Using Pipeline: TCP to textfile

4 minutes read

Using Pipeline

In an earlier blog, I introduced Pipeline, a multi-function telemetry collection service written in Go. In this tutorial, I’ll cover how to set up Pipeline for the simplest of tasks: ingesting telemetry data over TCP and writing it to a file as a JSON object.

Preparing the Router

This tutorial assumes that you’ve already configured your router for model-driven telemetry (MDT) with TCP dial-out using the instructions in this tutorial. The IP address and port that you specify in the destination-group in the router config should match the IP address and port on which Pipeline is listening.

Getting Pipeline

Pipeline is available from github.

Pipeline.conf

The pipeline.conf file contains all the configuration necessary to get Pipeline running. In many cases, the default pipeline.conf can be used with little or no modification.

The pipeline configuration is divided up into sections. Each section is delineated by an arbitrary name enclosed in square brackets. Each section defines either an input stage (“stage = xport_input”) or an output stage (“stage = export_output”). Other parameters in the section tell Pipeline what to listen for (in the case of an input stage) or how to output the data (for an output stage).

The easiest way to understand this is to look at a simple example.

Configuring the Input Stage for TCP Dial-Out

Let’s take a look at the TCP dial-out section in the default pipeline.conf.

scadora@darcy:~/bigmuddy-network-telemetry-pipeline$ grep -A20 "Example of a TCP dialout" pipeline.conf
# Example of a TCP dialout (router connects to pipeline over TCP)
#
[testbed]
stage = xport_input
#
# Module type, the type dictates the rest of the options in the section.
# TCP can only be used as xport_iinput currently. UDP works similarly.
#
type = tcp
#
# Supported encapsulation is 'st' for streaming telemetry header. This
# is the header used to carry streaming telemetry payload in TCP and UDP.
#
encap = st
#
# TCP option dictating which binding to listen on. Can be a host name
# or address and port, or just port.
#
listen = :5432

This [testbed] section shown above will work “as is” for MDT with TCP dial-out. If you want to change the port that Pipeline listens on to something other than “5432”, you can edit this section of the pipeline.conf. Otherwise, we’re good to go for the input stage.

Configuring the Output Stage for Text File

To dump the received data to a file, we need a “tap” stage in Pipeline. The default pipeline.conf file comes with a tap stage section called [inspector] as you can see below.

scadora@darcy:~/bigmuddy-network-telemetry-pipeline$ grep -A20 "Example of a tap stage" pipeline.conf
# Example of a tap stage; dump content to file, or at least count messages
#
[inspector]
stage = xport_output
#
# Module type: tap is only supported in xport_output stage currently.
#
type = tap
#
# File to dump decoded messages
#
file = /data/dump.txt
#
# encoding = json
#

This [inspector] section shown above will work “as is” for dumping data to a file. If you want to change the file that Pipeline writes to (default is /data/dump.txt) or write with a different encoding (default is JSON), you can edit this section of the pipeline.conf. Otherwise, we’re good to go for the output stage as well.

Running Pipeline

Running pipeline is trivial. Just execute the binary in the bin directory. Pipeline will use the pipeline.conf file by default.

scadora@darcy:~/bigmuddy-network-telemetry-pipeline$ bin/pipeline &
[1] 21975
scadora@darcy:~/bigmuddy-network-telemetry-pipeline$ Startup pipeline
Load config from [pipeline.conf], logging in [pipeline.log]
Wait for ^C to shutdown

scadora@darcy:~/bigmuddy-network-telemetry-pipeline$

Seeing the Data

Assuming your router is properly configured, the router should initiate the TCP session to Pipeline and stream the data specified in the sensor-group configuration. To see the data as it comes in, use the Linux “tail” utility on the file that the [inspector] stage was configured to write to.

scadora@darcy:~/bigmuddy-network-telemetry-pipeline$ tail -f dump.txt

------- 2017-04-03 20:37:06.763244782 -0700 PDT -------
Summary: GPB(common) Message [172.30.8.53:15457(SunC)/Cisco-IOS-XR-infra-statsd-oper:infra-statistics/interfaces/interface/latest/generic-counters msg len: 5984]
{
    "Source": "172.30.8.53:15457",
    "Telemetry": {
        "node_id_str": "SunC",
        "subscription_id_str": "Sub1",
        "encoding_path": "Cisco-IOS-XR-infra-statsd-oper:infra-statistics/interfaces/interface/latest/generic-counters",
        "collection_id": 712163,
        "collection_start_time": 1491277026499,
        "msg_timestamp": 1491277026499,
        "collection_end_time": 1491277026507
    },
    "Rows": [
        {   
            "Timestamp": 1491277026506,
            "Keys": {
                "interface-name": "MgmtEth0/RP0/CPU0/0"
            },
            "Content": {
                "applique": 0,
                "availability-flag": 0,
                "broadcast-packets-received": 65679,
                "broadcast-packets-sent": 0,
                "bytes-received": 272894774,
                "bytes-sent": 20829696017,
                "carrier-transitions": 1,
                <output snipped for brevity>

Why Did We Do That Again?

To leverage the real power of telemetry, you need to get the data into an analytics stack like InfluxDB or Prometheus…or to multiple consumers via a pub/sub mechanism like Kafka. Pipeline can do all that and I’ll show you how in future tutorials.

But having the power to dump encoded telemetry data into a text file does come in handy, especially when you’re setting up telemetry and Pipeline for the first time. The tap output module gives you a quick and easy way to validate that the router is sending the data you think it should be sending. Once that’s settled, it’s a simple matter of configuring a different output module to send the data some place really useful.

Give Pipeline a try and let us know what you think!

Leave a Comment