Configuration files and certificates

app_config.json

The configuration file ① can be specified during installation of the AI Inference Servers or with the Update configuration function of Industrial Edge Management. (Update configuration applies only to pipelines that are not being executed. The values of the executed pipeline cannot be changed.)

Fields:

• stopPipelineAtDataLoss: Configures whether the currently executed pipeline is to be stopped directly after the first deleted data point. This applies both to orchestratorInternalQueueSize as and to runtimeInternalQueueSize.

• orchestratorInternalQueueSize: Configures the internal queue size of the internal component named Orchestrator, which forwards the data points.

• runtimeInternalQueueSize: Configures the size of the internal queues of the internal runtime components. They are used between the pipeline steps for the data communication.

config.json

This file ② is necessary so that the functions connected to the AI Model Manager work correctly.

Fields:

• embedderApi: The URL under which an internal component can reach the API of the AI Inference Server. The endpoints for the status request and the deployment manipulation are called with this base URL.

• sharedVolume: The absolute path at which the shared volume is attached within the internal component. This component creates subdirectories named after the deployment IDs here and extracts the packages in these directories. The AI Inference Server can then access every deployment because the data storage medium is shared.

• statusUpdateInterval: The interval at which the component asks the AI Inference Server for the status of the deployments. Only time values between "1 s" (one second) and "1h" (one hour) are accepted. For example:

  • "1m" → 1 minute

  • "10s" → 10 seconds

  • "5m3s" → 5 minutes and 3 seconds

• sqliteDsn: The absolute path to the database file and some database connection properties. This path must point to somewhere in the folder /ai-ml-manager-agent/database. We recommend retaining it as in the above example and referencing a file called "agent.db" directly under the node agent-db-volume. The mode connection property must be set to "rwc" so that the application can read/write/create the database. Other properties are recommendations for proper functioning and performance.

Here, you will find the correct configuration content of the "config.json" file:

action.json

This file ③ is updated automatically by the AI Model Manager and contains relevant information on deployment. No change has to be made, the AI Inference Server only needs to have this file in order to be available.

aimm.crt

This file ⑥ contains the public key so that an internal component can verify the authenticity of the SSL certificate of AI Model Manager.

The file was loaded once on system start of the Inference Server and is completely optional. If you want to upload another certificate into the app, click + and select the file with the uploaded certificate.

Keywords used by AI Inference Server

The following keyword cannot be a variable name in a Python script:

timestamp

Activating certificates

After fine tuning of the configuration files is complete, the required configuration files must be selected in order to be included in the AI Inference Server installation. Select the certificates by selecting the corresponding check box ⑤. To complete the installation, click "Next" ⑦ and select the location where you want to install the app.