gRPC: 如何合理管理日志配置?

2021-12-13 00:19:10 浏览数 (4)

介绍

本文介绍如何在 gRPC 场景下使用日志。将使用 rk-boot 来管理日志配置。

有什么使用场景?

日志自动滚动 分成多个日志文件 日志格式修改 等等

请访问如下地址获取完整教程:

  • https://rkdev.info/cn
  • https://rkdocs.netlify.app/cn (备用)

安装

代码语言:txt复制
go get github.com/rookie-ninja/rk-boot
go get github.com/rookie-ninja/rk-grpc

简述概念

rk-boot 使用如下两个库管理日志。

  • zap 管理日志实例
  • lumberjack 管理日志滚动

rk-boot 定义了两种日志类型,会在后面详细介绍,这里先做个简短介绍。

  • ZapLogger: 标准日志,用于记录 Error, Info 等。
  • EventLogger: JSON 或者 Console 格式,用于记录 Event,例如 RPC 请求。

快速开始

在这个例子中,我们会试着改变 zap 日志的路径和格式。

1.创建 boot.yaml

代码语言:txt复制
---
zapLogger:
  - name: zap-log                      # Required
    zap:
      encoding: json
      outputPaths: ["logs/zap.log"]
grpc:
  - name: greeter
    port: 8080
    enabled: true

2.创建 main.go

往 zap-log 日志实例中写个日志。

代码语言:txt复制
package main

import (
	"context"
	"github.com/rookie-ninja/rk-boot"
	"github.com/rookie-ninja/rk-entry/entry"
	_ "github.com/rookie-ninja/rk-grpc/boot"
)

func main() {
	// Create a new boot instance.
	boot := rkboot.NewBoot()

	// Bootstrap
	boot.Bootstrap(context.Background())

	// Write zap log
	rkentry.GlobalAppCtx.GetZapLoggerEntry("zap-log").GetLogger().Info("This is zap-log")

	// Wait for shutdown sig
	boot.WaitForShutdownSig(context.Background())
}

3.启动 main.go

代码语言:txt复制
$ go run main.go

4.验证

文件夹结构以及日志内容。

代码语言:txt复制
├── boot.yaml
├── go.mod
├── go.sum
├── logs
│   └── zap.log
└── main.go
代码语言:txt复制
{"level":"INFO","ts":"2021-10-21T02:10:09.279 0800","msg":"This is zap-log"}

配置 EventLogger

上面的例子中,我们配置了 zap 日志,这回我们修改一下 EventLogger。

1.创建 boot.yaml

代码语言:txt复制
---
eventLogger:
  - name: event-log                 # Required
    encoding: console
    outputPaths: ["logs/event.log"]
grpc:
  - name: greeter
    port: 8080
    enabled: true

2.创建 main.go

往 event-log 实例中写入日志。

代码语言:txt复制
package main

import (
	"context"
	"github.com/rookie-ninja/rk-boot"
	"github.com/rookie-ninja/rk-entry/entry"
	_ "github.com/rookie-ninja/rk-grpc/boot"
)

func main() {
	// Create a new boot instance.
	boot := rkboot.NewBoot()

	// Bootstrap
	boot.Bootstrap(context.Background())

	// Write event log
	helper := rkentry.GlobalAppCtx.GetEventLoggerEntry("event-log").GetEventHelper()
	event := helper.Start("demo-event")
	event.AddPair("key", "value")
	helper.Finish(event)

	// Wait for shutdown sig
	boot.WaitForShutdownSig(context.Background())
}

3.启动 main.go

代码语言:txt复制
$ go run main.go

4.验证

文件夹结构以及日志内容。

代码语言:txt复制
├── boot.yaml
├── go.mod
├── go.sum
├── logs
│   └── event.log
└── main.go
代码语言:txt复制
{"endTime": "2021-10-21T02:22:58.118 0800", "startTime": "2021-10-21T02:22:58.118 0800", "elapsedNano": 409, "timezone": "CST", "ids": {"eventId":"510a050f-c31e-4f50-8d6e-3b836ba1ef17"}, "app": {"appName":"demo","appVersion":"master-7d51358","entryName":"","entryType":""}, "env": {"arch":"amd64","az":"*","domain":"*","hostname":"lark.local","localIP":"10.8.0.2","os":"darwin","realm":"*","region":"*"}, "payloads": {}, "error": {}, "counters": {}, "pairs": {"key":"value"}, "timing": {}, "remoteAddr": "localhost", "operation": "demo-event", "eventStatus": "Ended", "resCode": "OK"}

