Tomcat 监控与诊断

Monitoring and Diagnostics

概述

有效的监控和诊断是确保Tomcat稳定运行的关键。本文介绍JMX监控、性能指标收集、APM工具集成和故障诊断方法。

1. JMX监控配置

1.1 启用JMX

# setenv.sh - JMX配置
export CATALINA_OPTS="$CATALINA_OPTS -Dcom.sun.management.jmxremote"
export CATALINA_OPTS="$CATALINA_OPTS -Dcom.sun.management.jmxremote.port=9999"
export CATALINA_OPTS="$CATALINA_OPTS -Dcom.sun.management.jmxremote.authenticate=false"
export CATALINA_OPTS="$CATALINA_OPTS -Dcom.sun.management.jmxremote.ssl=false"
export CATALINA_OPTS="$CATALINA_OPTS -Djava.rmi.server.hostname=192.168.1.10"

1.2 JMX安全配置

# 启用JMX认证
export CATALINA_OPTS="$CATALINA_OPTS -Dcom.sun.management.jmxremote.authenticate=true"
export CATALINA_OPTS="$CATALINA_OPTS -Dcom.sun.management.jmxremote.password.file=$CATALINA_HOME/conf/jmxremote.password"
export CATALINA_OPTS="$CATALINA_OPTS -Dcom.sun.management.jmxremote.access.file=$CATALINA_HOME/conf/jmxremote.access"

2. 性能监控工具

2.1 JVM监控脚本

#!/bin/bash
# jvm-monitor.sh

JAVA_PID=$(jps -l | grep Bootstrap | awk '{print $1}')

if [ -z "$JAVA_PID" ]; then
    echo "Tomcat进程未找到"
    exit 1
fi

echo "=== JVM监控 $(date) ==="

# 内存使用情况
echo "内存使用:"
jstat -gc $JAVA_PID | awk 'NR==2 {
    printf "Eden: %.2fMB, Survivor: %.2fMB, Old: %.2fMB\n", 
           $3/1024, ($2+$1)/1024, $4/1024
}'

# GC统计
echo "GC统计:"
jstat -gccapacity $JAVA_PID | awk 'NR==2 {
    printf "新生代容量: %.2fMB, 老年代容量: %.2fMB\n", $4/1024, $10/1024
}'

# 线程信息
echo "线程信息:"
jstack $JAVA_PID | grep "java.lang.Thread.State" | sort | uniq -c

# CPU使用率
echo "进程CPU使用率:"
ps -p $JAVA_PID -o pid,ppid,pcpu,pmem,cmd

2.2 应用监控

// TomcatMonitor.java
package com.example.monitor;

import javax.management.*;
import java.lang.management.ManagementFactory;

public class TomcatMonitor {

    private MBeanServer mbs = ManagementFactory.getPlatformMBeanServer();

    public void monitorTomcat() throws Exception {
        System.out.println("=== Tomcat监控报告 ===");

        // 连接器监控
        monitorConnectors();

        // 会话监控
        monitorSessions();

        // 线程池监控
        monitorThreadPools();
    }

    private void monitorConnectors() throws Exception {
        for (ObjectName connector : mbs.queryNames(new ObjectName("Catalina:type=Connector,*"), null)) {
            String port = connector.getKeyProperty("port");
            Integer currentThreads = (Integer) mbs.getAttribute(connector, "currentThreadCount");
            Integer maxThreads = (Integer) mbs.getAttribute(connector, "maxThreads");
            Long requestCount = (Long) mbs.getAttribute(connector, "requestCount");

            System.out.printf("连接器 %s: 线程 %d/%d, 请求数 %d%n", 
                              port, currentThreads, maxThreads, requestCount);
        }
    }

    private void monitorSessions() throws Exception {
        for (ObjectName manager : mbs.queryNames(new ObjectName("Catalina:type=Manager,*"), null)) {
            String path = manager.getKeyProperty("path");
            Integer activeSessions = (Integer) mbs.getAttribute(manager, "activeSessions");
            System.out.printf("应用 %s: 活跃会话 %d%n", path.isEmpty() ? "/" : path, activeSessions);
        }
    }

    private void monitorThreadPools() throws Exception {
        for (ObjectName executor : mbs.queryNames(new ObjectName("Catalina:type=Executor,*"), null)) {
            String name = executor.getKeyProperty("name");
            Integer activeCount = (Integer) mbs.getAttribute(executor, "activeCount");
            Integer poolSize = (Integer) mbs.getAttribute(executor, "poolSize");
            System.out.printf("线程池 %s: 活跃 %d, 总数 %d%n", name, activeCount, poolSize);
        }
    }
}

3. APM工具集成

3.1 Prometheus + Grafana

<!-- 添加JMX Exporter -->
<Context>
    <JarScanner>
        <JarScanFilter defaultTldScan="false"/>
    </JarScanner>
</Context>

# 启动JMX Exporter
java -javaagent:jmx_prometheus_javaagent-0.17.0.jar=8080:config.yaml \
     -jar your-application.jar

3.2 Micrometer集成

// MetricsConfiguration.java
@Configuration
public class MetricsConfiguration {

    @Bean
    public MeterRegistry meterRegistry() {
        return Metrics.globalRegistry;
    }

    @Bean
    public TomcatMetricsBinder tomcatMetricsBinder() {
        return new TomcatMetricsBinder();
    }
}

4. 健康检查

4.1 健康检查端点

// HealthCheckServlet.java
@WebServlet("/health")
public class HealthCheckServlet extends HttpServlet {

    @Override
    protected void doGet(HttpServletRequest request, HttpServletResponse response) 
            throws ServletException, IOException {

        boolean healthy = true;
        Map<String, String> checks = new HashMap<>();

        // 数据库检查
        try {
            // 数据库连接测试
            checks.put("database", "UP");
        } catch (Exception e) {
            healthy = false;
            checks.put("database", "DOWN: " + e.getMessage());
        }

        // 内存检查
        Runtime runtime = Runtime.getRuntime();
        long usedMemory = runtime.totalMemory() - runtime.freeMemory();
        double memoryUsage = (double) usedMemory / runtime.maxMemory();

        if (memoryUsage > 0.9) {
            healthy = false;
            checks.put("memory", "HIGH: " + String.format("%.1f%%", memoryUsage * 100));
        } else {
            checks.put("memory", "OK: " + String.format("%.1f%%", memoryUsage * 100));
        }

        response.setContentType("application/json");
        response.setStatus(healthy ? 200 : 503);

        Gson gson = new Gson();
        response.getWriter().write(gson.toJson(Map.of(
            "status", healthy ? "UP" : "DOWN",
            "checks", checks,
            "timestamp", System.currentTimeMillis()
        )));
    }
}

4.2 自动健康检查脚本

#!/bin/bash
# health-check.sh

TOMCAT_URL="http://localhost:8080"
ALERT_EMAIL="admin@example.com"

check_tomcat_health() {
    local status_code=$(curl -s -o /dev/null -w "%{http_code}" "$TOMCAT_URL/health")

    if [ "$status_code" = "200" ]; then
        echo "✓ Tomcat健康检查通过"
        return 0
    else
        echo "✗ Tomcat健康检查失败 (状态码: $status_code)"
        echo "Tomcat健康检查失败" | mail -s "Tomcat告警" "$ALERT_EMAIL"
        return 1
    fi
}

check_response_time() {
    local response_time=$(curl -s -o /dev/null -w "%{time_total}" "$TOMCAT_URL/")
    local time_ms=$(echo "$response_time * 1000" | bc)

    if (( $(echo "$time_ms > 5000" | bc -l) )); then
        echo "⚠ 响应时间过长: ${time_ms}ms"
        return 1
    else
        echo "✓ 响应时间正常: ${time_ms}ms"
        return 0
    fi
}

main() {
    echo "=== Tomcat健康检查 $(date) ==="

    local failed=0

    check_tomcat_health || failed=1
    check_response_time || failed=1

    if [ $failed -eq 1 ]; then
        echo "健康检查失败"
        exit 1
    else
        echo "所有检查通过"
        exit 0
    fi
}

main

5. 性能分析工具

5.1 线程分析

#!/bin/bash
# thread-analyzer.sh

JAVA_PID=$(jps -l | grep Bootstrap | awk '{print $1}')

analyze_threads() {
    echo "=== 线程分析 ==="

    # 生成线程转储
    jstack $JAVA_PID > /tmp/thread_dump_$(date +%Y%m%d_%H%M%S).txt

    # 分析线程状态
    echo "线程状态分布:"
    jstack $JAVA_PID | grep "java.lang.Thread.State" | \
    sort | uniq -c | sort -nr

    # 检查死锁
    if jstack $JAVA_PID | grep -q "Found Java-level deadlock"; then
        echo "⚠️ 检测到死锁!"
    fi

    # 分析阻塞线程
    echo "阻塞的线程:"
    jstack $JAVA_PID | grep -A 5 "BLOCKED"
}

analyze_threads

5.2 内存分析

#!/bin/bash
# memory-analyzer.sh

JAVA_PID=$(jps -l | grep Bootstrap | awk '{print $1}')

analyze_memory() {
    echo "=== 内存分析 ==="

    # 堆内存使用
    jmap -heap $JAVA_PID

    # 对象直方图
    echo "对象使用排行:"
    jmap -histo $JAVA_PID | head -20

    # 检查内存泄漏风险
    local old_gen_usage=$(jstat -gc $JAVA_PID | awk 'NR==2 {print int($4*100/($1+$2+$3+$4+$5))}')

    if [ $old_gen_usage -gt 80 ]; then
        echo "⚠️ 老年代使用率过高: ${old_gen_usage}%"
    fi
}

analyze_memory

6. 监控告警

6.1 告警规则

#!/bin/bash
# alert-rules.sh

JAVA_PID=$(jps -l | grep Bootstrap | awk '{print $1}')
ALERT_EMAIL="admin@example.com"

# 内存告警
check_memory_alert() {
    local memory_usage=$(jstat -gc $JAVA_PID | awk 'NR==2 {
        used=($2+$3+$4); total=($1+$2+$3+$4+$5); 
        print int(used*100/total)
    }')

    if [ $memory_usage -gt 85 ]; then
        echo "内存使用率告警: ${memory_usage}%" | \
        mail -s "Tomcat内存告警" "$ALERT_EMAIL"
    fi
}

# 线程告警
check_thread_alert() {
    local thread_count=$(jstack $JAVA_PID | grep "^\"" | wc -l)

    if [ $thread_count -gt 500 ]; then
        echo "线程数量告警: $thread_count" | \
        mail -s "Tomcat线程告警" "$ALERT_EMAIL"
    fi
}

# GC告警
check_gc_alert() {
    local full_gc_count=$(jstat -gc $JAVA_PID | awk 'NR==2 {print $9}')
    local gc_time=$(jstat -gc $JAVA_PID | awk 'NR==2 {print $10}')

    # 检查Full GC频率
    if [ $full_gc_count -gt 10 ]; then
        echo "Full GC频率过高: $full_gc_count" | \
        mail -s "Tomcat GC告警" "$ALERT_EMAIL"
    fi
}

# 执行所有检查
check_memory_alert
check_thread_alert  
check_gc_alert

6.2 监控仪表板

#!/bin/bash
# dashboard.sh

generate_dashboard() {
    local html_file="/tmp/tomcat_dashboard.html"

    cat > "$html_file" << 'EOF'
<!DOCTYPE html>
<html>
<head>
    <title>Tomcat监控仪表板</title>
    <meta http-equiv="refresh" content="30">
</head>
<body>
    <h1>Tomcat监控仪表板</h1>
    <div id="metrics">
        <h2>系统指标</h2>
        <pre id="system-info"></pre>

        <h2>JVM指标</h2>
        <pre id="jvm-info"></pre>

        <h2>应用指标</h2>
        <pre id="app-info"></pre>
    </div>

    <script>
        // 定期刷新数据
        function updateMetrics() {
            fetch('/metrics')
                .then(response => response.json())
                .then(data => {
                    document.getElementById('system-info').textContent = JSON.stringify(data.system, null, 2);
                    document.getElementById('jvm-info').textContent = JSON.stringify(data.jvm, null, 2);
                    document.getElementById('app-info').textContent = JSON.stringify(data.application, null, 2);
                });
        }

        setInterval(updateMetrics, 5000);
        updateMetrics();
    </script>
</body>
</html>
EOF

    echo "仪表板已生成: $html_file"
}

generate_dashboard

小结

通过本文学习，你应该掌握：

JMX监控配置和安全设置
性能监控脚本和工具使用
APM工具的集成方法
健康检查端点的实现
线程和内存分析技术
监控告警规则配置
监控仪表板的构建

下一篇文章将介绍Tomcat代理配置技术。

监控与诊断

Tomcat 监控与诊断

Monitoring and Diagnostics

概述

1. JMX监控配置

1.1 启用JMX

1.2 JMX安全配置

2. 性能监控工具

2.1 JVM监控脚本

2.2 应用监控

3. APM工具集成

3.1 Prometheus + Grafana

3.2 Micrometer集成

4. 健康检查

4.1 健康检查端点

4.2 自动健康检查脚本

5. 性能分析工具

5.1 线程分析

5.2 内存分析

6. 监控告警

6.1 告警规则

6.2 监控仪表板

小结

results matching ""

No results matching ""