Use automatic deployments while preserving device or module specific settings

Question

Use automatic deployments while preserving device or module specific settings

Kim Streich 0

Hi,

I am trying to figure out the best way to use automatic deployments with Azure IoT Edge.

The automatic deployments work great. I can target all my devices, but for each device I need to set some device-specific settings.

For example, I am using the local blob storage module with the cloud upload function. I need to set a different connection string for each device to make sure devices can't read each other's uploads.

I created a deployment with some default settings that targets all devices. Then I created another module twin deployment that overrides the connection string and upload bucket that targets a single device and module.

This is working as I want it, but it means I will need to create one config deployment per device.

But according to this document there is a limit of 100 configurations per basic or standard SKU hub for Automatic device and module configurations.

This means I won't be able to configure more than 99 devices with this approach.

How can I solve this?

Is there another way for me to set device-specific settings and ensure they don't get overwritten by the automatic deployments?

Thanks, Kim

Manas Mohanty 8,150 Reputation points Microsoft External Staff Moderator

2025-03-21T12:19:08.93+00:00

Hi Kim Streich

Actually, we can modify existing deployment multiple times or use multiple deployment to target devices with unique target conditions based on device twin or properties.

But we have to set a higher priority number then base deployment prior to render the effects. Please refer target condition section from here. https://learn.microsoft.com/en-us/azure/iot-edge/how-to-deploy-at-scale#step-4-target-devices

You can increase max deployment quotas or throttle limits by increasing the number of provisioned units in an IoT hub.

Hope it helps.

Thank you.
Kim Streich 0 Reputation points

2025-03-21T15:04:06.8866667+00:00
Hi Manas. Thank you so much for the super quick response.

Yes, I am already doing that. I have...

A base deployment that sets some default settings and is applied to all devices

One Module Twin Configuration per device that overrides the default values

This is working well, but the issue is that you can only have 100 deployments at a time.

When I delete the module twin configuration, it deletes the customizations from the device.

The end-effect is that this method would only allow me to manage at most 98 devices in one hub.

Is there a way to configure device specific desired properties without a layered deployment that uses one of the 100 deployment slots?

1 answer

Your answer

Manas Mohanty 8,150 Reputation points Microsoft External Staff Moderator

2025-03-21T12:19:08.93+00:00

Hi Kim Streich

Actually, we can modify existing deployment multiple times or use multiple deployment to target devices with unique target conditions based on device twin or properties.

But we have to set a higher priority number then base deployment prior to render the effects. Please refer target condition section from here. https://learn.microsoft.com/en-us/azure/iot-edge/how-to-deploy-at-scale#step-4-target-devices

You can increase max deployment quotas or throttle limits by increasing the number of provisioned units in an IoT hub.

Hope it helps.

Thank you.
Kim Streich 0 Reputation points

2025-03-21T15:04:06.8866667+00:00

Hi Manas. Thank you so much for the super quick response.

Yes, I am already doing that. I have...

A base deployment that sets some default settings and is applied to all devices

One Module Twin Configuration per device that overrides the default values

This is working well, but the issue is that you can only have 100 deployments at a time.

When I delete the module twin configuration, it deletes the customizations from the device.

The end-effect is that this method would only allow me to manage at most 98 devices in one hub.

Is there a way to configure device specific desired properties without a layered deployment that uses one of the 100 deployment slots?

Answer 1

Sander van de Velde | MVP 36,941 MVP Volunteer Moderator

Hello @Kim Streich ,

welcome to this moderated Azure community forum.

We achieved this using the specific module twin configuration like:

"properties": {
  "desired": {
    "url": 'http://realurl.com',
    "SendInterval": 5,
    "layeredProperties": {
      "url": 'http://dummy.com',
      "SendInterval": 60,
    }
  }
}

So the 'layeredProperties' are deployed via the layered deployments. If a module needs customization, we add the same property next to it.

In the code of the module that handles twin changes, we test for both properties where the layered-properties are overruled by non-layered properties.

So, the layered deployments only deploy generic values which can be overruled by custom properties.

If the response helped, do "Accept Answer". If it doesn't work, please let us know the progress. All community members with similar issues will benefit by doing so. Your contribution is highly appreciated.

Kim Streich 0 Reputation points

2025-03-21T15:07:57.1066667+00:00

Hi @Sander van de Velde | MVP . Thank you so much for the super quick response.

Yes, I am already doing that. I have...

A base deployment that sets some default settings and is applied to all devices

One Module Twin Configuration per device that overrides the default values

This is working well, but the issue is that you can only have 100 deployments at a time.

When I delete the module twin configuration, it deletes the customizations from the device.

The end-effect is that this method would only allow me to manage at most 98 devices in one hub.

Is there a way to configure device specific desired properties without a layered deployment that uses one of the 100 deployment slots?

Or can I somehow set the desired properties on a module twin without using layered deployment and make sure that the automatic deployment doesn't overwrite them?
Sander van de Velde | MVP 36,941 Reputation points MVP Volunteer Moderator

2025-03-21T16:49:12.6+00:00

Hello @Kim Streich ,

you need to think in 'feature' layers instead of device layers.

You are probably deploying logic on edge devices what are somewhat repetitive.

So you probably ingest data from protocols, you transform data, you put it in a local database, etc.

These are features. So tag your edge device as having one or more features on top of the base deployment like 'hasFeatureA = true'.

Each feature has it's own set of modules which are only deployed to devices supporting that feature in the tags.

Once set up, you can duplicate feature layers if there is an update with an higher priority. Once that updated feature is deployed to all related devices, you can remove that old layer.

This keeps the number of layers in use quite low.

As an alternative, creating a second IoT Hub is a simple way to double the number of devices. Then, use static assignment in the Device Provisioning Service to redirect devices to the right IoT Hub.
Kim Streich 0 Reputation points

2025-03-21T19:36:45.7066667+00:00

Thank you so much, @Sander van de Velde .

The reason I am thinking in terms of devices is because I need to configure the local blob storage to upload to cloud blob storage.

One of the requirements is that each edge device shouldn't be able to see the files of the other devices. So, I need to configure different access credentials on each of them.

I have now come up with a way of doing that.

I removed the properties.desired.deviceToCloudUploadProperties from the automatic deployment and I am setting it manually using the azure cli with az iot hub module-twin update.

When I update the deployment, it doesn't overwrite the desired properties since they are missing from the deployment manifest.

I am planning on automating this step with ansible during device provisioning.

Does doing it this way have any downsides that I am missing? Or are there any other best practices for doing device-specific configuration?
Sander van de Velde | MVP 36,941 Reputation points MVP Volunteer Moderator

2025-03-21T21:06:54.9933333+00:00

Hello @Kim Streich ,

this is basically the approach with the 'splitted desired properties'.

You can automate this using AZ CLI or with the Server SDK.

Because you need IoT hub level access the secret 'root' key should probably not leave the save Azure environment so that I would prefer using an Azure Function.

If the response helped, do "Accept Answer". If it doesn't work, please let us know the progress. All community members with similar issues will benefit by doing so. Your contribution is highly appreciated.
Kim Streich 0 Reputation points

2025-03-24T16:54:38.46+00:00

Thank you very much for all your help @Sander van de Velde | MVP .
Sander van de Velde | MVP 36,941 Reputation points MVP Volunteer Moderator

2025-03-25T20:08:34.8233333+00:00

You're welcome, @Kim Streich.

Please accept the answer so others with the same questions are made aware of it. Thanks.

Share via

Use automatic deployments while preserving device or module specific settings

1 answer

Your answer