第七模塊 ：微服務監控告警Prometheus架構和實踐

時間  2020-05-26
標籤 第七 模塊 微服 監控 告警 prometheus 架構 實踐 欄目 系統架構 简体版
原文   原文鏈接

119.監控模式分類~1.mp4html

 

logging：日誌監控，Logging 的特色是，它描述一些離散的(不連續的)事件。 例如：應用經過一個滾動的文件輸出 Debug 或 Error 信息，並經過日誌收集系統，存儲到 Elasticsearch 中; 審批明細信息經過 Kafka，存儲到數據庫(BigTable)中; 又或者，特定請求的元數據信息，從服務請求中剝離出來，發送給一個異常收集服務，如 NewRelic。java

 

tracing：鏈路追蹤 ，例如skywalking、cat、zipkin專門作分佈式鏈路追蹤mysql

 

metrics：重要的數據指標、例如統計當前http的一個請求量，數據是能夠度量、累加的，數據是能夠聚合和累加的。經過對數據點進行聚合查看重要的指標，裏面包括一些計數器、測量儀器、直方圖等，還能夠react

在上面打標籤git

 

 

  • Metrics - 用於記錄可聚合的數據。例如，隊列的當前深度可被定義爲一個度量值，在元素入隊或出隊時被更新；HTTP 請求個數可被定義爲一個計數器，新請求到來時進行累加。

普羅米修斯重點屬於metrics的監控github

上面重點講下幾個工具部署的成本：golang

capEx表示研發人員開發的成本：metrics研發人員開發成功重點，elk最低，掌握skywalking須要必定的基礎web

OpEx表示運維成本，elk在運維的時候須要不斷擴容，運行成功最高redis

reaction表示當出問題的時候哪些功能可以第一時間進行告警，metries工具最高。skywalking的告警能力通常spring

出了問題具體分析問題哪些工具最有效：查看調用鏈解決問題最有效

 

Logging，Metrics 和 Tracing

Logging，Metrics 和 Tracing 有各自專一的部分。

  • Logging - 用於記錄離散的事件。例如，應用程序的調試信息或錯誤信息。它是咱們診斷問題的依據。
  • Metrics - 用於記錄可聚合的數據。例如，隊列的當前深度可被定義爲一個度量值，在元素入隊或出隊時被更新；HTTP 請求個數可被定義爲一個計數器，新請求到來時進行累加。
  • Tracing - 用於記錄請求範圍內的信息。例如，一次遠程方法調用的執行過程和耗時。它是咱們排查系統性能問題的利器。

這三者也有相互重疊的部分，以下圖所示。

經過上述信息，咱們能夠對已有系統進行分類。例如，Zipkin 專一於 tracing 領域；Prometheus 開始專一於 metrics，隨着時間推移可能會集成更多的 tracing 功能，但不太可能深刻 logging 領域； ELK，阿里雲日誌服務這樣的系統開始專一於 logging 領域，但同時也不斷地集成其餘領域的特性到系統中來，正向上圖中的圓心靠近。

三者關係的一篇論文：http://peter.bourgon.org/blog/2017/02/21/metrics-tracing-and-logging.html

關於三者關係的更詳細信息可參考 Metrics, tracing, and logging。下面咱們重點介紹下 tracing。

Tracing 的誕生

Tracing 是在90年代就已出現的技術。但真正讓該領域流行起來的仍是源於 Google 的一篇論文"Dapper, a Large-Scale Distributed Systems Tracing Infrastructure"，而另外一篇論文"Uncertainty in Aggregate Estimates from Sampled Distributed Traces"中則包含關於採樣的更詳細分析。論文發表後一批優秀的 Tracing 軟件孕育而生，比較流行的有：

  • Dapper(Google) : 各 tracer 的基礎
  • StackDriver Trace (Google)
  • Zipkin(twitter)
  • Appdash(golang)
  • 鷹眼(taobao)
  • 諦聽(盤古，阿里云云產品使用的Trace系統)
  • 雲圖(螞蟻Trace系統)
  • sTrace(神馬)
  • X-ray(aws)

分佈式追蹤系統發展很快，種類繁多，但核心步驟通常有三個：代碼埋點，數據存儲、查詢展現。

下圖是一個分佈式調用的例子，客戶端發起請求，請求首先到達負載均衡器，接着通過認證服務，計費服務，而後請求資源，最後返回結果。

opentracing1.png

數據被採集存儲後，分佈式追蹤系統通常會選擇使用包含時間軸的時序圖來呈現這個 Trace。

opentracing2.png

但在數據採集過程當中，因爲須要侵入用戶代碼，而且不一樣系統的 API 並不兼容，這就致使了若是您但願切換追蹤系統，每每會帶來較大改動。

OpenTracing

爲了解決不一樣的分佈式追蹤系統 API 不兼容的問題，誕生了 OpenTracing 規範。
OpenTracing 是一個輕量級的標準化層，它位於應用程序/類庫和追蹤或日誌分析程序之間。

+-------------+  +---------+  +----------+  +------------+
| Application | | Library | | OSS | | RPC/IPC | | Code | | Code | | Services | | Frameworks | +-------------+ +---------+ +----------+ +------------+ | | | | | | | | v v v v +------------------------------------------------------+ | OpenTracing | +------------------------------------------------------+ | | | | | | | | v v v v +-----------+ +-------------+ +-------------+ +-----------+ | Tracing | | Logging | | Metrics | | Tracing | | System A | | Framework B | | Framework C | | System D | +-----------+ +-------------+ +-------------+ +-----------+

OpenTracing 的優點

  • OpenTracing 已進入 CNCF，正在爲全球的分佈式追蹤，提供統一的概念和數據標準。
  • OpenTracing 經過提供平臺無關、廠商無關的 API，使得開發人員可以方便的添加（或更換）追蹤系統的實現。

OpenTracing 數據模型

OpenTracing 中的 Trace（調用鏈）經過歸屬於此調用鏈的 Span 來隱性的定義。
特別說明，一條 Trace（調用鏈）能夠被認爲是一個由多個 Span 組成的有向無環圖（DAG圖），Span 與 Span 的關係被命名爲 References。

例如：下面的示例 Trace 就是由8個 Span 組成：

單個 Trace 中，span 間的因果關係


        [Span A]  ←←←(the root span)
            | +------+------+ | | [Span B] [Span C] ←←←(Span C 是 Span A 的孩子節點, ChildOf) | | [Span D] +---+-------+ | | [Span E] [Span F] >>> [Span G] >>> [Span H] ↑ ↑ ↑ (Span G 在 Span F 後被調用, FollowsFrom)

有些時候，使用下面這種，基於時間軸的時序圖能夠更好的展示 Trace（調用鏈）：

單個 Trace 中，span 間的時間關係


––|–––––––|–––––––|–––––––|–––––––|–––––––|–––––––|–––––––|–> time [Span A···················································] [Span B··············································] [Span D··········································] [Span C········································] [Span E·······] [Span F··] [Span G··] [Span H··]

每一個 Span 包含如下的狀態:（譯者注：因爲這些狀態會反映在 OpenTracing API 中，因此會保留部分英文說明）

  • An operation name，操做名稱
  • A start timestamp，起始時間
  • A finish timestamp，結束時間
  • Span Tag，一組鍵值對構成的 Span 標籤集合。鍵值對中，鍵必須爲 string，值能夠是字符串，布爾，或者數字類型。
  • Span Log，一組 span 的日誌集合。
    每次 log 操做包含一個鍵值對，以及一個時間戳。

鍵值對中，鍵必須爲 string，值能夠是任意類型。
可是須要注意，不是全部的支持 OpenTracing 的 Tracer，都須要支持全部的值類型。

  • SpanContext，Span 上下文對象 (下面會詳細說明)
  • References(Span間關係)，相關的零個或者多個 Span（Span 間經過 SpanContext 創建這種關係）

每個 SpanContext 包含如下狀態：

  • 任何一個 OpenTracing 的實現，都須要將當前調用鏈的狀態（例如：trace 和 span 的 id），依賴一個獨特的 Span 去跨進程邊界傳輸
  • Baggage Items，Trace 的隨行數據，是一個鍵值對集合，它存在於 trace 中，也須要跨進程邊界傳輸

更多關於 OpenTracing 數據模型的知識，請參考 OpenTracing語義標準。

OpenTracing 實現

這篇文檔列出了全部 OpenTracing 實現。在這些實現中，比較流行的爲 Jaeger 和 Zipkin。

 

 

 

 

metrs主要用於監控告警，出了問題以後再經過鏈路追蹤或者elk來查看發現問題解決問題

metries能夠對系統層、應用層、業務層進行監控

121.Prometheus 簡介~1.mp4

時間數據庫

在t0產生數據v0，在t1產生數據v1，在t2產生數據v2，將這些數據點弄起來就是一個時間序列

infuxdb和普羅米修斯都是時間序列數據庫

122.Prometheus 架構設計~1.mp4

123.Prometheus 基本概念~1.mp4

第一種計數器：例如統計的http數目、下單數目等

測量儀器：例如當前同時在線用戶數目。磁盤使用率

直方圖：響應時間在某個區間內的分佈狀況

彙總：90%的響應時間

target：能夠是操做系統、機器、應用、服務等須要暴露metrries端點，每隔15秒中經過/metries抓取數據

應用直接採用：應用直接採集，直接在應用程序中埋點，直接使用普羅米修斯採集

第二種間接採集使用，使用exporter採集，執行的exporter以下所示，redis、apache、操做系統等

 

 

 web ui和grafan經過promql就能夠來查詢對於的數據

 

 

 

 124.【實驗】Prometheus 起步查詢實驗（上）~1.mp4

第一步

將http-simulatror導入到eclipse中

springboot集成prometheus

Maven pom.xml引入依賴

<dependency>
<groupId>io.prometheus</groupId>
<artifactId>simpleclient_spring_boot</artifactId>
</dependency>

2 啓動類引入註解

import io.prometheus.client.spring.boot.EnablePrometheusEndpoint;
import io.prometheus.client.spring.boot.EnableSpringBootMetricsCollector;
import org.springframework.boot.SpringApplication;
import org.springframework.boot.autoconfigure.SpringBootApplication;

@SpringBootApplication
@EnablePrometheusEndpoint
@EnableSpringBootMetricsCollector
public class Application {

public static void main(String[] args) {
SpringApplication.run(Application.class, args);
}
3 Controller類寫須要監控的指標，好比Counter

import io.prometheus.client.Counter; import org.springframework.web.bind.annotation.RequestMapping; import org.springframework.web.bind.annotation.RestController; import java.util.Random; @RestController public class SampleController { private static Random random = new Random(); private static final Counter requestTotal = Counter.build() .name("my_sample_counter") .labelNames("status") .help("A simple Counter to illustrate custom Counters in Spring Boot and Prometheus").register(); @RequestMapping("/endpoint") public void endpoint() { if (random.nextInt(2) > 0) { requestTotal.labels("success").inc(); } else { requestTotal.labels("error").inc(); } } } 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

4 設置springboot應用的服務名和端口，在application.properties

spring.application.name=mydemo server.port=8888

5 配置prometheus.yml

global: scrape_interval: 15s # By default, scrape targets every 15 seconds. evaluation_interval: 15s # By default, scrape targets every 15 seconds. # scrape_timeout is set to the global default (10s). # Attach these labels to any time series or alerts when communicating with # external systems (federation, remote storage, Alertmanager). external_labels: monitor: 'codelab-monitor' # Load rules once and periodically evaluate them according to the global 'evaluation_interval'. rule_files: # - "first.rules" # - "second.rules" # A scrape configuration containing exactly one endpoint to scrape: # Here it's Prometheus itself.
scrape_configs: # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config. - job_name: 'prometheus' # Override the global default and scrape targets from this job every 5 seconds. scrape_interval: 5s # metrics_path defaults to '/metrics' # scheme defaults to 'http'. static_configs: - targets: ['localhost:9090'] - job_name: 'mydemo' # Override the global default and scrape targets from this job every 5 seconds. scrape_interval: 5s metrics_path: '/prometheus' # scheme defaults to 'http'. static_configs: - targets: ['10.94.20.52:8888']

最關鍵的配置就是targets: [‘10.94.20.52:8888’]，就是springboot應用的ip和端口

注：在application.xml裏設置屬性：spring.metrics.servo.enabled=false，去掉重複的metrics，否則在prometheus的控制檯的targets頁籤裏，會一直顯示此endpoint爲down狀態。

 

pom.xml

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>

    <groupId>io.spring2go.promdemo</groupId>
    <artifactId>http-simulator</artifactId>
    <version>0.0.1-SNAPSHOT</version>
    <packaging>jar</packaging>

    <name>http-simulator</name>
    <description>Demo project for Spring Boot</description>

    <parent>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-parent</artifactId>
        <version>1.5.17.RELEASE</version>
        <relativePath/> <!-- lookup parent from repository -->
    </parent>

    <properties>
        <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
        <project.reporting.outputEncoding>UTF-8</project.reporting.outputEncoding>
        <java.version>1.8</java.version>
    </properties>

    <dependencies>
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-web</artifactId>
        </dependency>
        
        <!-- The prometheus client -->
       <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient_spring_boot</artifactId> <version>0.5.0</version>
        </dependency>

        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-test</artifactId>
            <scope>test</scope>
        </dependency>
    </dependencies>

    <build>
        <plugins>
            <plugin>
                <groupId>org.springframework.boot</groupId>
                <artifactId>spring-boot-maven-plugin</artifactId>
            </plugin>
        </plugins>
    </build>


</project>

 啓動類：

package io.spring2go.promdemo.httpsimulator; import org.springframework.beans.factory.annotation.Autowired; import org.springframework.boot.CommandLineRunner; import org.springframework.boot.SpringApplication; import org.springframework.boot.autoconfigure.SpringBootApplication; import org.springframework.context.ApplicationListener; import org.springframework.context.annotation.Bean; import org.springframework.context.event.ContextClosedEvent; import org.springframework.core.task.SimpleAsyncTaskExecutor; import org.springframework.core.task.TaskExecutor; import org.springframework.stereotype.Controller; import org.springframework.web.bind.annotation.PathVariable; import org.springframework.web.bind.annotation.RequestMapping; import org.springframework.web.bind.annotation.RequestMethod; import org.springframework.web.bind.annotation.ResponseBody; import io.prometheus.client.spring.boot.EnablePrometheusEndpoint; @Controller @SpringBootApplication @EnablePrometheusEndpoint public class HttpSimulatorApplication implements ApplicationListener<ContextClosedEvent> { @Autowired private SimulatorOpts opts; private ActivitySimulator simulator; public static void main(String[] args) { SpringApplication.run(HttpSimulatorApplication.class, args); } @RequestMapping(value = "/opts") public @ResponseBody String getOps() { return opts.toString(); } @RequestMapping(value = "/spike/{mode}", method = RequestMethod.POST) public @ResponseBody String setSpikeMode(@PathVariable("mode") String mode) { boolean result = simulator.setSpikeMode(mode); if (result) { return "ok"; } else { return "wrong spike mode " + mode; } } @RequestMapping(value = "error_rate/{error_rate}", method = RequestMethod.POST) public @ResponseBody String setErrorRate(@PathVariable("error_rate") int errorRate) { simulator.setErrorRate(errorRate); return "ok"; } @Bean public TaskExecutor taskExecutor() { return new SimpleAsyncTaskExecutor(); } @Bean public CommandLineRunner schedulingRunner(TaskExecutor executor) { return new CommandLineRunner() { public void run(String... args) throws Exception { simulator = new ActivitySimulator(opts); executor.execute(simulator); System.out.println("Simulator thread started..."); } }; } @Override public void onApplicationEvent(ContextClosedEvent event) { simulator.shutdown(); System.out.println("Simulator shutdown..."); } }

 

 application.properties

management.security.enabled=false opts.endpoints=/login, /login, /login, /login, /login, /login, /login, /users, /users, /users, /users/{id}, /register, /register, /logout, /logout, /logout, /logout opts.request_rate=1000 opts.request_rate_uncertainty=70 opts.latency_min=10 opts.latency_p50=25 opts.latency_p90=150 opts.latency_p99=750 opts.latency_max=10000 opts.latency_uncertainty=70 opts.error_rate=1 opts.spike_start_chance=5 opts.spike_end_chance=30

 

最關鍵的核心類

ActivitySimulator

package io.spring2go.promdemo.httpsimulator; import java.util.Random; import io.prometheus.client.Counter; import io.prometheus.client.Histogram; public class ActivitySimulator implements Runnable { private SimulatorOpts opts; private Random rand = new Random(); private boolean spikeMode = false; private volatile boolean shutdown = false; private final Counter httpRequestsTotal = Counter.build() .name("http_requests_total") .help("Total number of http requests by response status code") .labelNames("endpoint", "status") .register(); private final Histogram httpRequestDurationMs = Histogram.build() .name("http_request_duration_milliseconds") .help("Http request latency histogram") .exponentialBuckets(25, 2, 7) .labelNames("endpoint", "status") .register(); public ActivitySimulator(SimulatorOpts opts) { this.opts = opts; System.out.println(opts); } public void shutdown() { this.shutdown = true; } public void updateOpts(SimulatorOpts opts) { this.opts = opts; } public boolean setSpikeMode(String mode) { boolean result = true; switch (mode) { case "on": opts.setSpikeMode(SpikeMode.ON); System.out.println("Spike mode is set to " + mode); break; case "off": opts.setSpikeMode(SpikeMode.OFF); System.out.println("Spike mode is set to " + mode); break; case "random": opts.setSpikeMode(SpikeMode.RANDOM); System.out.println("Spike mode is set to " + mode); break; default: result = false; System.out.println("Can't recognize spike mode " + mode); } return result; } public void setErrorRate(int rate) { if (rate > 100) { rate = 100; } if (rate < 0) { rate = 0; } opts.setErrorRate(rate); System.out.println("Error rate is set to " + rate); } public SimulatorOpts getOpts() { return this.opts; } public void simulateActivity() { int requestRate = this.opts.getRequestRate(); if (this.giveSpikeMode()) { requestRate *= (5 + this.rand.nextInt(10)); } int nbRequests = this.giveWithUncertainty(requestRate, this.opts.getRequestRateUncertainty()); for (int i = 0; i < nbRequests; i++) { String statusCode = this.giveStatusCode(); String endpoint = this.giveEndpoint(); this.httpRequestsTotal.labels(endpoint, statusCode).inc(); int latency = this.giveLatency(statusCode); if (this.spikeMode) { latency *= 2; } this.httpRequestDurationMs.labels(endpoint, statusCode).observe(latency); } } public boolean giveSpikeMode() { switch (this.opts.getSpikeMode()) { case ON: this.spikeMode = true; break; case OFF: this.spikeMode = false; break; case RANDOM: int n = rand.nextInt(100); if (!this.spikeMode && n < this.opts.getSpikeStartChance()) { this.spikeMode = true; } else if (this.spikeMode && n < this.opts.getSpikeEndChance()) { this.spikeMode = false; } break; } return this.spikeMode; } public int giveWithUncertainty(int n, int u) { int delta = this.rand.nextInt(n * u / 50) - (n * u / 100); return n + delta; } public String giveStatusCode() { if (this.rand.nextInt(100) < this.opts.getErrorRate()) { return "500"; } else { return "200"; } } public String giveEndpoint() { int n = this.rand.nextInt(this.opts.getEndopints().length); return this.opts.getEndopints()[n]; } public int giveLatency(String statusCode) { if (!"200".equals(statusCode)) { return 5 + this.rand.nextInt(50); } int p = this.rand.nextInt(100); if (p < 50) { return this.giveWithUncertainty(this.opts.getLatencyMin() + this.rand.nextInt(this.opts.getLatencyP50() - this.opts.getLatencyMin()), this.opts.getLatencyUncertainty()); } if (p < 90) { return this.giveWithUncertainty(this.opts.getLatencyP50() + this.rand.nextInt(this.opts.getLatencyP90() - this.opts.getLatencyP50()), this.opts.getLatencyUncertainty()); } if (p < 99) { return this.giveWithUncertainty(this.opts.getLatencyP90() + this.rand.nextInt(this.opts.getLatencyP99() - this.opts.getLatencyP90()), this.opts.getLatencyUncertainty()); } return this.giveWithUncertainty(this.opts.getLatencyP99() + this.rand.nextInt(this.opts.getLatencyMax() - this.opts.getLatencyP99()), this.opts.getLatencyUncertainty()); } @Override public void run() { while(!shutdown) { System.out.println("Simulator is running..."); this.simulateActivity(); try { Thread.sleep(1000); } catch (InterruptedException e) { // TODO Auto-generated catch block
 e.printStackTrace(); } } } }

 SimulatorOpts

 

package io.spring2go.promdemo.httpsimulator; import java.util.Arrays; import org.springframework.beans.factory.annotation.Value; import org.springframework.context.annotation.Configuration; import com.fasterxml.jackson.annotation.JsonAutoDetect; @Configuration @JsonAutoDetect(fieldVisibility = JsonAutoDetect.Visibility.ANY) public class SimulatorOpts { // Endpoints, Weighted map of endpoints to simulate
    @Value("${opts.endpoints}") private String[] endopints; // RequestRate, requests per second
    @Value("${opts.request_rate}") private int requestRate; // RequestRateUncertainty, Percentage of uncertainty when generating requests (+/-)
    @Value("${opts.request_rate_uncertainty}") private int requestRateUncertainty; // LatencyMin in milliseconds
    @Value("${opts.latency_min}") private int latencyMin; // LatencyP50 in milliseconds
    @Value("${opts.latency_p50}") private int latencyP50; // LatencyP90 in milliseconds
    @Value("${opts.latency_p90}") private int latencyP90; // LatencyP99 in milliseconds
    @Value("${opts.latency_p99}") private int latencyP99; // LatencyMax in milliseconds
    @Value("${opts.latency_max}") private int latencyMax; // LatencyUncertainty, Percentage of uncertainty when generating latency (+/-)
    @Value("${opts.latency_uncertainty}") private int latencyUncertainty; // ErrorRate, Percentage of chance of requests causing 500
    @Value("${opts.error_rate}") private int errorRate; // SpikeStartChance, Percentage of chance of entering spike mode
    @Value("${opts.spike_start_chance}") private int spikeStartChance; // SpikeStartChance, Percentage of chance of exiting spike mode
    @Value("${opts.spike_end_chance}") private int spikeEndChance; // SpikeModeStatus ON/OFF/RANDOM
    private SpikeMode spikeMode = SpikeMode.OFF; public String[] getEndopints() { return endopints; } public void setEndopints(String[] endopints) { this.endopints = endopints; } public int getRequestRate() { return requestRate; } public void setRequestRate(int requestRate) { this.requestRate = requestRate; } public int getRequestRateUncertainty() { return requestRateUncertainty; } public void setRequestRateUncertainty(int requestRateUncertainty) { this.requestRateUncertainty = requestRateUncertainty; } public int getLatencyMin() { return latencyMin; } public void setLatencyMin(int latencyMin) { this.latencyMin = latencyMin; } public int getLatencyP50() { return latencyP50; } public void setLatencyP50(int latencyP50) { this.latencyP50 = latencyP50; } public int getLatencyP90() { return latencyP90; } public void setLatencyP90(int latencyP90) { this.latencyP90 = latencyP90; } public int getLatencyP99() { return latencyP99; } public void setLatencyP99(int latencyP99) { this.latencyP99 = latencyP99; } public int getLatencyMax() { return latencyMax; } public void setLatencyMax(int latencyMax) { this.latencyMax = latencyMax; } public int getLatencyUncertainty() { return latencyUncertainty; } public void setLatencyUncertainty(int latencyUncertainty) { this.latencyUncertainty = latencyUncertainty; } public int getErrorRate() { return errorRate; } public void setErrorRate(int errorRate) { this.errorRate = errorRate; } public int getSpikeStartChance() { return spikeStartChance; } public void setSpikeStartChance(int spikeStartChance) { this.spikeStartChance = spikeStartChance; } public int getSpikeEndChance() { return spikeEndChance; } public void setSpikeEndChance(int spikeEndChance) { this.spikeEndChance = spikeEndChance; } public SpikeMode getSpikeMode() { return spikeMode; } public void setSpikeMode(SpikeMode spikeMode) { this.spikeMode = spikeMode; } @Override public String toString() { return "SimulatorOpts [endopints=" + Arrays.toString(endopints) + ", requestRate=" + requestRate + ", requestRateUncertainty=" + requestRateUncertainty + ", latencyMin=" + latencyMin + ", latencyP50="
                + latencyP50 + ", latencyP90=" + latencyP90 + ", latencyP99=" + latencyP99 + ", latencyMax="
                + latencyMax + ", latencyUncertainty=" + latencyUncertainty + ", errorRate=" + errorRate + ", spikeStartChance=" + spikeStartChance + ", spikeEndChance=" + spikeEndChance + ", spikeMode="
                + spikeMode + "]"; } }

