Telegraf-rust is a lightweight client library for general metrics writing using Telegraf. Telegraf is a micro-service provided by InfluxData for making metrics reporting easy for distributed services - see their docs for more information.
This library does not provide querying or other InfluxDB client-library features. This is meant to be lightweight and simple for services to report metrics.
Telegraf-rust supports all socket connection types, such as UDS (unix domain socket):
- TCP (
tcp://) - UDP (
udp://) - UDS Stream (
unix://) - UDS Datagram (
unixgram://)
Add it to your Cargo.toml:
[dependencies]
telegraf = "*"Using this library assumes you have a socket listener setup in your Telegraf configuration file. An example TCP connection looks like so:
[[inputs.socket_listener]]
service_address = "tcp://localhost:8094"All usage will start by creating a socket connection via a Client. This supports multiple connection protocols - which one you use will be determined by how your Telegraf input.socket_listener configuration is setup.
Once a client is setup there are multiple different ways to write points:
use telegraf::*;
let mut client = Client::new("tcp://localhost:8094").unwrap();
#[derive(Metric)]
struct MyMetric {
field1: i32,
#[telegraf(tag)]
tag1: String,
}
let point = MyMetric { field1: 1, tag1: "tag" };
client.write(&point);By default the measurement name will be the same as the struct. You can override this via derive attributes:
use telegraf::*;
#[derive(Metric)]
#[measurement = "custom_name"]
struct MyMetric {
field1: i32,
}As with any Telegraf point, tags are optional but at least one field is required.
Timestamps are optional and can be set via the timestamp attribute:
use telegraf::*;
#[derive(Metric)]
struct MyMetric {
#[telegraf(timestamp)]
ts: u64,
field1: i32,
}use telegraf::*;
let mut client = Client::new("tcp://localhost:8094").unwrap();
let p = point!("measurement", ("tag1", "tag1Val"), ("field1", "val") ("field2", 10); 100);
client.write_point(&p);The macro syntax is the following format:
(<measurement>, [(<tagName>, <tagVal>)], [(<fieldName>, <fieldVal>)]; <timestamp>)
Measurement name, tag set, and field set are comma separated. Tag and field tuples are space separated. Timestamp is semicolon separated. The tag set and timestamp are optional.
use telegraf::{Client, Point};
let c = Client::new("tcp://localhost:8094").unwrap();
let p = Point::new(
String::from("measurement"),
vec![
(String::from("tag1"), String::from("tag1value"))
],
vec![
(String::from("field1"), Box::new(10)),
(String::from("field2"), Box::new(20.5)),
(String::from("field3"), Box::new("anything!"))
],
Some(100),
);
c.write_point(p)Any attribute that will be the value of a field must implement the IntoFieldData trait provided by this library.
pub trait IntoFieldData {
fn field_data(&self) -> FieldData;
}Out of the box implementations are provided for many common data types, but manual implementation is possible for other data types.
Timestamps are optional. If not present, the Telegraf daemon will set the timestamp using the current time.
Timestamps are specified in nanosecond-precision Unix time, therefore u64 must implement the From<T> trait for the field type, if the implementation is not already present:
use telegraf::*;
#[derive(Copy, Clone)]
struct MyType {
// ...
}
impl From<MyType> for u64 {
fn from(my_type: MyType) -> Self {
todo!()
}
}
#[derive(Metric)]
struct MyMetric {
#[telegraf(timestamp)]
ts: MyType,
field1: i32,
}More information about timestamps can be found here.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent