Skip to content

quwan-sre/observability-go-contrib

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

22 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Go Contrib

CI codecov PkgGoDev Go Report Card

为 Go Packages 提供可观测性的扩展插件。

目标与边界

针对特定的 Go Package 提供 middlewareinterceptorpluginhookcallback 等形式的可观测性埋点支持,Go 应用在启动、初始化时加载注册本项目组件,使特定操作(如网络调用、中间件调用)上报可观测性数据。

扩展插件只实现于特定 Package 提供的扩展接口之上,并且统一定义数据模型(如指标名、指标 Labels),但不干预用户在服务内做自定义的埋点。

扩展插件使用以下 Package 进行埋点:

  • metrics:github.com/prometheus/client_golang
  • traces:github.com/open-telemetry/opentelemetry-go
  • logs: #TODO

接入使用

请阅读具体扩展插件目录下的 README.md 了解如何使用,如 go-contrib/metrics/gin/README.md 提供了对 github.com/gin-gonic/gin HTTP 框架 metrics 插装的指引。

Metrics 定义

RPC Metrics

接收请求指标

指标名称:

  • apm_rpc_receive_request_duration_milliseconds_count
  • apm_rpc_receive_request_duration_milliseconds_sum
  • apm_rpc_receive_request_duration_milliseconds_bucket

包含 Labels:

  • sdk:插装 SDK,枚举值见 quwan-sre/observability-go-contrib
  • request_protocol:请求协议,枚举值:http | grpc
  • request_target:请求目标,HTTP 请求为域名,gRPC 请求为服务名,示例值:www.baidu.com | RouteGuide | unknown
  • request_path:请求路径,示例值:/api/v1/foo/bar | /routeguide.RouteGuide/RouteChat
  • grpc_response_status:RPC 状态码,遵循 GRPC Core: Status codes and their use in gRPC,通常 0 为 OK;
  • response_code:HTTP 状态码,遵循 HTTP/1.1: Status Code Definitions,通常 200 为 OK

代码定义:

	DefaultRPCReceiveRequestMetric = prometheus.NewHistogramVec(prometheus.HistogramOpts{
		Name:                            DefaultRPCReceiveRequestMetricName,
		Buckets:                         []float64{0.5, 1, 5, 10, 25, 50, 100, 250, 500, 1000, 2500, 5000, 10000, 30000, 60000, 300000, 600000, 1800000, 3600000},
		NativeHistogramBucketFactor:     1.4,
		NativeHistogramZeroThreshold:    0.5,
		NativeHistogramMinResetDuration: 5 * time.Minute,
		NativeHistogramMaxZeroThreshold: 50,
		NativeHistogramMaxBucketNumber:  20,
	}, []string{"sdk", "request_protocol", "request_target", "request_path", "grpc_response_status", "response_code"})

发送请求指标

指标名称:

  • apm_rpc_send_request_duration_milliseconds_count
  • apm_rpc_send_request_duration_milliseconds_sum
  • apm_rpc_send_request_duration_milliseconds_bucket

包含 Labels:

  • sdk:插装 SDK,枚举值见 quwan-sre/observability-go-contrib
  • request_protocol:请求协议,枚举值:http | grpc
  • request_target:请求目标,HTTP 请求为域名,gRPC 请求为服务名,示例值:www.baidu.com | RouteGuide | unknown
  • request_path:请求路径,示例值:127.0.0.1:8080/health | /routeguide.RouteGuide/RouteChat
  • grpc_response_status:RPC 状态码,遵循 GRPC Core: Status codes and their use in gRPC,通常 0 为 OK;
  • response_code:HTTP 状态码,遵循 HTTP/1.1: Status Code Definitions,通常 200 为 OK

代码定义:

	DefaultRPCSendRequestMetric = prometheus.NewHistogramVec(prometheus.HistogramOpts{
		Name:                            DefaultRPCSendRequestMetricName,
		Buckets:                         []float64{0.5, 1, 5, 10, 25, 50, 100, 250, 500, 1000, 2500, 5000, 10000, 30000, 60000, 300000, 600000, 1800000, 3600000},
		NativeHistogramBucketFactor:     1.4,
		NativeHistogramZeroThreshold:    0.5,
		NativeHistogramMinResetDuration: 5 * time.Minute,
		NativeHistogramMaxZeroThreshold: 50,
		NativeHistogramMaxBucketNumber:  20,
	}, []string{"sdk", "request_protocol", "request_target", "request_path", "grpc_response_status", "response_code"})

备注

  1. 由于 etcd 同样使用 gRPC 协议通信,并且使用 gRPC Interceptor 埋点采集,因此指标中包含与 etcd 进行通信的数据;
  2. 对于 gRPC Streaming 类型请求,同样记录在以上指标中,但是 gRPC Streaming 是双向通信(可单向接收、可单向发送、可同时收发),因此在请求耗时上会不同(测试时显著低)于 Unary gRPC 请求响应;
  3. 对于发送请求指标中的 request_target, gRPC 请求填入的值为 gRPC 请求方法(格式为:/package.service/method)中的 service,由 proto 定义,而非 Kubernetes、Istio 中的 Service,它们可能并不相同;
  4. 对于 request_pathrequest_target,若无法获取,则填入 unknown

