Cleanup of Exec Inputs documentation - redux

Hi @sparrc

(Sorry for the noise - new pr)

Please find some improvements to readability including the \n for the exec/telegraf line-protocol input.

I hope you (and others) find it easier to read.

/Mark

This is an ammend
This commit is contained in:
marknmel 2016-03-15 09:29:05 -04:00 committed by Mark Brown
parent 2cb32a683e
commit 698ceff2d7
1 changed files with 35 additions and 25 deletions

View File

@ -1,28 +1,20 @@
# Exec Input Plugin # Exec Input Plugin
The exec plugin can execute arbitrary commands which output: Please also see: [Telegraf Input Data Formats](https://github.com/influxdata/telegraf/blob/master/docs/DATA_FORMATS_INPUT.md)
* JSON The exec input plugin can execute arbitrary commands which output:
* InfluxDB [line-protocol](https://docs.influxdata.com/influxdb/v0.9/write_protocols/line/)
* JSON [javascript object notation](http://www.json.org/)
* InfluxDB [line-protocol](https://docs.influxdata.com/influxdb/v0.10/write_protocols/line/)
* Graphite [graphite-protocol](http://graphite.readthedocs.org/en/latest/feeding-carbon.html) * Graphite [graphite-protocol](http://graphite.readthedocs.org/en/latest/feeding-carbon.html)
> Graphite understands messages with this format:
> ``` ### Example 1 - JSON
metric_path value timestamp\n
```
> __metric_path__ is the metric namespace that you want to populate. #### Configuration
> __value__ is the value that you want to assign to the metric at this time. In this example a script called ```/tmp/test.sh``` and a script called ```/tmp/test2.sh```
are configured for ```[[inputs.exec]]``` in JSON format.
> __timestamp__ is the unix epoch time.
If using JSON, only numeric values are parsed and turned into floats. Booleans
and strings will be ignored.
### Configuration
``` ```
# Read flattened metrics from one or more commands that output JSON to stdout # Read flattened metrics from one or more commands that output JSON to stdout
@ -64,8 +56,6 @@ Other options for modifying the measurement names are:
name_prefix = "prefix_" name_prefix = "prefix_"
``` ```
### Example 1
Let's say that we have the above configuration, and mycollector outputs the Let's say that we have the above configuration, and mycollector outputs the
following JSON: following JSON:
@ -85,10 +75,16 @@ The collected metrics will be stored as fields under the measurement
``` ```
exec_mycollector a=0.5,b_c=0.1,b_d=5 1452815002357578567 exec_mycollector a=0.5,b_c=0.1,b_d=5 1452815002357578567
``` ```
If using JSON, only numeric values are parsed and turned into floats. Booleans
and strings will be ignored.
### Example 2 ### Example 2 - Influx Line-Protocol
Now let's say we have the following configuration: In this example an application called ```/usr/bin/line_protocol_collector```
and a script called ```/tmp/test2.sh``` are configured for ```[[inputs.exec]]```
in influx line-protocol format.
#### Configuration
``` ```
[[inputs.exec]] [[inputs.exec]]
@ -103,7 +99,7 @@ Now let's say we have the following configuration:
data_format = "influx" data_format = "influx"
``` ```
And line_protocol_collector outputs the following line protocol: The line_protocol_collector application outputs the following line protocol:
``` ```
cpu,cpu=cpu0,host=foo,datacenter=us-east usage_idle=99,usage_busy=1 cpu,cpu=cpu0,host=foo,datacenter=us-east usage_idle=99,usage_busy=1
@ -118,15 +114,18 @@ cpu,cpu=cpu6,host=foo,datacenter=us-east usage_idle=99,usage_busy=1
You will get data in InfluxDB exactly as it is defined above, You will get data in InfluxDB exactly as it is defined above,
tags are cpu=cpuN, host=foo, and datacenter=us-east with fields usage_idle tags are cpu=cpuN, host=foo, and datacenter=us-east with fields usage_idle
and usage_busy. They will receive a timestamp at collection time. and usage_busy. They will receive a timestamp at collection time.
Each line must end in \n, just as the Influx line protocol does.
### Example 3 ### Example 3 - Graphite
We can also change the data_format to "graphite" to use the metrics collecting scripts such as (compatible with graphite): We can also change the data_format to "graphite" to use the metrics collecting scripts such as (compatible with graphite):
* Nagios [Mertics Plugins] (https://exchange.nagios.org/directory/Plugins) * Nagios [Mertics Plugins] (https://exchange.nagios.org/directory/Plugins)
* Sensu [Mertics Plugins] (https://github.com/sensu-plugins) * Sensu [Mertics Plugins] (https://github.com/sensu-plugins)
In this example a script called /tmp/test.sh and a script called /tmp/test2.sh are configured for [[inputs.exec]] in graphite format.
#### Configuration #### Configuration
``` ```
# Read flattened metrics from one or more commands that output JSON to stdout # Read flattened metrics from one or more commands that output JSON to stdout
@ -161,6 +160,17 @@ We can also change the data_format to "graphite" to use the metrics collecting s
"measurement*" "measurement*"
] ]
``` ```
Graphite messages are in this format:
```
metric_path value timestamp\n
```
__metric_path__ is the metric namespace that you want to populate.
__value__ is the value that you want to assign to the metric at this time.
__timestamp__ is the unix epoch time.
And test.sh/test2.sh will output: And test.sh/test2.sh will output: