted/enterprise/deployment-manager/making-changes/trigger-a-stream.md) | Run a Workflow on demand against a Triggered stream. | | [Terminate a Stream](/deployment/self-hosted/enterprise/deployment-manager/making-changes/delete-a-stream.md) | Permanently remove a stream and its configuration. | | [Redeploy Deployment Manager](/deployment/self-hosted/enterprise/deployment-manager/making-changes/redeploy-deployment-manager.md) | Recover a device's configuration after a failure. | | [Delete a Device](/deployment/self-hosted/enterprise/deployment-manager/making-changes/delete-a-device.md) | Permanently remove a device you no longer use. | | [API Keys for Deployment Manager](/deployment/self-hosted/enterprise/deployment-manager/making-changes/api-keys-for-device-manager.md) | How device API keys are generated, scoped, and revoked. | ### Monitoring The [Monitoring](/deployment/self-hosted/enterprise/deployment-manager/monitoring.md) guides show how to check stream status, logs, resources, and Event Store health. | Guide | Description | | -------------------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------- | | [View a Stream](/deployment/self-hosted/enterprise/deployment-manager/monitoring/view-a-stream.md) | View a stream's status, latest frame, and Workflow details. | | [View Device Logs](/deployment/self-hosted/enterprise/deployment-manager/monitoring/view-device-logs.md) | Search, filter, and download logs from Roboflow services. | | [View the Resource Monitor](/deployment/self-hosted/enterprise/deployment-manager/monitoring/view-the-resource-monitor.md) | Monitor disk, memory, CPU, GPU, and service container health. | | [View Event Store Status](/deployment/self-hosted/enterprise/deployment-manager/monitoring/view-event-store-status.md) | Monitor Event Store usage, backup progress, and history. | ## HTTP API The Deployment Manager API lets you monitor and manage your Roboflow Deployment Manager (RFDM) devices programmatically. All endpoints are mounted under `/:workspace/devices/v2` on the public API host (`https://api.roboflow.com`). Read endpoints require a [scoped API key](https://docs.roboflow.com/reference/authentication/authentication/scoped-api-keys) with the `device:read` scope; endpoints that change a device require `device:update`. Workspace API keys with no explicit scopes list are implicitly granted all scopes (legacy behavior); keys that carry an explicit `scopes` array must include the relevant scope. With the Deployment Manager API, you can: * [List and Get Devices](#list-and-get-devices) * [Create a Device](#create-a-device) * [Device Config](#device-config) * [Device Commands](#device-commands) * [Device Streams](#device-streams) * [Device Logs and Telemetry](#device-logs-and-telemetry) * [Device Events](#device-events) * [Device Services](#device-services) * [Device Groups](#device-groups) * [Device Management](#device-management) The optional edge services (Event Store, PLC Relay, RTSP Simulator, OPC UA Server) serve their own HTTP APIs on the device's address rather than through `api.roboflow.com`. They do not accept a Roboflow API key, and their authentication and error shapes differ from everything below. See [Services](/deployment/self-hosted/enterprise/deployment-manager/services.md#using-the-apis). ### Authentication All endpoints accept a workspace API key via either: * Query string: `?api_key=YOUR_API_KEY` * Header: `Authorization: Bearer YOUR_API_KEY` #### Device-scoped API keys API keys minted for a specific device (for example, by RFDM during install) are narrowed to that device. They may only call routes whose `:deviceId` path parameter equals the key's bound device. They will receive **403** on the workspace-wide list and create endpoints, and on any `:deviceId` path that doesn't match. ### Cross-workspace isolation A workspace API key can only read or modify devices that belong to its workspace. Requests for a device id owned by a different workspace return **404**, even if the id is otherwise valid. ### Errors | Status | Meaning | | ------ | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | 400 | Invalid query parameter (unknown `time_period`, malformed cursor or date, invalid `direction`, etc.) or invalid request body. | | 401 | Missing or invalid API key. | | 403 | Device-scoped API key targeting a different device, the workspace-wide list/create routes, a source device from another workspace during duplication, or a feature-gated request rejected (for example, AI1 creation or `offline_mode` not enabled for the workspace). | | 404 | Device, stream, config, or source device not found. Read requests for device ids owned by another workspace also return 404. | | 429 | Rate limit exceeded. Log reads are limited to 120 requests per minute per workspace; telemetry reads are limited to 60 requests per minute per device with a 10-request burst over 10 seconds. | Error responses use one of two shapes: * Handler-level errors (typically 400, 403, 404, and 429 from the handlers themselves) return `{ "error": "" }`. * Authentication and workspace-validation failures (typically 401) return a structured `error` object: `{ "error": { "message": "...", "status": 401, "type": "OAuthException", "hint": "..." } }`. The logs rate limiter additionally returns a plain string body (not JSON) when it triggers a **429**. ### List and Get Devices #### List Devices List every device registered in the workspace. Device-scoped API keys cannot call this endpoint and receive **403**. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Get Device Retrieve a single device by id. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} ### Create a Device Create a new device in your workspace and return the identifiers needed to install it. Device-scoped API keys cannot call this endpoint and receive **403**. **Required scope:** `device:update` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} Use the returned `installId` to bootstrap the device with the Roboflow Deployment Manager installer. #### Notes on AI1 devices * Setting `device_type` to `"ai1"` requires the workspace to have the `deviceAio` feature enabled, otherwise the request returns **403**. * `offline_mode` is only valid for AI1 devices on workspaces with `roboflowLiteMode`. Other combinations return **400** or **403**. * For AI1 devices that are not duplicating an existing device and do not specify a `workflow_id`, the slugified `device_name` must contain at least one alphanumeric character, otherwise the request returns **400**. * When AI1 + offline mode applies, the response also includes an `offlineProvisioningQrPayload` field encoding the QR payload for offline provisioning. #### Notes on duplication When `sourceDeviceId` is supplied, the new device is created from a copy of the source device's config. The source device must belong to the same workspace as the request, otherwise the request returns **403**. A non-existent `sourceDeviceId` returns **404**. ### Device Config #### Get Workspace Default Config Returns the workspace-level default device configuration. This is the base default config with the workspace's config patch (if any) merged in. Use it as a template when provisioning a new device before calling `POST /:workspace/devices/v2`. **Required scope:** `device:read` {% hint style="info" %} This is a workspace-wide endpoint. Device-scoped API keys will receive **403** since there is no `:deviceId` in the path. {% endhint %} **Example Request** ```bash curl "https://api.roboflow.com/{workspace}/devices/v2/default-config?api_key=$ROBOFLOW_API_KEY" ``` **Example Response** ```json { "config": { "version": "1.0.0", "config": { "inference": { "confidence": 0.7, "threshold": 0.3 }, "device_type": "edge" }, "services": {} }, "patch": { "config": { "inference": { "confidence": 0.7 } } } } ``` * `config` -- the merged result (base default + workspace patch). * `patch` -- the workspace's stored patch. Empty object `{}` if the workspace has no custom patch. #### Update Workspace Default Config Replace the workspace's config patch. New devices start from the merged result. Send the whole patch each time, because the new one replaces the old one. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/default-config" method="patch" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Get Config Returns the device's current runtime configuration. The response shape mirrors the Roboflow Deployment Manager (RFDM) config spec. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/config" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% hint style="warning" %} The response includes the full configuration as authored. Per-service `environment_variables` and any integration credentials embedded in the config are returned unredacted. Treat the response body as sensitive and do not log it in plaintext. {% endhint %} #### Update Config Change part of a device's configuration (ex: a stream's Workflow, a confidence threshold, a service setting). Send only the fields you want to change: nested objects merge, `null` removes a key, and an array replaces the whole array. You do not need to send the full config back. On a device with [maintenance windows](/deployment/self-hosted/enterprise/deployment-manager/setting-up/setup-maintenance-windows.md), changes that affect the deployment wait for the next window, and the response returns `deferred: true`. Add `forceApply=true` to apply the change now along with everything already queued, or `immediateDelta=true` to apply only this change now and keep the rest queued. **Required scope:** `device:update` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/config" method="patch" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Config History List prior configuration revisions for a device, newest first. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/config/history" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Set Service Version Set the inference or manager version on one device without sending the whole config. Read the service's current `image` from [Get Config](#get-config) and send it with the new version. If the image changed since you read it, the request fails, so you never pin a version to the wrong image. On a device with [maintenance windows](/deployment/self-hosted/enterprise/deployment-manager/setting-up/setup-maintenance-windows.md), the change waits for the next window. To update many devices, call this once per device. Required scope: `device:update` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/config/service-version" method="patch" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} ### Device Commands Actions you start from the web app (ex: restart a device, pause a stream, scan for cameras) are queued as commands. The device picks them up on its next poll. #### Get Pending Commands Returns the commands waiting for a device, oldest first. Reading does not remove them: a command stays queued until the device reports back that it ran, so a device that restarts mid-command sees it again. Commands queued more than 5 minutes ago expire and are never delivered, because running them later would not match what you asked for. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/commands" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Queue a Command Queue a command for a device to run on its next poll. This does from the API what the web app does when you restart a device or start a stream. **Required scope:** `device:update` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/commands" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} ### Device Streams #### List Streams List every stream configured on a device. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/streams" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Get Stream Retrieve a single stream. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/streams/{streamId}" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Report Pipeline Status Report the inference pipelines a device is running. Each reported pipeline updates the matching stream, along with its FPS and sharpness metrics. A stream declared in the device config is matched by its stream key, so it stops showing as provisioning once the device reports it. RFDM posts this on devices that run without the device-manager container. Devices that run that container report the same data in their healthcheck, so they do not need this endpoint. **Required scope:** `device:update` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/pipelines-status" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} You can also post the inference server's `/inference_pipelines/list` response unchanged, as `{"success": true, "data": {"pipelines": [...]}}`. If that body has `"success": false`, the device failed to list its pipelines, so the request is a no-op and leaves the existing streams alone. ### Device Logs and Telemetry #### Logs Returns paginated device logs. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/logs" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} **Rate limit:** 120 requests per minute per workspace. Log reads and log ingest use separate rate-limit buckets, so polling for logs will not count against your ingest quota (and vice versa). #### Telemetry Returns aggregated hardware metrics (CPU, memory, disk, GPU) and per-stream FPS for a device, bucketed over a fixed time window. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/telemetry" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} **Rate limits** This endpoint is rate-limited per device to **60 requests per minute** with an additional **10 requests per 10 seconds** burst limiter. Excess requests return **429**. The same per-device telemetry quota is shared with the device's telemetry ingest path, so abusive reads can affect ingest for the same device. ### Device Events Returns device and stream lifecycle events (for example device boots, stream starts and stops, errors, config changes). **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/events" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} By default the feed collapses repeated events into state changes, which hides periodic telemetry that reuses one event name. Pass `dedupe=false` with both `entity_type` and `entity_id` to read every raw row. To chart one value over time, use the [service series endpoint](#device-services) instead: it resolves the entity ID for you. ### Device Services Returns the [services](/deployment/self-hosted/enterprise/deployment-manager/services.md) a device runs, joined to their live status: container state, image version, and per-kind status such as PLC connection state or Event Store capacity. **Required scope:** `device:read` {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/services" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/services/{serviceName}" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} Use these endpoints to poll a fleet: ask for `include=status` to check that a service is up and reporting, and add `include=metrics` for its CPU and memory. {% hint style="info" %} `include=config` returns environment variable values as they are stored, and those often hold credentials such as broker passwords and API keys. It is off by default for that reason. Treat any response that includes `config` as sensitive. {% endhint %} Services report telemetry under an entity ID that comes from the service kind. Two services of the same kind on one device therefore collide unless you give each its own `DEVICE_EVENT_ENTITY_ID`. When that happens, the API lists the shared IDs in `signal_conflicts` and returns `status` as `null`, because it cannot tell which container sent the reading. A custom entity ID must use only letters, digits, hyphens, and underscores. The characters `.`, `/`, `*`, `~`, `[`, and `]` break the status cache, and the service then reports no status at all. #### Service Metric Series Returns one metric for one service over time, for charting. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/services/{serviceName}/series" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} ### Device Groups Group devices to find and manage them together (ex: every camera on one line). {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/groups" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/groups" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/groups/{groupId}" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/groups/{groupId}" method="patch" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/groups/{groupId}" method="delete" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} To add or remove a few devices, use these instead of sending the full member list to Update a Device Group. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/groups/{groupId}/devices" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/groups/{groupId}/devices" method="delete" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} ### Device Management These endpoints do from the API what the device page in the web app does. Delete and redeploy need a workspace API key, so a device cannot delete or reinstall itself with its own key. #### Update a Device Rename a device or set its environment variables. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}" method="patch" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Delete a Device {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}" method="delete" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Redeploy a Device Use the new `installId` to reinstall Deployment Manager on the device, the same way as after [Create a Device](#create-a-device). {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/redeploy" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Apply Pending Config On a device with [maintenance windows](/deployment/self-hosted/enterprise/deployment-manager/setting-up/setup-maintenance-windows.md), config changes wait for the next window. Call this to apply them now. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/config/apply-pending" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Capture a Frame to a Project Add a stream's latest frame to a project, to collect training images from a deployed camera. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/streams/{streamId}/capture" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Sample a Live Frame Get a recent frame from a camera as an image URL (ex: to check what a camera sees, or to pass the image to a Workflow). If the stream has a frame less than 10 seconds old, you get it at once. If not, Roboflow asks the device for a new frame and waits up to `timeout_seconds`. The URL points to a copy of the frame, so it stays valid for an hour. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/streams/{streamId}/sample" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Snooze Alerts {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/{deviceId}/alerts/snooze" method="post" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} #### Fleet-Wide Reads These read data across all devices in the workspace, so they need a workspace API key. {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/events" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/inference-versions" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} {% openapi src="/files/H5x9CDQa7dPNDXC9dDQe" path="/{workspace}/devices/v2/stream-options" method="get" %} [deployment-manager.yaml](https://1583372177-files.gitbook.io/~/files/v0/b/gitbook-x-prod.appspot.com/o/spaces%2FoksNmwps1HmYr8TqHDnX%2Fuploads%2Fgit-blob-34b7351d1ba187316930d80b9839ae0e19531a8c%2Fdeployment-manager.yaml?alt=media) {% endopenapi %} ## MCP Server Connect an AI agent (ex: Claude, Cursor, ChatGPT, Codex) to the [Roboflow MCP Server](https://docs.roboflow.com/agents/mcp-server) and manage your devices by asking in plain language. The agent uses the same API as the web app, with your account's permissions. {% hint style="info" %} [Set up the Roboflow MCP Server](https://docs.roboflow.com/agents/mcp-server#adding-mcp) in a few minutes, then ask your agent about your devices. {% endhint %} With the MCP Server connected, your agent can: * Check on your fleet: find offline devices, read logs and telemetry, and explain recent errors. * Look through a camera: grab a live frame, see the scene, and test a Workflow on it before you deploy. * Change a device: update its config, pin an inference version, or rename it, with maintenance windows respected. * Run streams: add, pause, resume, trigger, or remove a stream, and power-cycle a stuck camera's PoE port. * Organize devices: create device groups and move devices between them. Try asking: * "Which of my devices are offline, and why?" * "Show me what the loading dock camera sees right now." * "Set inference to version 1.2.3 on every device in the Line 3 group." See every device tool in the [MCP Server tool list](https://docs.roboflow.com/agents/mcp-server#devices-and-streams).