概念

上面的例子中,我们尝试了 ZapLogger 和 EventLogger。接下来我们看看 rk-boot 是如何实现的,并且怎么使用。

架构

ZapLoggerEntry

ZapLoggerEntry 是 zap 实例的一个封装。

代码语言:txt复制
// ZapLoggerEntry contains bellow fields.
// 1: EntryName: Name of entry.
// 2: EntryType: Type of entry which is ZapLoggerEntryType.
// 3: EntryDescription: Description of ZapLoggerEntry.
// 4: Logger: zap.Logger which was initialized at the beginning.
// 5: LoggerConfig: zap.Logger config which was initialized at the beginning which is not accessible after initialization..
// 6: LumberjackConfig: lumberjack.Logger which was initialized at the beginning.
type ZapLoggerEntry struct {
	EntryName        string             `yaml:"entryName" json:"entryName"`
	EntryType        string             `yaml:"entryType" json:"entryType"`
	EntryDescription string             `yaml:"entryDescription" json:"entryDescription"`
	Logger           *zap.Logger        `yaml:"-" json:"-"`
	LoggerConfig     *zap.Config        `yaml:"zapConfig" json:"zapConfig"`
	LumberjackConfig *lumberjack.Logger `yaml:"lumberjackConfig" json:"lumberjackConfig"`
}

如何在 boot.yaml 里配置 ZapLoggerEntry?

ZapLoggerEntry 完全兼容 zap 和 lumberjack 的 YAML 结构。

用户可以根据需求,配置多个 ZapLogger 实例,并且通过 name 来访问。

完整配置:

代码语言:txt复制
---
zapLogger:
  - name: zap-logger                      # Required
    description: "Description of entry"   # Optional
    zap:
      level: info                         # Optional, default: info, options: [debug, DEBUG, info, INFO, warn, WARN, dpanic, DPANIC, panic, PANIC, fatal, FATAL]
      development: true                   # Optional, default: true
      disableCaller: false                # Optional, default: false
      disableStacktrace: true             # Optional, default: true
      sampling:                           # Optional, default: empty map
        initial: 0
        thereafter: 0
      encoding: console                   # Optional, default: "console", options: [console, json]
      encoderConfig:
        messageKey: "msg"                 # Optional, default: "msg"
        levelKey: "level"                 # Optional, default: "level"
        timeKey: "ts"                     # Optional, default: "ts"
        nameKey: "logger"                 # Optional, default: "logger"
        callerKey: "caller"               # Optional, default: "caller"
        functionKey: ""                   # Optional, default: ""
        stacktraceKey: "stacktrace"       # Optional, default: "stacktrace"
        lineEnding: "n"                  # Optional, default: "n"
        levelEncoder: "capitalColor"      # Optional, default: "capitalColor", options: [capital, capitalColor, color, lowercase]
        timeEncoder: "iso8601"            # Optional, default: "iso8601", options: [rfc3339nano, RFC3339Nano, rfc3339, RFC3339, iso8601, ISO8601, millis, nanos]
        durationEncoder: "string"         # Optional, default: "string", options: [string, nanos, ms]
        callerEncoder: ""                 # Optional, default: ""
        nameEncoder: ""                   # Optional, default: ""
        consoleSeparator: ""              # Optional, default: ""
      outputPaths: [ "stdout" ]           # Optional, default: ["stdout"], stdout would be replaced if specified
      errorOutputPaths: [ "stderr" ]      # Optional, default: ["stderr"], stderr would be replaced if specified
      initialFields:                      # Optional, default: empty map
        key: "value"
    lumberjack:                           # Optional
      filename: "rkapp-event.log"         # Optional, default: It uses <processname>-lumberjack.log in os.TempDir() if empty.
      maxsize: 1024                       # Optional, default: 1024 (MB)
      maxage: 7                           # Optional, default: 7 (days)
      maxbackups: 3                       # Optional, default: 3 (days)
      localtime: true                     # Optional, default: true
      compress: true                      # Optional, default: true

如何在代码里获取 ZapLogger?

通过 name 来访问。

代码语言:txt复制
// Access entry
rkentry.GlobalAppCtx.GetZapLoggerEntry("zap-logger")

// Access zap logger
rkentry.GlobalAppCtx.GetZapLoggerEntry("zap-logger").GetLogger()

// Access zap logger config
rkentry.GlobalAppCtx.GetZapLoggerEntry("zap-logger").GetLoggerConfig()

// Access lumberjack config
rkentry.GlobalAppCtx.GetZapLoggerEntry("zap-logger").GetLumberjackConfig()

EventLoggerEntry

rk-boot 把每一个 RPC 请求看作一个 Event,并且使用 rk-query 中的 Event 类型来记录日志。

代码语言:txt复制
// EventLoggerEntry contains bellow fields.
// 1: EntryName: Name of entry.
// 2: EntryType: Type of entry which is EventLoggerEntryType.
// 3: EntryDescription: Description of EventLoggerEntry.
// 4: EventFactory: rkquery.EventFactory was initialized at the beginning.
// 5: EventHelper: rkquery.EventHelper was initialized at the beginning.
// 6: LoggerConfig: zap.Config which was initialized at the beginning which is not accessible after initialization.
// 7: LumberjackConfig: lumberjack.Logger which was initialized at the beginning.
type EventLoggerEntry struct {
	EntryName        string                `yaml:"entryName" json:"entryName"`
	EntryType        string                `yaml:"entryType" json:"entryType"`
	EntryDescription string                `yaml:"entryDescription" json:"entryDescription"`
	EventFactory     *rkquery.EventFactory `yaml:"-" json:"-"`
	EventHelper      *rkquery.EventHelper  `yaml:"-" json:"-"`
	LoggerConfig     *zap.Config           `yaml:"zapConfig" json:"zapConfig"`
	LumberjackConfig *lumberjack.Logger    `yaml:"lumberjackConfig" json:"lumberjackConfig"`
}