 

SpikeMode

package io.spring2go.promdemo.httpsimulator; public enum SpikeMode { OFF, ON, RANDOM }

 

咱們將應用運行起來

模擬一個簡單的HTTP微服務，生成Prometheus Metrics，能夠Spring Boot方式運行

Metrics

運行時訪問端點：

http://SERVICE_URL:8080/prometheus

包括：

  • http_requests_total：請求計數器，endpoint和status爲label
  • http_request_duration_milliseconds：請求延遲分佈(histogram)

運行時options

Spike Mode

在Spike模式下，請求數會乘以一個因子(5~15)，延遲加倍

Spike模式能夠是on, off或者random, 改變方式：

# ON
curl -X POST http://SERVICE_URL:8080/spike/on

# OFF
curl -X POST http://SERVICE_URL:8080/spike/off

# RANDOM
curl -X POST http://SERVICE_URL:8080/spike/random

Error rate

缺省錯誤率1%，能夠調整(0~100)，方法：

# Setting error to 50%
curl -X POST http://SERVICE_URL:8080/error_rate/50

其它參數

配置在application.properties中

opts.endpoints=/login, /login, /login, /login, /login, /login, /login, /users, /users, /users, /users/{id}, /register, /register, /logout, /logout, /logout, /logout
opts.request_rate=1000
opts.request_rate_uncertainty=70
opts.latency_min=10
opts.latency_p50=25
opts.latency_p90=150
opts.latency_p99=750
opts.latency_max=10000
opts.latency_uncertainty=70

opts.error_rate=1
opts.spike_start_chance=5
opts.spike_end_chance=30

運行時校驗端點：

http://SERVICE_URL:8080/opts

參考

https://github.com/PierreVincent/prom-http-simulator

 

咱們在瀏覽器輸入http://localhost:8080/prometheus，咱們能夠查看收集到了metrie信息

# HELP http_requests_total Total number of http requests by response status code # TYPE http_requests_total counter http_requests_total{endpoint="/login",status="500",} 188.0 http_requests_total{endpoint="/register",status="500",} 55.0 http_requests_total{endpoint="/login",status="200",} 18863.0 http_requests_total{endpoint="/register",status="200",} 5425.0 http_requests_total{endpoint="/users/{id}",status="500",} 26.0 http_requests_total{endpoint="/users/{id}",status="200",} 2663.0 http_requests_total{endpoint="/logout",status="200",} 10722.0 http_requests_total{endpoint="/users",status="200",} 8034.0 http_requests_total{endpoint="/users",status="500",} 94.0 http_requests_total{endpoint="/logout",status="500",} 93.0 # HELP http_request_duration_milliseconds Http request latency histogram # TYPE http_request_duration_milliseconds histogram http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="25.0",} 85.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="50.0",} 174.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="100.0",} 188.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="200.0",} 188.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="400.0",} 188.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="800.0",} 188.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="1600.0",} 188.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="+Inf",} 188.0 http_request_duration_milliseconds_count{endpoint="/login",status="500",} 188.0 http_request_duration_milliseconds_sum{endpoint="/login",status="500",} 5499.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="25.0",} 27.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="50.0",} 50.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="100.0",} 55.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="200.0",} 55.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="400.0",} 55.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="800.0",} 55.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="1600.0",} 55.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="500",le="+Inf",} 55.0 http_request_duration_milliseconds_count{endpoint="/register",status="500",} 55.0 http_request_duration_milliseconds_sum{endpoint="/register",status="500",} 1542.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="25.0",} 8479.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="50.0",} 11739.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="100.0",} 14454.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="200.0",} 17046.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="400.0",} 17882.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="800.0",} 18482.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="1600.0",} 18705.0 http_request_duration_milliseconds_bucket{endpoint="/login",status="200",le="+Inf",} 18863.0 http_request_duration_milliseconds_count{endpoint="/login",status="200",} 18863.0 http_request_duration_milliseconds_sum{endpoint="/login",status="200",} 2552014.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="25.0",} 2388.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="50.0",} 3367.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="100.0",} 4117.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="200.0",} 4889.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="400.0",} 5136.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="800.0",} 5310.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="1600.0",} 5379.0 http_request_duration_milliseconds_bucket{endpoint="/register",status="200",le="+Inf",} 5425.0 http_request_duration_milliseconds_count{endpoint="/register",status="200",} 5425.0 http_request_duration_milliseconds_sum{endpoint="/register",status="200",} 739394.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="25.0",} 14.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="50.0",} 25.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="100.0",} 26.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="200.0",} 26.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="400.0",} 26.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="800.0",} 26.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="1600.0",} 26.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="500",le="+Inf",} 26.0 http_request_duration_milliseconds_count{endpoint="/users/{id}",status="500",} 26.0 http_request_duration_milliseconds_sum{endpoint="/users/{id}",status="500",} 752.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="25.0",} 1220.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="50.0",} 1657.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="100.0",} 2030.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="200.0",} 2383.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="400.0",} 2508.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="800.0",} 2608.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="1600.0",} 2637.0 http_request_duration_milliseconds_bucket{endpoint="/users/{id}",status="200",le="+Inf",} 2663.0 http_request_duration_milliseconds_count{endpoint="/users/{id}",status="200",} 2663.0 http_request_duration_milliseconds_sum{endpoint="/users/{id}",status="200",} 402375.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="25.0",} 4790.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="50.0",} 6634.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="100.0",} 8155.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="200.0",} 9609.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="400.0",} 10113.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="800.0",} 10493.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="1600.0",} 10622.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="200",le="+Inf",} 10722.0 http_request_duration_milliseconds_count{endpoint="/logout",status="200",} 10722.0 http_request_duration_milliseconds_sum{endpoint="/logout",status="200",} 1502959.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="25.0",} 3622.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="50.0",} 4967.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="100.0",} 6117.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="200.0",} 7254.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="400.0",} 7624.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="800.0",} 7866.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="1600.0",} 7966.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="200",le="+Inf",} 8034.0 http_request_duration_milliseconds_count{endpoint="/users",status="200",} 8034.0 http_request_duration_milliseconds_sum{endpoint="/users",status="200",} 1100809.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="25.0",} 41.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="50.0",} 88.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="100.0",} 94.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="200.0",} 94.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="400.0",} 94.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="800.0",} 94.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="1600.0",} 94.0 http_request_duration_milliseconds_bucket{endpoint="/users",status="500",le="+Inf",} 94.0 http_request_duration_milliseconds_count{endpoint="/users",status="500",} 94.0 http_request_duration_milliseconds_sum{endpoint="/users",status="500",} 2685.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="25.0",} 41.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="50.0",} 85.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="100.0",} 93.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="200.0",} 93.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="400.0",} 93.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="800.0",} 93.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="1600.0",} 93.0 http_request_duration_milliseconds_bucket{endpoint="/logout",status="500",le="+Inf",} 93.0 http_request_duration_milliseconds_count{endpoint="/logout",status="500",} 93.0 http_request_duration_milliseconds_sum{endpoint="/logout",status="500",} 2683.0

