Application Lifecycle

Prerequisites

NMX-T (Telemetry) requires NMX-C (Controller) to be up and running within the same network domain. 

Running NMX-T using NVOS

To start and stop the application, perform the following:

1. Make sure the cluster is enabled:

   nv show cluster 
   

2. If the cluster is disabled, enable it by running: 

   nv set cluster state enabled
   nv config apply

3. Start the telemetry application:

   nv action start cluster app nmx-telemetry
   

4. Verify the NMX-T is running among the other cluster applications.
   

   nv show cluster apps running
   

5. Enable the NMX-M registration:

   nv action update cluster apps nmx-telemetry manager enabled

To stop the application, run:

nv action stop cluster app nmx-telemetry

Call Application Interfaces

• Health check callback - expected to see "OK" in the output

  curl -s http://localhost:9350/healthcheck
  

• Application package information - list of components with their versions

  curl -s http://localhost:9350/info/package | jq
  

• Configuration callback - expected to see security configuration details

  curl -s http://localhost:9350/config | jq
  

• Prometheus metrics - expected to see metrics data

  curl -s http://localhost:9352/xcset/nvlink_domain_telemetry
  

• Custom gRPC - expected to see metrics data
  The form of the call depends on the installation method. 

  ./test/nmx_client.sh --target localhost:9351 subscribe
  

Running NMX-T in a Standalone Mode (non-NVOS Users)

When running NMX-T in a standalone mode, i.e. not integrated into the NVOS bundle, the application lifecycle is controlled using the helper scripts distributed as a part of the package.

All the integration scripts are executed from the package “untarred” folder, making it an implicit script argument.

• install.py: installs an application from the uncompressed package folder. Accepts path to the application registry as a first positional command line argument.

• start.py: starts the application running. The script needs to receive as input the argument full path to the folder with the user_config.json file.

• stop.py: stops the application. The script receives as input argument the full path to the application registry

• uninstall.py: removes the installed components of the application, including the untarred package. Accepts path to application registry as a first positional command line argument.

• support.py: a maintenance script. The script receives as input argument the full path to the application registry.

Cluster Application Registry

The cluster application registry is a JSON file keeping information of the cluster applications installed. While the path to the folder containing the application registry may vary, the name of the file remains constant installed_apps.json 

{
    "nmx-telemetry": {
        "configCallback": "http://0.0.0.0:9350/config",
        "healthcheck": "http://0.0.0.0:9350/healthcheck",
        "packageFilesFolder": "/etc/nmx/nmx-telemetry"
    }
}

Standalone Application Lifecycle Example

1. Install the application. 

   python3 ./install.py $(pwd)/..
   

2. Start the application.

   python3 ./start.py $(pwd)/..
   

3. Check the docker containers are up and running - nmx-t container should be present and running healthy.

   docker ps | grep nmx-t
   

4. Stop the application.

   python3 ./stop.py $(pwd)/..
   

5. Uninstall the cluster application.

   python3 ./uninstall.py $(pwd)/..
   

Lifecycle Management Script Configuration

The application lifecycle management scripts could be individually configured at runtime by setting the following environment variables: 

For example, run uninstall script, showing detailed logs to the standard output and keeping the package file:

NMX_LOGGER=stdout NMX_REMOVE_FILES=false NMX_DEBUG=true python3 ./stop.py $(pwd)

Last updated: July 24, 2025