EventLogger 字段

我们可以看到 EventLogger 打印出来的日志里,包含字段,介绍一下这些字段。

字段

详情

endTime

结束时间

startTime

开始时间

elapsedNano

Event 时间开销(Nanoseconds)

timezone

时区

ids

包含 eventId, requestId 和 traceId。如果原数据拦截器被启动,或者 event.SetRequest() 被用户调用,新的 RequestId 将会被使用,同时 eventId 与 requestId 会一模一样。 如果调用链拦截器被启动,traceId 将会被记录。

app

包含 appName, appVersion, entryName, entryType。

env

包含 arch, az, domain, hostname, localIP, os, realm, region. realm, region, az, domain 字段。这些字段来自系统环境变量(REALM,REGION,AZ,DOMAIN)。 "*" 代表环境变量为空。

payloads

包含 RPC 相关信息。

error

包含错误。

counters

通过 event.SetCounter() 来操作。

pairs

通过 event.AddPair() 来操作。

timing

通过 event.StartTimer() 和 event.EndTimer() 来操作。

remoteAddr

RPC 远程地址。

operation

RPC 名字。

resCode

RPC 返回码。

eventStatus

Ended 或者 InProgress

例子------------------------------------------------------------------------ endTime=2021-07-10T03:00:12.153392 08:00 startTime=2021-07-10T03:00:12.153261 08:00 elapsedNano=130727 timezone=CST ids={"eventId":"c9a1f6b0-b9ec-4e46-9ed4-238c3c6759ab","requestId":"c9a1f6b0-b9ec-4e46-9ed4-238c3c6759ab","traceId":"5441ff5c3855f03b573e95d81139123b"} app={"appName":"rk-demo","appVersion":"master-f414049","entryName":"greeter","entryType":"GrpcEntry"} env={"arch":"amd64","az":"*","domain":"*","hostname":"lark.local","localIP":"10.8.0.2","os":"darwin","realm":"*","region":"*"} payloads={"grpcMethod":"Greeter","grpcService":"api.v1.Greeter","grpcType":"unaryServer","gwMethod":"GET","gwPath":"/v1/greeter","gwScheme":"http","gwUserAgent":"curl/7.64.1"} error={} counters={} pairs={} timing={} remoteAddr=localhost:59631 operation=/api.v1.Greeter/Greeter resCode=OK eventStatus=Ended EOE

如何在 boot.yaml 里配置 EventLoggerEntry?

EventLoggerEntry 将会把 Application 名字注入到 Event 中。启动器会从 go.mod 文件中提取 Application 名字。 如果没有 go.mod 文件,启动器会使用默认的名字。

用户可以根据需求,配置多个 EventLogger 实例,并且通过 name 来访问。

完整配置:

代码语言:txt复制
---
eventLogger:
  - name: event-logger                 # Required
    description: "This is description" # Optional
    encoding: console                  # Optional, default: console, options: console and json
    outputPaths: ["stdout"]            # Optional
    lumberjack:                        # Optional
      filename: "rkapp-event.log"      # Optional, default: It uses <processname>-lumberjack.log in os.TempDir() if empty.
      maxsize: 1024                    # Optional, default: 1024 (MB)
      maxage: 7                        # Optional, default: 7 (days)
      maxbackups: 3                    # Optional, default: 3 (days)
      localtime: true                  # Optional, default: true
      compress: true                   # Optional, default: true

如何在代码里获取 EventLogger?

通过 name 来访问。

代码语言:txt复制
// Access entry
rkentry.GlobalAppCtx.GetEventLoggerEntry("event-logger")

// Access event factory
rkentry.GlobalAppCtx.GetEventLoggerEntry("event-logger").GetEventFactory()

// Access event helper
rkentry.GlobalAppCtx.GetEventLoggerEntry("event-logger").GetEventHelper()

// Access lumberjack config
rkentry.GlobalAppCtx.GetEventLoggerEntry("event-logger").GetLumberjackConfig()

如何使用 Event?

Event 是一个 interface,包含了若干方法,请参考:Event

常用方法:

代码语言:txt复制
// Get EventHelper to create Event instance
helper := rkentry.GlobalAppCtx.GetEventLoggerEntry("event-log").GetEventHelper()

// Start and finish event
event := helper.Start("demo-event")
helper.Finish(event)

// Add K/V
event.AddPair("key", "value")

// Start and end timer
event.StartTimer("my-timer")
event.EndTimer("my-timer")

// Set counter
event.SetCounter("my-counter", 1)

1 人点赞