 咱們來看具體的代碼

metric有Counter、Gauge、Histogram和Summary四種類型

this.httpRequestsTotal.labels(endpoint, statusCode).inc();這裏是建立一個計數器，統計http請求的信息

對不一樣的http請求的端點，改端點下不一樣的響應都會進行記錄

http_requests_total{endpoint="/login",status="500",} 188.0
http_requests_total{endpoint="/register",status="500",} 55.0
http_requests_total{endpoint="/login",status="200",} 18863.0

= Histogram.build()
.name("http_request_duration_milliseconds")
.help("Http request latency histogram")
.exponentialBuckets(25, 2, 7)
.labelNames("endpoint", "status")
.register();

這裏是建立一個直方圖，用來統計延遲數據分佈

http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="25.0",} 85.0
http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="50.0",} 174.0
http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="100.0",} 188.0
http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="200.0",} 188.0
http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="400.0",} 188.0
http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="800.0",} 188.0
http_request_duration_milliseconds_bucket{endpoint="/login",status="500",le="1600.0",} 188.0

 125.【實驗】Prometheus起步查詢實驗（中）~1.mp4

首先安裝普羅米修斯

接下來咱們要讓普羅米修斯抓取咱們上面的springboot http-simulation的數據http://localhost:8080/，須要修改prometheus.ym

# my global config global: scrape_interval: 5s # Set the scrape interval to every 15 seconds. Default is every 1 minute. evaluation_interval: 5s # Evaluate rules every 15 seconds. The default is every 1 minute. # scrape_timeout is set to the global default (10s). # Alertmanager configuration alerting: alertmanagers: - static_configs: - targets: # - alertmanager:9093 # Load rules once and periodically evaluate them according to the global 'evaluation_interval'. rule_files: # - "first_rules.yml" # - "second_rules.yml" # A scrape configuration containing exactly one endpoint to scrape: # Here it's Prometheus itself.
scrape_configs: # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config. - job_name: 'prometheus' # metrics_path defaults to '/metrics' # scheme defaults to 'http'. static_configs: - targets: ['localhost:9090'] - job_name: 'http-simulation' metrics_path: /prometheus # scheme defaults to 'http'. static_configs: - targets: ['localhost:8080']

 建立一個新的job，job的名稱是'http-simulation'，業務metrics的路徑是/prometheus，對於的tartgets是localhost:8080'

 下面咱們將普羅米修斯運行起來

咱們使用git bash窗口將普羅米修斯運行起來。這裏不要使用windows的cmd窗口

啓動成功以後，瀏覽器輸入http://localhost:9090/graph

點擊查看target能夠看到當前監控了哪些target

能夠看到當前監控了哪些實例

咱們要查看http-simulation的請求數目，如何作的統計了:http_requests_total{job="http-simulation"}

對於統計咱們能夠點擊上方的graph，查看一個圖形的一個統計，查看的時間是能夠手動修改的

校驗http-simulator在1狀態 up{job="http-simulator"} 查詢http請求數 http_requests_total{job="http-simulator"} 查詢成功login請求數 http_requests_total{job="http-simulator", status="200", endpoint="/login"} 查詢成功請求數，以endpoint區分 http_requests_total{job="http-simulator", status="200"} 查詢總成功請求數 sum(http_requests_total{job="http-simulator", status="200"}) 查詢成功請求率，以endpoint區分 rate(http_requests_total{job="http-simulator", status="200"}[5m]) 查詢總成功請求率 sum(rate(http_requests_total{job="http-simulator", status="200"}[5m]))

 

 

 http_requests_total{job="http-simulator", status="200", endpoint="/login"}
126.【實驗】Prometheus起步查詢實驗（下）~1.mp4
4. 延遲分佈(Latency distribution)查詢 查詢http-simulator延遲分佈 http_request_duration_milliseconds_bucket{job="http-simulator"} 查詢成功login延遲分佈 http_request_duration_milliseconds_bucket{job="http-simulator", status="200", endpoint="/login"} 不超過200ms延遲的成功login請求佔比 sum(http_request_duration_milliseconds_bucket{job="http-simulator", status="200", endpoint="/login", le="200.0"}) / sum(http_request_duration_milliseconds_count{job="http-simulator", status="200", endpoint="/login"}) 成功login請求延遲的99百分位 histogram_quantile(0.99, rate(http_request_duration_milliseconds_bucket{job="http-simulator", status="200", endpoint="/log

 

127.【實驗】Prometheus + Grafana 展現實驗（上）~1.mp4

安裝grafana，下載以後解壓完成以後就能夠了

運行grafana以前，保證普羅米修斯以及咱們要監控的應用已經正常啓動起來了

咱們使用git bash窗口運行

密碼是admin/admin

登陸成功以後，咱們要給grafana設置普羅米修斯的數據源

直接點擊add data source

其餘參數默認不作修改

添加Proemethes數據源

Name -> prom-datasource
Type -> Prometheus
HTTP URL -> http://localhost:9090
其它缺省便可

 

http://localhost:9090是普羅米修斯運行的程序端口

Save & Test確保鏈接成功

3. 建立一個Dashboard
點擊**+圖標建立一個Dashbaord，點擊保存**圖標保存Dashboard，使用缺省Folder，給Dashboard起名爲prom-demo。

 

 

4. 展現請求率

點擊Add panel圖標，點擊Graph圖標添加一個Graph，

點擊Graph上的Panel Title->Edit進行編輯

修改Title：General -> Title = Request Rate

設置Metrics

sum(rate(http_requests_total{job="http-simulator"}[5m]))

調整Lagend

  • 以表格展現As Table
  • 顯示Min/Max/Avg/Current/Total
  • 根據須要調整Axis

注意保存Dahsboard。

設置完成metries以後點擊下右邊的那個小三角

128.【實驗】Prometheus + Grafana 展現實驗（下）~1.mp4

5. 展現實時錯誤率

點擊Add panel圖標，點擊Singlestat圖標添加一個Singlestat，

點擊Graph上的Panel Title->Edit進行編輯

修改Title：General -> Title = Live Error Rate

設置Metrics

sum(rate(http_requests_total{job="http-simulator", status="500"}[5m])) / sum(rate(http_requests_total{job="http-simulator"}[5m]))

調整顯示單位unit：Options->Unit，設置爲None->percent(0.0-1.0)

調整顯示值(目前爲平均)爲當前值(now)：Options->Value->Stat，設置爲Current

添加閥值和顏色：Options->Coloring，選中Value，將Threshold設置爲0.01,0.05，表示

  • 綠色：0-1%
  • 橙色：1-5%
  • 紅色：>5%

添加測量儀效果：Options->Gauge，選中Show，並將Max設爲1

添加錯誤率演變曲線：選中Spark lines -> Show

注意保存Dahsboard。

sum(rate(http_requests_total{job="http-simulation",status="500"}[5m])) /sum(rate(http_requests_total{job="http-simulation"}[5m]))

調整顯示單位unit：Options->Unit，設置爲None->percent(0.0-1.0)

調整顯示值(目前爲平均)爲當前值(now)：Options->Value->Stat，設置爲Current

設置測量儀效果

6. 展現Top requested端點
點擊Add panel圖標，點擊Table圖標添加一個Table，

設置Metrics

sum(rate(http_requests_total{job="http-simulator"}[5m])) by (endpoint)
減小表中數據項，在Metrics下，選中Instant只顯示當前值