Database Metrics

发送请求指标

指标名称:

  • apm_database_send_request_duration_milliseconds_count
  • apm_database_send_request_duration_milliseconds_sum
  • apm_database_send_request_duration_milliseconds_bucket

包含 Labels:

  • sdk:插装 SDK,枚举值见 quwan-sre/observability-go-contrib
  • database_type:数据库类型,枚举值:mysql | mongodb
  • database_addr:数据库地址,通过 SDK 元数据获取或用户填入,示例值:127.0.0.1:3306
  • response_status:请求响应状态:枚举值:成功 0 | 错误 1
  • query_type:语句类型,随不同数据库类型而不同,示例值: create | dropDatabase | ping

代码定义:

	DefaultDatabaseSendRequestMetric = prometheus.NewHistogramVec(prometheus.HistogramOpts{
		Name:                            DefaultDatabaseSendRequestMetricName,
		Buckets:                         []float64{1, 2.5, 5, 10, 20, 50, 100, 500, 1000, 2500, 5000, 7500, 10000},
		NativeHistogramBucketFactor:     1.4,
		NativeHistogramZeroThreshold:    1,
		NativeHistogramMinResetDuration: 5 * time.Minute,
		NativeHistogramMaxZeroThreshold: 50,
		NativeHistogramMaxBucketNumber:  20,
	}, []string{"sdk", "database_type", "database_addr", "response_status", "query_type"})

缓存指标

发送请求指标

指标名称:

  • apm_cache_send_request_duration_milliseconds_count
  • apm_cache_send_request_duration_milliseconds_sum
  • apm_cache_send_request_duration_milliseconds_bucket

包含 Labels:

  • sdk:插装 SDK,枚举值见 quwan-sre/observability-go-contrib
  • cache_type:数据库类型,枚举值:redis | memcached
  • cache_addr:数据库地址,通过 SDK 元数据获取或用户填入,示例值:127.0.0.1:6379
  • response_status:请求响应状态:枚举值:成功 0 | 错误 1
  • command:语句类型,随不同缓存类型而不同,示例值: get | pipeline

代码定义:

	DefaultCacheRequestMetric = prometheus.NewHistogramVec(prometheus.HistogramOpts{
		Name:                            DefaultCacheRequestMetricName,
		Buckets:                         []float64{0.25, 0.5, 1, 2, 5, 10, 25, 50, 100, 250, 500, 1000, 3000, 5000},
		NativeHistogramBucketFactor:     1.4,
		NativeHistogramZeroThreshold:    0.1,
		NativeHistogramMinResetDuration: 5 * time.Minute,
		NativeHistogramMaxZeroThreshold: 10,
		NativeHistogramMaxBucketNumber:  20,
	}, []string{"sdk", "cache_type", "cache_addr", "command", "response_status"})

消息队列指标

发送消息指标

  • apm_mq_send_msg_count

包含 Labels:

  • sdk:插装 SDK,枚举值见 quwan-sre/observability-go-contrib
  • mq_type:MQ 类型,枚举值:kafka
  • mq_addr:MQ Broker 地址,通过 SDK 元数据获取或用户填入,示例值:127.0.0.1:9092
  • mq_topic:MQ Topic,示例值:my_kafka_topic
  • mq_partition:MQ 分区,若发送时未知则为 unknown

代码定义:

	DefaultMQSendMsgMetric = prometheus.NewCounterVec(prometheus.CounterOpts{
		Name: DefaultMQSendMsgMetricName,
	}, []string{"sdk", "mq_type", "mq_addr", "mq_topic", "mq_partition"})

消费消息指标

  • apm_mq_receive_msg_count

包含 Labels:

  • sdk:插装 SDK,枚举值见 quwan-sre/observability-go-contrib
  • mq_type:MQ 类型,枚举值:kafka
  • mq_addr:MQ Broker 地址,通过 SDK 元数据获取或用户填入,示例值:127.0.0.1:9092
  • mq_topic:MQ Topic,示例值:my_kafka_topic
  • mq_partition:MQ 分区,示例值:0 | 1 | 2

代码定义:

	DefaultMQReceiveMsgMetric = prometheus.NewCounterVec(prometheus.CounterOpts{
		Name: DefaultMQReceiveMsgMetricName,
	}, []string{"sdk", "mq_type", "mq_addr", "mq_topic", "mq_partition"})

Go Packages 支持列表

分类 Package 数据类型 & 稳定性状态
HTTP github.com/gin-gonic/gin alpha: metrics
HTTP github.com/go-resty/resty/v2 alpha: metrics
RPC google.golang.org/grpc alpha: metrics
Redis github.com/go-redis/redis/v9 alpha: metrics
Kafka github.com/Shopify/sarama alpha: metrics
MongoDB go.mongodb.org/mongo-driver alpha: metrics
etcd go.etcd.io/etcd/client/v3 alpha: metrics

About

Collection of observability extensions.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages