Jaeger (Distributed Tracing System)

Overview

jaeger란? Uber Technologies에서 만든 분산형 추적시스템 오픈소스로서 마이크로 서비스 기반 분산 시스템 모니터링을 할수가 있다.

현재는 CNCF(Cloud Native Computing Foundation)의 프로젝트이다.

- 분산형 추적시스템에 대표적인 오픈소스로는 zipkin, jagger, appdash 가 대표적이다.

- 분산 컨텍스트 전파

- 분산 트랜잭션 모니터링

- 근본 원인 분석

- 서비스 의존성 분석

- 성능 / 대기 시간 최적화

Installation

client

application.yml

?

1 2	compile group: 'io.opentracing.contrib', name:'opentracing-spring-web-autoconfigure', version: '0.0.4' compile group: 'com.uber.jaeger', name: 'jaeger-core', version: '0.27.0

---

jaegerConfig.java
 

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30

import com.uber.jaeger.Configuration;   import com.uber.jaeger.samplers.ProbabilisticSampler;   import org.springframework.beans.factory.annotation.Value;   import org.springframework.boot.web.client.RestTemplateBuilder;   import org.springframework.context.annotation.Bean;   import org.springframework.web.client.RestTemplate;   @org.springframework.context.annotation.Configuration   public class JaegerConfig {       @Value("trace.url")       String JaggerUrl;       @Bean       public io.opentracing.Tracer jaggerTracer() {           return new Configuration("ios", new         Configuration.SamplerConfiguration(ProbabilisticSampler.TYPE, 1,         JaggerUrl),         new Configuration.ReporterConfiguration()).getTracer();     } }

--

Server

Helm chart

value.yaml
 

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271

# Default values for jaeger. # Default values for jaeger. # This is a YAML-formatted file. # Jaeger values are grouped by component. Cassandra values override subchart values    provisionDataStore:   cassandra: false   elasticsearch: true    tag: 1.9.0    storage:   # allowed values (cassandra, elasticsearch)   type: elasticsearch   cassandra:     host: cassandra     port: 9042     user: user     password: password   elasticsearch:     scheme: http     host: elasticsearch     port: 9200     user: elastic     password: changeme     nodesWanOnly: false    # Begin: Override values on the Cassandra subchart to customize for Jaeger cassandra:   image:     tag: 3.11.3   persistence:     # To enable persistence, please see the documentation for the Cassandra chart     enabled: false   config:     cluster_name: jaeger     seed_size: 1     dc_name: dc1     rack_name: rack1     endpoint_snitch: GossipingPropertyFileSnitch # End: Override values on the Cassandra subchart to customize for Jaeger    # Begin: Default values for the various components of Jaeger # This chart has been based on the Kubernetes integration found in the following repo: # https://github.com/jaegertracing/jaeger-kubernetes/blob/master/production/jaeger-production-template.yml # # This is the jaeger-cassandra-schema Job which sets up the Cassandra schema for # use by Jaeger schema:   annotations: {}   image: jaegertracing/jaeger-cassandra-schema   pullPolicy: IfNotPresent   # Acceptable values are test and prod. Default is for production use.   mode: prod   resources: {}     # limits:     #   cpu: 500m     #   memory: 512Mi     # requests:     #   cpu: 256m     #   memory: 128Mi   ## Additional pod labels   ## ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/   podLabels: {}    # Begin: Override values on the Elasticsearch subchart to customize for Jaeger elasticsearch:   image:     tag: "5.4"   cluster:     name: "tracing"   data:     persistence:       enabled: false   rbac:     create: false    agent:   enabled: true   annotations: {}   image: jaegertracing/jaeger-agent   pullPolicy: IfNotPresent   collector:     host: null     port: null   cmdlineParams: {}   daemonset:     useHostPort: false   service:     annotations: {}     # List of IP ranges that are allowed to access the load balancer (if supported)     loadBalancerSourceRanges: []     type: LoadBalancer     # zipkinThriftPort :accept zipkin.thrift over compact thrift protocol     zipkinThriftPort: 5775     # compactPort: accept jaeger.thrift over compact thrift protocol     compactPort: 6831     # binaryPort: accept jaeger.thrift over binary thrift protocol     binaryPort: 6832     # samplingPort: (HTTP) serve configs, sampling strategies     samplingPort: 5778   resources: {}     # limits:     #   cpu: 500m     #   memory: 512Mi     # requests:     #   cpu: 256m     #   memory: 128Mi   nodeSelector: {}   podAnnotations: {}   ## Additional pod labels   ## ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/   podLabels: {}   ## Allow the scheduling on tainted nodes (requires Kubernetes >= 1.6)   ## Ref: https://kubernetes.io/docs/concepts/configuration/taint-and-toleration/   tolerations: []   useHostNetwork: false   dnsPolicy: ClusterFirst    collector:   enabled: true   annotations: {}   image: jaegertracing/jaeger-collector   pullPolicy: IfNotPresent   dnsPolicy: ClusterFirst   cmdlineParams: {}   replicaCount: 1   service:     annotations: {}     # List of IP ranges that are allowed to access the load balancer (if supported)     loadBalancerSourceRanges: []     type: ClusterIP     # tchannelPort: used by jaeger-agent to send spans in jaeger.thrift format     tchannelPort: 14267     # httpPort: can accept spans directly from clients in jaeger.thrift format     httpPort: 14268     # can accept Zipkin spans in JSON or Thrift     zipkinPort: 9411     healthCheckPort: 14269   resources: {}     # limits:     #   cpu: 1     #   memory: 1Gi     # requests:     #   cpu: 500m     #   memory: 512Mi   nodeSelector: {}   podAnnotations: {}   ## Additional pod labels   ## ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/   podLabels: {}   ## Allow the scheduling on tainted nodes (requires Kubernetes >= 1.6)   ## Ref: https://kubernetes.io/docs/concepts/configuration/taint-and-toleration/   tolerations: []    query:   enabled: true   annotations: {}   image: jaegertracing/jaeger-query   pullPolicy: IfNotPresent   dnsPolicy: ClusterFirst   cmdlineParams: {}   healthCheckPort: 16687   basePath: /   replicaCount: 1   service:     annotations: {}     type: ClusterIP     # List of IP ranges that are allowed to access the load balancer (if supported)     loadBalancerSourceRanges: []     # queryPort: externally accessible port for UI and API     queryPort: 80     # targetPort: the internal port the UI and API are exposed on     targetPort: 16686   ingress:     enabled: false     annotations: {}     # Used to create an Ingress record.     # hosts:     #   - chart-example.local     # annotations:       # kubernetes.io/ingress.class: nginx       # kubernetes.io/tls-acme: "true"     # tls:       # Secrets must be manually created in the namespace.       # - secretName: chart-example-tls       #   hosts:       #     - chart-example.local   resources: {}     # limits:     #   cpu: 500m     #   memory: 512Mi     # requests:     #    cpu: 256m     #    memory: 128Mi   nodeSelector: {}   podAnnotations: {}   ## Additional pod labels   ## ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/   podLabels: {}   ## Allow the scheduling on tainted nodes (requires Kubernetes >= 1.6)   ## Ref: https://kubernetes.io/docs/concepts/configuration/taint-and-toleration/   tolerations: []    spark:   enabled: false   annotations: {}   image: jaegertracing/spark-dependencies   tag: latest   pullPolicy: Always   schedule: "49 23 * * *"   successfulJobsHistoryLimit: 5   failedJobsHistoryLimit: 5   resources: {}     # limits:     #   cpu: 500m     #   memory: 512Mi     # requests:     #   cpu: 256m     #   memory: 128Mi   nodeSelector: {}   ## Additional pod labels   ## ref: https://kubernetes.io/docs/concepts/overview/working-with-objects/labels/   podLabels: {}   ## Allow the scheduling on tainted nodes (requires Kubernetes >= 1.6)   ## Ref: https://kubernetes.io/docs/concepts/configuration/taint-and-toleration/   tolerations: [] # End: Default values for the various components of Jaeger    hotrod:   enabled: false   replicaCount: 1   image:     repository: jaegertracing/example-hotrod     tag: latest     pullPolicy: Always   service:     annotations: {}     name: hotrod     type: ClusterIP     # List of IP ranges that are allowed to access the load balancer (if supported)     loadBalancerSourceRanges: []     externalPort: 80     internalPort: 8080   ingress:     enabled: false     # Used to create Ingress record (should be used with service.type: ClusterIP).     hosts:       - chart-example.local     annotations: {}       # kubernetes.io/ingress.class: nginx       # kubernetes.io/tls-acme: "true"     tls:       # Secrets must be manually created in the namespace.       # - secretName: chart-example-tls       #   hosts:       #     - chart-example.local   resources: {}     # We usually recommend not to specify default resources and to leave this as a conscious     # choice for the user. This also increases chances charts run on environments with little     # resources, such as Minikube. If you do want to specify resources, uncomment the following     # lines, adjust them as necessary, and remove the curly braces after 'resources:'.     # limits:     #   cpu: 100m     #   memory: 128Mi     # requests:     #   cpu: 100m     #   memory: 128Mi   tracing:     host: null     port: 6831