隱藏Time列，在Column Sytle下，Apply to columns named爲Time，將Type->Type設置爲Hidden

將Value列重命名，添加一個Column Style，Apply to columns named爲Value，將Column Header設置爲Requests/s

點擊表中的Requests/s header，讓其中數據根據端點活躍度進行排序。

注意調整Widget位置並保存Dahsboard。

sum(rate(http_requests_total{job="http-simulation"}[5m])) by (endpoint)

勾選instance只看當前存在的請求類型

點擊value能夠按照值大小進行排序，讓當前請求最大的值在前面

這樣能夠實時的查看當前的一個請求的統計。訪問最頻繁的五個端點信息等

 129.【實驗】Prometheus + Alertmanager 告警實驗（上）~1.mp4

注意啓用--web.enable-lifecycle，讓Prometheus支持經過web端點動態更新配置

接下來咱們作這樣的一個功能，當咱們以前運行的http-simulation這個應用掛了，咱們發出一個告警

. HttpSimulatorDown告警
在Prometheus目錄下:

 

添加simulator_alert_rules.yml告警配置文件

groups: - name: simulator-alert-rule rules: - alert: HttpSimulatorDown expr: sum(up{job="http-simulation"}) == 0
    for: 1m labels: severity: critical

 

一分鐘內統計 

 sum(up{job="http-simulation"})的值都是0，說明1分鐘內實例都沒有啓動，發出告警
修改，引用文件
prometheus.ymlsimulator_alert_rules.yml
# my global config global: scrape_interval: 5s # Set the scrape interval to every 15 seconds. Default is every 1 minute. evaluation_interval: 5s # Evaluate rules every 15 seconds. The default is every 1 minute. # scrape_timeout is set to the global default (10s). # Alertmanager configuration alerting: alertmanagers: - static_configs: - targets: # - alertmanager:9093 # Load rules once and periodically evaluate them according to the global 'evaluation_interval'. rule_files: - "simulator_alert_rules.yml" # - "second_rules.yml" # A scrape configuration containing exactly one endpoint to scrape: # Here it's Prometheus itself.
scrape_configs: # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config. - job_name: 'prometheus' # metrics_path defaults to '/metrics' # scheme defaults to 'http'. static_configs: - targets: ['localhost:9090'] - job_name: 'http-simulation' metrics_path: /prometheus static_configs: - targets: ['localhost:8080']

 

這樣說明配置已經成功，咱們將http-simulation應用中止，一分鐘以後會觸發報警

狀態爲firing表示告警已經觸發了

3. ErrorRateHigh告警

假設已經執行上面的步驟2，則從新運行Prometheus HTTP Metrics Simulator

在simulator_alert_rules.yml文件中增長告警配置

- alert: ErrorRateHigh expr: sum(rate(http_requests_total{job="http-simulator", status="500"}[5m])) / sum(rate(http_requests_total{job="http-simulator"}[5m])) > 0.02
    for: 1m labels: severity: major annotations: summary: "High Error Rate detected" description: "Error Rate is above 2% (current value is: {{ $value }}"

 

整個文件的內容以下
groups: - name: simulator-alert-rule rules: - alert: HttpSimulatorDown expr: sum(up{job="http-simulation"}) == 0
    for: 1m labels: severity: critical - alert: ErrorRateHigh expr: sum(rate(http_requests_total{job="http-simulator", status="500"}[5m])) / sum(rate(http_requests_total{job="http-simulator"}[5m])) > 0.02
    for: 1m labels: severity: major annotations: summary: "High Error Rate detected" description: "Error Rate is above 2% (current value is: {{ $value }}"

130.【實驗】Prometheus + Alertmanager 告警實驗（下）~1.mp4

上面咱們已經設置了告警，接下來當產生告警的時候，可以發送郵件

首先要下載下載Alertmanager 0.15.2 for Windows，並解壓到本地目錄。

配置好郵箱地址以後，要啓動

啓動Alertmanager

./alertmanager.exe

在Prometheus目錄下，修改prometheus.yml配置Alertmanager地址，默認是9093端口

# Alertmanager configuration
alerting:
alertmanagers:
- static_configs:
- targets:
- localhost:9093

經過Prometheus->Status的Configuration和Rules確認配置和告警設置生效

經過Alertmanager UI界面和設置的郵箱，校驗ErrorRateHigh告警觸發

Alertmanager UI訪問地址：

http://localhost:9093

131.【實驗】Java 應用埋點和監控實驗~1.mp4

 

咱們能夠經過http api給隊列添加job，worker進行幹活

實驗4、Java應用埋點和監控實驗
實驗步驟
1. Review和運行埋點樣例代碼
將instrumentation-example導入Eclipse IDE
Review代碼理解模擬任務系統原理和埋點方式
以Spring Boot方式運行埋點案例
經過http://localhost:8080/prometheus查看metrics

 

 

pom.xml
<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>

    <groupId>io.spring2go.promdemo</groupId>
    <artifactId>instrumentation-example</artifactId>
    <version>0.0.1-SNAPSHOT</version>
    <packaging>jar</packaging>

    <name>instrumentation-example</name>
    <description>Demo project for Spring Boot</description>

    <parent>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-parent</artifactId>
        <version>1.5.17.RELEASE</version>
        <relativePath/> <!-- lookup parent from repository -->
    </parent>

    <properties>
        <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
        <project.reporting.outputEncoding>UTF-8</project.reporting.outputEncoding>
        <java.version>1.8</java.version>
    </properties>

    <dependencies>
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-web</artifactId>
        </dependency>
        
       <!-- The prometheus client --> <dependency> <groupId>io.prometheus</groupId> <artifactId>simpleclient_spring_boot</artifactId> <version>0.5.0</version>
        </dependency>

        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-test</artifactId>
            <scope>test</scope>
        </dependency>
    </dependencies>

    <build>
        <plugins>
            <plugin>
                <groupId>org.springframework.boot</groupId>
                <artifactId>spring-boot-maven-plugin</artifactId>
            </plugin>
        </plugins>
    </build>


</project>

 

InstrumentApplication
package io.spring2go.promdemo.instrument; import org.springframework.boot.CommandLineRunner; import org.springframework.boot.SpringApplication; import org.springframework.boot.autoconfigure.SpringBootApplication; import org.springframework.context.annotation.Bean; import org.springframework.core.task.SimpleAsyncTaskExecutor; import org.springframework.core.task.TaskExecutor; import org.springframework.stereotype.Controller; import org.springframework.web.bind.annotation.RequestMapping; import org.springframework.web.bind.annotation.RequestMethod; import org.springframework.web.bind.annotation.ResponseBody; import io.prometheus.client.spring.boot.EnablePrometheusEndpoint; @Controller @SpringBootApplication @EnablePrometheusEndpoint public class InstrumentApplication { private JobQueue queue = new JobQueue(); private WorkerManager workerManager; public static void main(String[] args) { SpringApplication.run(InstrumentApplication.class, args); } @RequestMapping(value = "/hello-world") public @ResponseBody String sayHello() { return "hello, world"; } @RequestMapping(value = "/jobs", method = RequestMethod.POST) public @ResponseBody String jobs() { queue.push(new Job()); return "ok"; } @Bean public TaskExecutor taskExecutor() { return new SimpleAsyncTaskExecutor(); } @Bean public CommandLineRunner schedulingRunner(TaskExecutor executor) { return new CommandLineRunner() { public void run(String... args) throws Exception { // 10 jobs per worker
                workerManager = new WorkerManager(queue, 1, 4, 10); executor.execute(workerManager); System.out.println("WorkerManager thread started..."); } }; } }

Job

package io.spring2go.promdemo.instrument; import java.util.Random; import java.util.UUID; public class Job { private String id; private Random rand = new Random(); public Job() { this.id = UUID.randomUUID().toString(); } public void run() { try { // Run the job (5 - 15 seconds)
            Thread.sleep((5 + rand.nextInt(10)) * 1000); } catch (InterruptedException e) { // TODO Auto-generated catch block
 e.printStackTrace(); } } public String getId() { return id; } public void setId(String id) { this.id = id; } }

