Next
Filter data
Capture and sync edge data
You can use the data management service to capture data from supported components and services, then sync it to the cloud. You can also sync data from arbitrary folders on your machine.
Data capture and sync in Viam involves two key pieces:
- The data management service that writes captured data to local edge device storage and syncs that data with the cloud.
- Individual resource configurations that specify what data to capture and how often.
How data capture and data sync works
The data management service writes data from your configured Viam resources to local storage on your edge device and syncs data from the edge device to the cloud:
- The data management service stores captured data locally in
~/.viam/capture by default. - Data is synced to the Viam cloud at a configured sync interval using encrypted gRPC calls and deleted from the disk once synced.
- You can capture and sync data independently, one can run without the other.
For more information, see How sync works.
Configure the data management service for your machine
From your machine’s CONFIGURE tab in the Viam app, add the data management service.
On the panel that appears, configure data capture and sync attributes as applicable.
To both capture data and sync it to the cloud, keep both Capturing and Syncing switched on.
Click the Save button in the top right corner of the page to save your config.
{
"components": [],
"services": [
{
"name": "my-data-manager",
"namespace": "rdk",
"type": "data_manager",
"attributes": {
"sync_interval_mins": 1,
"capture_dir": "",
"tags": [],
"capture_disabled": false,
"sync_disabled": true,
"delete_data_on_part_deletion": true,
"delete_every_nth_when_disk_full": 5,
"maximum_num_sync_threads": 250
}
}
]
}
{
"components": [],
"services": [
{
"name": "my-data-manager",
"namespace": "rdk",
"type": "data_manager",
"attributes": {
"capture_dir": "",
"tags": [],
"additional_sync_paths": [],
"sync_interval_mins": 3
}
}
]
}
The following attributes are available for the data management service:
Configure data capture for individual resources
You can capture data for any resource that supports it, including resources on remote parts.
Configure data capture for individual resources in their configuration by:
- Selecting which resource methods to capture data from
- Setting the capture frequency for each method
For each resource you can capture data for, there is a Data capture section in its panel.
Select a Method and specify a capture Frequency in hertz, for example to 0.1 to capture an image every 10 seconds.
You can add multiple methods with different capture frequencies.
Some methods will prompt you to add additional parameters.
The available methods, and corresponding additional parameters, will depend on the component or service type.
For example, a camera has the options ReadImage and NextPointCloud.
Keep in mind that some models do not support all options, for example webcams do not capture point clouds, and choose the method accordingly.
Caution
Avoid configuring data capture to higher rates than your hardware can handle, as this leads to performance degradation.
This example configuration captures data from the ReadImage method of a camera:
{
"services": [
...
,
{
"name": "data_manager",
"type": "data_manager",
"attributes": {
"sync_interval_mins": 5,
"capture_dir": "",
"sync_disabled": false,
"tags": []
}
}
],
"remotes": [
{
...
}
],
"components": [
...
,
{
"service_configs": [
{
"type": "data_manager",
"attributes": {
"capture_methods": [
{
"capture_frequency_hz": 0.333,
"disabled": false,
"method": "ReadImage",
"additional_params": {
"reader_name": "cam1",
"mime_type": "image/jpeg"
}
}
],
"retention_policy": {
"days": 5
}
}
}
],
"model": "webcam",
"name": "cam",
"type": "camera",
"attributes": {
"video_path": "video0"
},
"depends_on": [
"local"
]
},
...
]
}
This example configuration captures data from the GetReadings method of a temperature sensor and wifi signal sensor:
{
"services": [
{
"attributes": {
"capture_dir": "",
"tags": [],
"additional_sync_paths": [],
"sync_interval_mins": 3
},
"name": "dm",
"namespace": "rdk",
"type": "data_manager"
}
],
"components": [
{
"type": "sensor",
"model": "tmp36",
"attributes": {
"analog_reader": "temp",
"num_readings": 15
},
"depends_on": [],
"service_configs": [
{
"attributes": {
"capture_methods": [
{
"capture_frequency_hz": 0.2,
"cache_size_kb": 10,
"additional_params": {},
"method": "Readings"
}
]
},
"type": "data_manager"
}
],
"name": "tmp36",
"namespace": "rdk"
},
{
"type": "sensor",
"model": "wifi-rssi",
"attributes": {},
"service_configs": [
{
"type": "data_manager",
"attributes": {
"capture_methods": [
{
"additional_params": {},
"method": "Readings",
"capture_frequency_hz": 0.1,
"cache_size_kb": 10
}
]
}
}
],
"name": "my-wifi-sensor",
"namespace": "rdk"
}
]
}
Example for a vision service:
This example configuration captures data from the CaptureAllFromCamera method of the vision service:
{
"components": [
{
"name": "camera-1",
"namespace": "rdk",
"type": "camera",
"model": "webcam",
"attributes": {}
}
],
"services": [
{
"name": "vision-1",
"namespace": "rdk",
"type": "vision",
"model": "mlmodel",
"attributes": {},
"service_configs": [
{
"type": "data_manager",
"attributes": {
"capture_methods": [
{
"method": "CaptureAllFromCamera",
"capture_frequency_hz": 1,
"additional_params": {
"mime_type": "image/jpeg",
"camera_name": "camera-1",
"min_confidence_score": "0.7"
}
}
]
}
}
]
},
{
"name": "data_manager-1",
"namespace": "rdk",
"type": "data_manager",
"attributes": {
"sync_interval_mins": 0.1,
"capture_dir": "",
"tags": [],
"additional_sync_paths": []
}
},
{
"name": "mlmodel-1",
"namespace": "rdk",
"type": "mlmodel",
"model": "viam:mlmodel-tflite:tflite_cpu",
"attributes": {}
}
],
"modules": [
{
"type": "registry",
"name": "viam_tflite_cpu",
"module_id": "viam:tflite_cpu",
"version": "0.0.3"
}
]
}
Viam supports data capture from resources on remote parts.
For example, if you use a part that does not have a Linux operating system or does not have enough storage or processing power to run viam-server, you can still process and capture the data from that part’s resources by adding it as a remote part.
Currently, you can only configure data capture from remote resources in your JSON configuration.
To add them to your JSON configuration you must explicitly add the remote resource’s type, model, name, and additional_params to the data_manager service configuration in the remotes configuration:
| Key | Description |
|---|---|
type | The type tells your machine what the resource is. For example, a board. |
model | The model is a colon-delimited-triplet that specifies the namespace, the type of the part, and the part itself. |
name | The name specifies the fully qualified name of the part. |
additional_params | The additional parameters specify the data sources when you are using a board. |
The following attributes are available for data capture configuration:
Click the Save button in the top right corner of the page to save your config.
If cloud sync is enabled, the data management service deletes captured data once it has successfully synced to the cloud.
Warning
If your robot is offline and can’t sync and your machine’s disk fills up beyond a certain threshold, the data management service will delete captured data to free up additional space and maintain a working machine.
Supported resources
The following components and services support data capture and cloud sync:
| Type | Method |
|---|---|
| Arm | EndPosition, JointPositions |
| Board | Analogs, Gpios |
| Camera | GetImages, ReadImage, NextPointCloud |
| Encoder | TicksCount |
| Gantry | Lengths, Position |
| Motor | Position, IsPowered |
| Movement sensor | AngularVelocity, CompassHeading, LinearAcceleration, LinearVelocity, Orientation, Position |
| Sensor | Readings |
| Servo | Position |
| Vision service | CaptureAllFromCamera |
View captured data
To view all the captured data you have access to, go to the DATA tab where you can filter by location, type of data, and more.
You can also access data from a resource or machine part menu.
Stop data capture and data sync
If you don’t need to capture data, for instance in a test scenario, you can turn off data capture to reduce unnecessary storage. Alternatively, see advanced data capture and sync configurations for other ways to control data usage, such as conditional sync or retention policies.
To turn off data capture for a specific resource’s capture method (for example, a camera component capturing through the GetImage capture method) navigate to the Data capture section of your resource’s configuration card and toggle the configured capture method’s switch to Off.
You can also globally turn off data capture on the data_manager service configuration card by toggling the Capturing switch to Off.
To turn off data sync, navigate to the data_manager service configuration card and toggle the Syncing switch to Off.
Click the Save button in the top right corner of the page to save your config.
Advanced data capture and sync configurations
Capture directly to MongoDB
You can configure direct capture of tabular data to a MongoDB instance alongside disk storage on your edge device. This can be useful for powering real-time dashboards before data is synced from the edge to the cloud.
Configure using the mongo_capture_config attributes in your data manager service.
Here is a sample configuration that will capture fake sensor readings both to the configured MongoDB URI as well as to the ~/.viam/capture directory on disk:
When mongo_capture_config.uri is configured, data capture will attempt to connect to the configured MongoDB server and write captured tabular data to the configured mongo_capture_config.database and mongo_capture_config.collection (or their defaults if unconfigured) after enqueuing that data to be written to disk.
If writes to MongoDB fail for any reason, data capture will log an error for each failed write and continue capturing.
Failing to write to MongoDB doesn’t affect capturing and syncing data to cloud storage other than adding capture latency.
Caution
- Capturing directly to MongoDB may write data to MongoDB that later fails to be written to disk (and therefore never gets synced to cloud storage).
- Capturing directly to MongoDB does not retry failed writes to MongoDB. As a consequence, it is NOT guaranteed all data captured will be written to MongoDB.
This can happen in cases such as MongoDB being inaccessible to
viam-serveror writes timing out. - Capturing directly to MongoDB may reduce the maximum frequency that data capture can capture data due to the added latency of writing to MongoDB. If your use case needs to support very high capture rates, this feature may not be appropriate.
Conditional sync
By default, viam-server checks for new data to sync at the configured interval (sync_interval_mins).
You can additionally configure sync to only happen when certain conditions are met.
For example:
- Only sync when on WiFi
- Sync when specific events are detected
- Sync during certain time windows
See Conditional cloud sync for how to implement conditional syncs.
Cloud data retention
Configure how long your synced data remains stored in the cloud:
- Retain data up to a certain size (for example, 100GB) or for a specific length of time (for example, 14 days): Set
retention_policiesat the resource level. See theretention_policyfield in data capture configuration attributes. - Delete data captured by a machine when you delete the machine: Control whether your cloud data is deleted when a machine or machine part is removed.
See the
delete_data_on_part_deletionfield in the data management service configuration attributes.
Sync optimization
Configurable sync threads: You can control how many concurrent sync operations occur by adjusting the maximum_num_sync_threads setting.
Higher values may improve throughput on more powerful hardware, but raising it too high may introduce instability on resource-constrained devices.
Wait time before syncing arbitrary files: If you choose to sync arbitrary files (beyond those captured by the data management service), the file_last_modified_millis configuration attribute specifies how long a file must remain unmodified before the data manager considers it for syncing.
The default is 10 seconds.
Was this page helpful?
Glad to hear it! If you have any other feedback please let us know:
We're sorry about that. To help us improve, please tell us what we can do better:
Thank you!