What can I help you with?
NMX Telemetry (NMX-T) Documentation v1.0.0

Installation Configuration

Description

x86_64 Linux host

  • RAM 4GB

  • Disk space 10GB

NVL5/IB fabrics enabled

  • ibstat shows at least one up and active IB interface

Tools

  • Docker 20.10.5+

  • Python 3.8+

OS integration parameters are controlled by the integration.json file, a part of the cluster application package. NMX-T package comes with predefined integration parameters aligned with the NVOS distribution it is a part of.

In case of running NMX-T in a standalone mode it may be practically useful to modify the integration control file to change the service port assignments and location of the TLS/mTLS certificate files.

Key

Description

Default

control_port

TCP port number the control http interface binds to

9350

manager_port

TCP port number for the gRPC interface of NMX Telemetry connector

9351

prometheus_exporter_port

TCP port number for the Prometheus metrics endpoint

9352

syslog_receiver_port

UDP port number for the syslog receiver to be published

9353

certificate_path

Path to the certificate directory to be mounted inside the containers

/etc/nmx-telemetry

log_direcotry_path

Path of the local host directory to keep application logs

/var/log/nmx/nmx-t

nmx_fib_port

Port number of the FIB configuration storage service running on the host

9372

syslog

URI defining syslog destination to write application logs

""

NMX Telemetry integration parameters include port number assigned to the service components:

Copy
Copied!
            

{     "manager_port"9351,     "prometheus_exporter_port"9352,     "syslog_receiver_port"9353,     "control_port"9350,     "certificate_path""/etc/nmx-telemetry",     "log_direcory_path""/var/log/nmx/nmx-t",     "nmx_fib_port"9372,     "syslog""udp://240.127.1.1:514" }

NMX Telemetry uses the listed parameters to bind services to the ports, while NVOS is expected to expose those ports to the public. The configuration file is read when the application starts, and the settings remain unchanged throughout the application's lifecycle.

When running NMX-Telemetry as a Docker container, bypassing the application lifecycle management hooks, the environment integration parameters are passed to the Telemetry instance through the container's environment.

For example:

Copy
Copied!
            

docker run --privileged --network=host --rm --detach \ --env NMX_SYSLOG_RECEIVER=udp://240.127.1.1:514 --name nmx-telemetry nmx-telemetry:0.9.5

Environment variables matching the integration.json parameters:

Environment variable

Description

Format

NMX_CONTROL_ADDRESS

IP address and port the control http interface binds to

":${control_port}"

NMX_CONNECTOR_ADDRESS

IP address and port for the gRPC interface of NMX Telemetry connector

":${manager_port}"

NMX_METRICS_ADDRESS

IP address and port for the Prometheus metrics endpoint

":${prometheus_exporter_port}"

NMX_SYSLOG_RECEIVER

IP address and portfor the syslog receiver to be published

":${syslog_receiver_port}"

NMX_CONFIGURATOR_FIB_ADDRESS

IP address and port of the FIB configuration storage service running on the host

":${nmx_fib_port}"

NMX_LOGS_FORWARDING

URI defining syslog destination to write application logs

"${var.syslog}"

© Copyright 2025, NVIDIA. Last updated on Apr 29, 2025.