 

JobQueue

package io.spring2go.promdemo.instrument; import java.util.Queue; import java.util.concurrent.LinkedBlockingQueue; import io.prometheus.client.Gauge; public class JobQueue { private final Gauge jobQueueSize = Gauge.build() .name("job_queue_size") .help("Current number of jobs waiting in queue") .register(); private Queue<Job> queue = new LinkedBlockingQueue<Job>(); public int size() { return queue.size(); } public void push(Job job) { queue.offer(job);  jobQueueSize.inc(); } public Job pull() { Job job = queue.poll(); if (job != null) {  jobQueueSize.dec(); } return job; } }

 

Worker

package io.spring2go.promdemo.instrument; import java.util.UUID; import io.prometheus.client.Histogram; public class Worker extends Thread { private static final Histogram jobsCompletionDurationSeconds = Histogram.build() .name("jobs_completion_duration_seconds") .help("Histogram of job completion time") .linearBuckets(4, 1, 16) .register(); private String id; private JobQueue queue; private volatile boolean shutdown; public Worker(JobQueue queue) { this.queue = queue; this.id = UUID.randomUUID().toString(); } @Override public void run() { System.out.println(String.format("[Worker %s] Starting", this.id)); while(!shutdown) { this.pullJobAndRun(); } System.out.println(String.format("[Worker %s] Stopped", this.id)); } public void shutdown() { this.shutdown = true; System.out.println(String.format("[Worker %s] Shutting down", this.id)); } public void pullJobAndRun() { Job job = this.queue.pull(); if (job != null) { long jobStart = System.currentTimeMillis(); System.out.println(String.format("[Worker %s] Starting job: %s", this.id, job.getId())); job.run(); System.out.println(String.format("[Worker %s] Finished job: %s", this.id, job.getId())); int duration = (int)((System.currentTimeMillis() - jobStart) / 1000);  jobsCompletionDurationSeconds.observe(duration); } else { System.out.println(String.format("[Worker %s] Queue is empty. Backing off 5 seconds", this.id)); try { Thread.sleep(5 * 1000); } catch (InterruptedException e) { // TODO Auto-generated catch block
 e.printStackTrace(); } } } }

 

WorkerManager

package io.spring2go.promdemo.instrument; import java.util.LinkedList; import java.util.Queue; public class WorkerManager extends Thread { private Queue<Worker> workers = new LinkedList<Worker>(); private JobQueue queue; private int minWorkers; private int maxWorkers; private int jobsWorkerRatio; public WorkerManager(JobQueue queue, int minWorkers, int maxWorkers, int jobsWorkerRatio) { this.queue = queue; this.minWorkers = minWorkers; this.maxWorkers = maxWorkers; this.jobsWorkerRatio = jobsWorkerRatio; // Initialize workerpool
        for (int i = 0; i < minWorkers; i++) { this.addWorker(); } } public void addWorker() { Worker worker = new Worker(queue); this.workers.offer(worker); worker.start(); } public void shutdownWorker() { if (this.workers.size() > 0) { Worker worker = this.workers.poll(); worker.shutdown(); } } public void run() { this.scaleWorkers(); } public void scaleWorkers() { while(true) { int queueSize = this.queue.size(); int workerCount = this.workers.size(); if ((workerCount + 1) * jobsWorkerRatio < queueSize && workerCount < this.maxWorkers) { System.out.println("[WorkerManager] Too much work, starting extra worker."); this.addWorker(); } if ((workerCount - 1) * jobsWorkerRatio > queueSize && workerCount > this.minWorkers) { System.out.println("[WorkerManager] Too much workers, shutting down 1 worker"); this.shutdownWorker(); } try { Thread.sleep(10 * 1000); } catch (InterruptedException e) { // TODO Auto-generated catch block
 e.printStackTrace(); } } } }

application.properties

management.security.enabled=false

 

咱們要監聽兩個指標：第一job隊列的大小，二是隊列中每一個任務執行的時間
job的大小咱們在job中添加監聽，從隊列中post job的時候隊列大小增長1，取出減小1

2. 配置和運行Promethus
添加針對instrumentation-example的監控job

- job_name: 'instrumentation-example'
metrics_path: /prometheus
static_configs:
- targets: ['localhost:8080']
運行Prometheus

./prometheus.exe
經過Prometheus->Status的configuration和targets校驗配置正確

3. 生成測試數據和查詢Metrics
查詢instrumentation-example在UP1狀態

up{job="instrumentation-example"}
運行queueUpJobs.sh產生100個job

./queueUpJobs.sh
查詢JobQueueSize變化曲線(調整時間範圍到5m)：

job_queue_size{job="instrumentation-example"}
查詢90分位Job執行延遲分佈：

histogram_quantile(0.90, rate(jobs_completion_duration_seconds_bucket{job="instrumentation-example"}[5m]))

132.【實驗】NodeExporter 系統監控實驗~1.mp4

1. 下載和運行wmi-exporter
下載wmi_exporter-amd64，並解壓到本地目錄

 

校驗metrics端點

http://localhost:9182/metrics

2. 配置和運行Promethus
在Prometheus安裝目錄下

在prometheus.yml 中添加針對wmi-exporter的監控job

# my global config global: scrape_interval: 5s # Set the scrape interval to every 15 seconds. Default is every 1 minute. evaluation_interval: 5s # Evaluate rules every 15 seconds. The default is every 1 minute. # scrape_timeout is set to the global default (10s). # Alertmanager configuration alerting: alertmanagers: - static_configs: - targets: # - alertmanager:9093 # Load rules once and periodically evaluate them according to the global 'evaluation_interval'. rule_files: - "simulator_alert_rules.yml" # - "second_rules.yml" # A scrape configuration containing exactly one endpoint to scrape: # Here it's Prometheus itself.
scrape_configs:  # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config. - job_name: 'wmi-exporter' # metrics_path defaults to '/metrics' # scheme defaults to 'http'. static_configs: - targets: ['localhost:9182']

 

3. Grafana Dashboard for wmi-exporter
在Grafana安裝目錄下啓動Grafana服務器

./bin/grafana-server.exe
登陸Grafana UI(admin/admin)

http://localhost:3000
經過Grafana的**+**圖標導入(Import) wmi-exporter dashboard：

grafana id = 2129
注意選中prometheus數據源
查看Windows Node dashboard。

 

4. 參考

Grafana Dashboard倉庫

https://grafana.com/dashboards

 

而後輸入2129，會自動導入變成下面的信息

 

這裏數據源要選擇普羅米修斯數據源

133.【實驗】Spring Boot Actuator 監控實驗~1.mp4

實驗步驟
1. 運行Spring Boot + Actuator
將actuatordemo應用導入Eclipse IDE

Review actuatordemo代碼

以Spring Boot方式運行actuatordemo

校驗metrics端點

http://localhost:8080/prometheus
2. 配置和運行Promethus
在Prometheus安裝目錄下

在prometheus.yml 中添加針對wmi-exporter的監控job

- job_name: 'actuator-demo'
metrics_path: '/prometheus'
static_configs:
- targets: ['localhost:8080']
運行Prometheus

./prometheus.exe
訪問Prometheus UI

http://localhost:9090
經過Prometheus->Status的configuration和targets校驗配置正確

經過Prometheus->Graph查詢actuator-demo在UP=1狀態

up{job="actuatordemo"}
3. Grafana Dashboard for JVM (Micrometer)
在Grafana安裝目錄下啓動Grafana服務器

