Telegraf `docker_log` 不发送所有消息答案

【问题标题】：Telegraf `docker_log` does not send all messagesTelegraf `docker_log` 不发送所有消息
【发布时间】：2019-09-26 13:43:18
【问题描述】：

我想用telegraf ``inputs.docker_log从几个go服务中收集influxdb的日志。

电报：电报：1.12-高山
influxdb: influxdb:1.7.7-alpine

我发现 influxdb 并不包含我在docker service logs 中看到的所有记录。

我做了简单的 go 脚本。并做几个不同延迟的实验。

package main

import (
    "time"

    "github.com/labstack/gommon/log"
)

func main() {

    log.SetLevel(log.DEBUG)
    for i := 0; i < 5; i++ {
        writeSomething()
        <-time.After(10 * time.Second)
    }
}

func writeSomething() {
    delay := 50 * time.Millisecond
    log.Debug("A")
    <-time.After(delay)

    log.Debug("B")
    <-time.After(delay)

    log.Debug("C")
    <-time.After(delay)

}

我发现延迟

例如。延迟 = 50 毫秒

Docker service logs 显示所有 3 个变体（A、B、C）5 次

liquineq_sender.1.0jvam7vg63cm@Ivan-Lenovo-ideapad    | {"time":"2019-09-26T13:47:17.115060311Z","level":"DEBUG","prefix":"-","file":"main.go","line":"20","message":"A"}
liquineq_sender.1.0jvam7vg63cm@Ivan-Lenovo-ideapad    | {"time":"2019-09-26T13:47:17.165383407Z","level":"DEBUG","prefix":"-","file":"main.go","line":"23","message":"B"}
liquineq_sender.1.0jvam7vg63cm@Ivan-Lenovo-ideapad    | {"time":"2019-09-26T13:47:17.21562549Z","level":"DEBUG","prefix":"-","file":"main.go","line":"26","message":"C"}

但我在 influxdb 中找不到其中一些：

> select message from docker_log;
name: docker_log
time                message
----                -------
1569505640000000000 2019-09-26T13:47:18Z I! Starting Telegraf 1.12.2
1569505640000000000 t=2019-09-26T13:47:12+0000 lvl=info msg="Starting Grafana" logger=server version=6.3.0-pre commit=unknown-dev branch=master compiled=2019-06-21T08:57:10+0000
1569505640000000000 {"time":"2019-09-26T13:47:17.115060311Z","level":"DEBUG","prefix":"-","file":"main.go","line":"20","message":"A"}
1569505647000000000 {"time":"2019-09-26T13:47:27.265995126Z","level":"DEBUG","prefix":"-","file":"main.go","line":"20","message":"A"}
1569505657000000000 {"time":"2019-09-26T13:47:37.417263994Z","level":"DEBUG","prefix":"-","file":"main.go","line":"20","message":"A"}
1569505658000000000 {"time":"2019-09-26T13:47:37.517861048Z","level":"DEBUG","prefix":"-","file":"main.go","line":"26","message":"C"}
1569505668000000000 {"time":"2019-09-26T13:47:47.568291482Z","level":"DEBUG","prefix":"-","file":"main.go","line":"20","message":"A"}
1569505678000000000 {"time":"2019-09-26T13:47:57.71926933Z","level":"DEBUG","prefix":"-","file":"main.go","line":"20","message":"A"}

如果将延迟增加到 1 秒，则 influxdb 包含所有记录。

有没有办法设置 telegraf / influxdb 也解析毫秒？

【问题讨论】：

标签： influxdb telegraf

【解决方案1】：

我通过在电报配置中添加以下设置解决了一个问题。

[agent]
  precision = "100ns"

说明：
telegraf 的日志累加器有方法getTime。该方法用一些precision 来循环日志的时间。

来自代理配置：

// By default or when set to "0s", precision will be set to the same
// timestamp order as the collection interval, with the maximum being 1s.
//   ie, when interval = "10s", precision will be "1s"
//       when interval = "250ms", precision will be "1ms"
// Precision will NOT be used for service inputs. It is up to each individual
// service input to set the timestamp at the appropriate precision.

【讨论】：