Skip to content

Latest commit

 

History

History
79 lines (50 loc) · 4.69 KB

README.md

File metadata and controls

79 lines (50 loc) · 4.69 KB

UFM stream To FluentD endpoint

This plugin is used to extract topology via UFM API and stream it via fluentd protocol to telemetry console

Overview

The UFM Enterprise product is Nvidia’s platform for IB fabric management. Through this platform, the various devices (switches, multi-chip systems, cables, etc.) are discovered, configured and the status of the entire fabric is reflected. As a fabric manager, the UFM Enterprise holds an internal representation model of the network topology. This topology should be reflected, over time (as it can change with time) towards telemetry console. In order to do so, we present stream To FluentD plugin

Prerequisites

To install all prerequisites, please run :

pip install -r requirements.txt

Fluentd configurations

  • We provide fluentd.conf as a fluentd configurations sample.

Run

1) using command line arguments :

python3 fluentd_topology_plugin.py --fluentd_host=fluentd_host --fluentd_port=fluentd_port --ufm_host=ufm_host --ufm_username=ufm_username --ufm_protocol=https --ufm_password=ufm_password

2) using configuration file:

  • copy config file sample fluentd_topology_plugin.sample.cfg to fluentd_topology_plugin.cfg

    cp fluentd_topology_plugin.sample.cfg fluentd_topology_plugin.cfg

  • Edit config file with relevant parameters

    vi fluentd_topology_plugin.cfg

  • Run

    python3 fluentd_topology_plugin.py

Running syntax

Argument Corresponding Config Value Required Description
--fluentd_host <fluentd_host> fluentd-config.host True Hostname or IP for FluentD endpoint
--fluentd_port <fluentd_port> fluentd-config.port True Port for FluentD endpoint
--fluentd_timeout <fluentd_timeout> fluentd-config.timeout True Timeout for FluentD endpoint streaming [Default is 120 seconds]
--fluentd_message_tag_name <fluentd_message_tag_name> fluentd-config.message_tag_name False Message Tag Name for FluentD endpoint message [Default is the ufm_host]
--ufm_host <ufm_host> ufm-remote-server-config.host True Hostname or IP for The UFM Enterprise
--ufm_protocol <ufm_protocol> ufm-server-config.ws_protocol True Web services protocol used by UFM Enterprise (HTTP, HTTPS)
--ufm_username <ufm_username> ufm-server-config.username True Username of UFM user
--ufm_password <ufm_password> ufm-server-config.password True Password of UFM user
--logs_file_name <logs_file_name> logs-config.logs_file_name False Log file name, if not provided a default stream wil lbe used
--logs_level <logs_level> logs-config.logs_level False Default is 'info'
--local_streaming <local_streaming> streaming-config.local_streaming False Enable/Disable local topology streaming [Default is 'False']
--streaming streaming-config.streaming False Enable/Disable topology streaming [Default is 'True']
--streaming_interval <streaming_interval> streaming-config.interval False The periodic interval [the script will be ran in every X minutes (Default is 5 minutes)]
--streaming_systems <streaming_systems> streaming-config.systems False Default is 'True'
--streaming_ports <streaming_ports> streaming-config.ports False Default is 'True'
--streaming_links <streaming_links> streaming-config.links False Default is 'True'
--streaming_alarms <streaming_alarms> streaming-config.alarms False Default is 'True'

*If command line argument is provided, the corresponding config value will be ignored

Use

This application is not a daemon; you should run it via time-based job scheduler (cron job).