./bin/grafana-server.exe
登陸Grafana UI(admin/admin)

http://localhost:3000
經過Grafana的**+**圖標導入(Import) JVM (Micrometer) dashboard：

grafana id = 4701
注意選中prometheus數據源
查看JVM (Micormeter) dashboard。

4. 參考
Grafana Dashboard倉庫

https://grafana.com/dashboards
Micrometer Prometheus支持

https://micrometer.io/docs/registry/prometheus
Micrometer Springboot 1.5支持

https://micrometer.io/docs/ref/spring/1.5

 133.【實驗】Spring Boot Actuator 監控實驗~1.mp4

整個程序的代碼以下

首先咱們來看下代碼

<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>

    <groupId>io.spring2go.promdemo</groupId>
    <artifactId>actuatordemo</artifactId>
    <version>0.0.1-SNAPSHOT</version>
    <packaging>jar</packaging>

    <name>actuatordemo</name>
    <description>Demo project for Spring Boot</description>

    <parent>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-parent</artifactId>
        <version>1.5.17.RELEASE</version>
        <relativePath/> <!-- lookup parent from repository -->
    </parent>

    <properties>
        <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
        <project.reporting.outputEncoding>UTF-8</project.reporting.outputEncoding>
        <java.version>1.8</java.version>
    </properties>

    <dependencies>
        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-actuator</artifactId>
        </dependency>

        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-web</artifactId>
        </dependency>

       <dependency> <groupId>io.micrometer</groupId> <artifactId>micrometer-spring-legacy</artifactId> <version>1.0.6</version> </dependency> <dependency> <groupId>io.micrometer</groupId> <artifactId>micrometer-registry-prometheus</artifactId> <version>1.0.6</version> </dependency> <dependency> <groupId>io.github.mweirauch</groupId> <artifactId>micrometer-jvm-extras</artifactId>
            <version>0.1.2</version>
        </dependency>

        <dependency>
            <groupId>org.springframework.boot</groupId>
            <artifactId>spring-boot-starter-test</artifactId>
            <scope>test</scope>
        </dependency>
    </dependencies>

    <build>
        <plugins>
            <plugin>
                <groupId>org.springframework.boot</groupId>
                <artifactId>spring-boot-maven-plugin</artifactId>
            </plugin>
        </plugins>
    </build>


</project>

 

在本身本地電腦上創建一個Prometheus和Grafana儀表盤，用來可視化監控Spring Boot應用產生的全部metrics。

Spring Boot使用Micrometer，一個應用metrics組件，將actuator metrics整合到外部監控系統中。
它支持不少種監控系統，好比Netflix Atalas, AWS Cloudwatch, Datadog, InfluxData, SignalFx, Graphite, Wavefront和Prometheus等。
爲了整合Prometheus，你須要增長micrometer-registry-prometheus依賴：

在SpringBoot2.X中，spring-boot-starter-actuator引入了io.micrometer，對1.X中的metrics進行了重構，主要特色是支持tag/label，配合支持tag/label的監控系統，使得咱們能夠更加方便地對metrics進行多維度的統計查詢及監控
spring-boot-starter-actuator，主要是提供了Prometheus端點，不用重複造輪子。 

ActuatordemoApplication

package io.spring2go.promdemo.actuatordemo; import org.springframework.boot.SpringApplication; import org.springframework.boot.autoconfigure.SpringBootApplication; import org.springframework.context.annotation.Bean; import org.springframework.stereotype.Controller; import org.springframework.web.bind.annotation.RequestMapping; import org.springframework.web.bind.annotation.ResponseBody; import io.micrometer.core.instrument.MeterRegistry; import io.micrometer.spring.autoconfigure.MeterRegistryCustomizer; @SpringBootApplication @Controller public class ActuatordemoApplication { public static void main(String[] args) { SpringApplication.run(ActuatordemoApplication.class, args); } @RequestMapping(value = "/hello-world") public @ResponseBody String sayHello() { return "hello, world"; }  @Bean MeterRegistryCustomizer<MeterRegistry> metricsCommonTags() { return registry -> registry.config().commonTags("application", "actuator-demo"); } }

 

上面要設置註冊功能

commonTags必須叫作application

application.properties

endpoints.sensitive=false

 

訪問actouor端點的時候不須要用戶名和密碼

接下來咱們啓動應用

 

若是要收到埋點，使用下面的micrometer提供的方法在springboot應用中進行埋點就能夠了

 

 
 
# my global config
global:
 scrape_interval: 5s # Set the scrape interval to every 15 seconds. Default is every 1 minute.
 evaluation_interval: 5s # Evaluate rules every 15 seconds. The default is every 1 minute.
 # scrape_timeout is set to the global default (10s).
 
 
 
 
# Alertmanager configuration
alerting:
 alertmanagers:
 - static_configs:
 - targets:
 # - alertmanager:9093
 
 
 
 
# Load rules once and periodically evaluate them according to the global 'evaluation_interval'.
rule_files:
 #- "simulator_alert_rules.yml"
 # - "second_rules.yml"
 
 
 
 
# A scrape configuration containing exactly one endpoint to scrape:
# Here it's Prometheus itself.
scrape_configs:
 # The job name is added as a label `job=<job_name>` to any timeseries scraped from this config.
 - job_name: 'actuator-demo'
 metrics_path: /prometheus
 # metrics_path defaults to '/metrics'
 # scheme defaults to 'http'.
 
 
 
 
 static_configs:
 - targets: ['localhost:8080']
 
 
 
 

 

 
 

這裏代碼  - job_name: 'actuator-demo'要和代碼  return registry -> registry.config().commonTags("application", "actuator-demo");中的對應起來
 這樣整個監控就啓動起來了
 

 

 
 

 
 

jvm dashboard對於的id爲4701
 

添加4701，數據源必定要選擇普羅米修斯的數據源
 

 

 
 

134.Prometheus 監控最佳實踐~1.mp4
 

例如http的請求數目，http請求的平均演示，http500的錯誤數目。cpu、內存、磁盤的使用狀況等
 

 

online爲請求響應的系統：前臺的web系統，請求、錯誤、db、緩存、請求延遲等進行跟蹤
 

offline serving system：任務隊列系統 隊裏工做線程的利用率、任務線程執行的狀況
 

Batch Jobs：批處理系統，普羅米修斯到網關上拉去數據
 

普羅米修斯的高可用
 

 

 
 

 
 

 
 


普羅米修斯默認只支持15天的數據，超過整個範圍的數據須要單獨作處理
 

 
 

135.主流開源時序數據庫比較~1.mp4
 

137.微服務監控體系總結~1.mp4
 

 

系統層監控：底層硬件，操做系統的監控，普羅米修斯作系統操做系統的監控
 

應用層的監控：消息隊列redis、mysql spingboot 普羅米修斯進行埋點，cat skywalking作應用層的監控 elk能夠統計系統的日誌作監控
 

 
 

業務層：主要桶業務進行統計，例如轉帳等業務指標，普羅米修斯就能夠搞定
 

端用戶體驗：客戶訪問網頁的時間，聽雲等
 

 

 
 

 微服務架構的監控體系
 

左邊是對一個微服務架構作監控
 

主要的監控類別有下面的三種：log、Trace、Metrics
 

log：微服務產生log以後使用logstash進行日誌的收集、而後使用kafka隊列作緩衝，保證消息不被丟失，而後存儲到es數據庫中
 

trace：追蹤分佈式鏈路調用的場景
 

metris：在微服務裏面進行埋點，抓取metris數據，而後作告警等操做
 

普羅米修斯也能夠對kafka、cat等中間件進行監控，而後進行告警操做
 

 
 

 

 
 

 
 

 


        

            

        	





    

		

            
相關文章

            

                

                

                    

                

            


            
相關標籤/搜索

            
        


    

        

            

        

        

        
        

        

        

    




	

    





    

    	

    

        每日一句
    

    
    
    	每一个你不满意的现在，都有一个你没有努力的曾经。
    







    

    


    



    

        本站公眾號
    

    

           歡迎關注本站公眾號,獲取更多信息

        
    




    

    




	


	







    

        聯繫我們
        最近搜索
        最新文章
        
        沪ICP备13005482号-10

            MyBatis教程
            SQL 教程
            MySQL教程
            Java 教程
            Thymeleaf 教程
            Hibernate教程
            Spring教程 
            Redis教程