local docker

?

1	docker run --rm -it --network=host jaegertracing/all-in-one

Zipikin vs Jaeger

Launguage

	Zipkin	Jaeger
C++	Unofficial	Official
C#	Official	Official
Go	Official	Official
Java	Official	Official
Python	Unofficial	Official
Ruby	Official	Unofficial
Scala	Official	Unofficial(or use the Java library)
Node.js	Official	Official
PHP	Unofficial	Unofficial

Framework & Library

zipkin의 경우 인기있는 프레임워크를 지원하고 데이터베이스 드라이버와 같은 라이브러리를 지원한다.

jagger의 경우 Open Tracinginstrumentation libraries 를 활용한 다양한 라이브러리로 구현이 가능하다

jager는 또 gRPC, Thrift 및 AWS SDK 도구가 포함되어있다.

둘다 백엔드는 카산드라와 엘라스틱서치를 지원한다.

Jaeger는 CNCF의 일부로 당연히 kubernetes환경에 적합하게 설계되어있다.

helm 차트 템플릿 과 Envoy 또는 Isito 와 같은 서비스 프록시를 Jaeger Suppoter와 함께 사용하면 컨테이너에서 호출을 추적하는 것이

용이하다.

즉 k8s 통합환경에서 분산 추적에 맞게 설계되어있다.

Zipkin은 Docker 이미지와 Java 라이브러리를 제공Jaeger와 달리 Zipkin은 수집기, 저장소, API 및 UI가 포함 된 단일 프레임워크이다.

Conclusion

production환경이 kubernetes면 무조건 jaeger를 단일 환경 및 모놀리틱 시스템에는 zipkin을 고려해볼만 하다

즉 컨테이너 환경에는 jaeger로 가야되는게 맞는거 같다.

또한 jaeger안에는 expoter로 zipkin도 있기때문에 zipkin도 통합할수 있다.

즉 jaeger를 써보고 맞지않는다면 zipkin을 고려해보면 좋겠다.