nGrinder二次开发之监控机器性能数据

1.背景

做性能测试时，统计性能数据分为被压系统的数据和被压系统所在机器的数据，被压系统所在机器的数据主要包括CPU利用率、内存使用率、网络IO、磁盘IO和负载load；nGrinder默认只收集CPU, Memory, Received Byte/s, Sent Byte Per Secode/s，同时支持自定义数据收集，之前介绍过相关内容：nGrinder对监控机器收集自定义数据及源码分析。

展示自定义数据需定时的写文件，比较麻烦；本文将介绍如何改动源码，直接展示CPU利用率、内存使用率、网络IO、磁盘IO和负载load这5类数据。

2.实现

nGrinder是通过sigar api统计机器性能数据，sigar项目主页：https://support.hyperic.com/display/SIGAR/Home， sigar api ：http://cpansearch.perl.org/src/DOUGM/hyperic-sigar-1.6.3-src/docs/javadoc/org/hyperic/sigar/package-summary.html。

2-1.新增数据类型

ngrinder-core中package：org.ngrinder.monitor.share.domain下SystemInfo类定义了收集的数据类型，新增load、read、write和memUsedPercentage数据字段：

/**
* modify by hugang
*/
public class SystemInfo extends MonitorInfo implements Serializable {

    private static final long serialVersionUID = -2995334644975166549L;

    /**
     * Header field of monitor status fields.
     */
    // 新增load、磁盘IO和内存使用率
    public static final String HEADER = "ip,system,collectTime,freeMemory,"
            + "totalMemory,cpuUsedPercentage,receivedPerSec,sentPerSec,load,read,write,memUsedPercentage";

    public boolean isParsed() {
        return true;
    }

    /**
     * Enum for the system type, linux or windows.
     */
    public enum System {
        LINUX, WINDOW
    }

    private System system;

    // 网络IO
    protected BandWidth bandWidth;

    private long totalCpuValue;

    private long idleCpuValue;

    private long freeMemory;

    private long totalMemory;

    private float cpuUsedPercentage;


    // load
    private double load;

    // 磁盘读
    private long read;

    // 磁盘写
    private long write;


    // 内存使用率
    private double memUsedPercentage;


    ...
    set()、get()方法


    public void parse(CompositeData cd) {
          // 新增
            this.load = getDouble(cd, "load");
            this.write = getLong(cd, "write");
            this.read = getLong(cd, "read");
            this.memUsedPercentage = getDouble(cd, "memUsedPercentage"); 
    }

    public String toRecordString() {
        StringBuilder sb = new StringBuilder();
        sb.append(ip).append(",").append(system).append(",");
        sb.append(DateUtils.getCollectTimeInLong(new Date(getCollectTime()))).append(",").append(freeMemory).append(",");

        sb.append(totalMemory).append(",").append(cpuUsedPercentage);
        if (bandWidth != null) {
            sb.append(",").append(bandWidth.getReceivedPerSec()).append(",").append(bandWidth.getSentPerSec());
        } 
        // 新类型load、read、write、memUsedPercentage数据拼接
        sb.append(",").append(load).append(",").append(read).append(",").append(write).append(",").append(memUsedPercentage);
        if (customValues != null) {
            sb.append(",").append(customValues);
        }
        return sb.toString();
    }

public static class NullSystemInfo extends SystemInfo {
        private static final NullSystemInfo instance = new NullSystemInfo();

        public static SystemInfo getNullSystemInfo() {
            return instance;
        }

        /**
         * Return the empty record string.
         *
         * @return null filled record string.
         * @see #toRecordString()
         */
        @Override
        public String toRecordString() {
            StringBuilder sb = new StringBuilder();
            sb.append("null").append(",").append("null").append(",");
            sb.append("null").append(",").append("null").append(",");
            sb.append("null").append(",").append("null");
            if (bandWidth != null) {
                sb.append(",").append("null").append(",").append("null");
            }
            // 拼接新类型数据为"null"
            sb.append(",").append("null").append(",").append("null").append(",").append("null");
            if (customValues != null) {
                int valueCount = StringUtils.countMatches(customValues, ",") + 1;
                for (int i = 0; i < valueCount; i++) {
                    sb.append(",").append("null");
                }
            }
            return sb.toString();
        }

        public boolean isParsed() {
            return false;
        }
    }



}

2-2.收集新增数据

ngrinder-core中package：org.ngrinder.monitor.collector下SystemDataCollector类中execute()方法实现了系统数据的收集：

/**
* modify by hugang
*/
public synchronized SystemInfo execute() {
        SystemInfo systemInfo = new SystemInfo();
        systemInfo.setCollectTime(System.currentTimeMillis());
        try {
            BandWidth networkUsage = getNetworkUsage();
            BandWidth bandWidth = networkUsage.adjust(prev.getBandWidth());
            systemInfo.setBandWidth(bandWidth);
            // getCombined()代表 Sum of User + Sys + Nice + Wait
            systemInfo.setCPUUsedPercentage((float) sigar.getCpuPerc().getCombined() * 100);
            Cpu cpu = sigar.getCpu();
            systemInfo.setTotalCpuValue(cpu.getTotal());
            systemInfo.setIdleCpuValue(cpu.getIdle());
            Mem mem = sigar.getMem();
            systemInfo.setTotalMemory(mem.getTotal() / 1024L);
            systemInfo.setFreeMemory(mem.getActualFree() / 1024L);

            // 新增load、磁盘IO和内存使用率
            // The system load averages for the past 1, 5, and 15 minutes.
            double load = 0;
            long read = 0l;
            long write = 0l;

            load = sigar.getLoadAverage()[0];

            //LOGGER.info("monitor system load:{}", load);

            // 所有的文件系统
            FileSystem[] fileSystems = sigar.getFileSystemList();
            // 获取本地文件系统
            List localDevNames = new ArrayList();
            for(FileSystem fileSystem : fileSystems) {
                if(fileSystem.getType() == FileSystem.TYPE_LOCAL_DISK) {
                    localDevNames.add(fileSystem.getDevName());
                }
            }


            for(String localDevName : localDevNames) {
                read += sigar.getDiskUsage(localDevName).getReadBytes();
                write += sigar.getDiskUsage(localDevName).getWriteBytes();
            }


            //LOGGER.info("monitor system read:{}, write:{}", read, write);

            systemInfo.setLoad(load);
            systemInfo.setRead(read / 1024L);
            systemInfo.setWrite(write / 1024L);
            systemInfo.setMemUsedPercentage(mem.getUsedPercent());

            systemInfo.setSystem(OperatingSystem.IS_WIN32 ? SystemInfo.System.WINDOW : SystemInfo.System.LINUX);
            systemInfo.setCustomValues(getCustomMonitorData());
        } catch (Throwable e) {
            LOGGER.error("Error while getting system perf data:{}", e.getMessage());
            LOGGER.debug("Error trace is ", e);
        }
        prev = systemInfo;
        return systemInfo;
    }

2-3.前端monitor模板

ngrinder-controller下src/main/webapp/WEB-INF/ftl/perftest/detail_report/monitor.ftl定义了系统数据的可视化，修改如下：

<#setting number_format="computer">
<#import "../../common/spring.ftl" as spring/>
<div class="page-header page-header">
    <h4>Monitorh4>
div>
<h6 id="cpu_usage_chart_header">CPU利用率(User + Sys + Nice + Wait)，建议值：小于75%h6>
<div class="chart" id="cpu_usage_chart">div>
<h6 id="mem_usage_chart_header">Memory使用率，建议值：小于80%h6>
<div class="chart" id="mem_usage_chart">div>
<h6 id="received_byte_per_sec_chart_header">Received Byte Per Secondh6>
<div class="chart" id="received_byte_per_sec_chart">div>
<h6 id="sent_byte_per_sec_chart_header">Sent Byte Per Secondh6>
<div class="chart" id="sent_byte_per_sec_chart">div>

<h6 id="custom_monitor_chart_1_header">Load-average(one-minute)h6>
<div class="chart" id="custom_monitor_chart_1">div>
<h6 id="custom_monitor_chart_2_header">Physical Disk readBytes(total number of physical disk reads)h6>
<div class="chart" id="custom_monitor_chart_2">div>
<h6 id="custom_monitor_chart_3_header">Physical Disk writesBytes(total number of physical disk writes)h6>
<div class="chart" id="custom_monitor_chart_3">div>


<script>
    //@ sourceURL=/perftest/detail_report/monitor
    function getMonitorDataAndDraw(testId, targetIP) {
        var ajaxObj = new AjaxObj("/perftest/api/" + testId + "/monitor");
        ajaxObj.params = {
            targetIP: targetIP,
            imgWidth: parseInt($("#cpu_usage_chart").width())
        };
        ajaxObj.success = function (data) {
            var interval = data.chartInterval;


            // modify by hugang
            // data.customData1为load, data.customData2为磁盘读， data.customData3为磁盘写
            drawChart('cpu_usage_chart', [data.cpu], formatPercentage, interval);
              // 调整为内存使用率
            drawChart('mem_usage_chart', [data.customData4], formatPercentage, interval);
            // load
            drawChart("custom_monitor_chart_1", [data.customData1], formatDouble, interval);
            drawChart("received_byte_per_sec_chart", [data.received], formatNetwork, interval);
            drawChart("sent_byte_per_sec_chart", [data.sent], formatNetwork, interval);
            // 磁盘读
            drawChart("custom_monitor_chart_2", [data.customData2], formatMemory, interval);
            // 磁盘写
            drawChart("custom_monitor_chart_3", [data.customData3], formatMemory, interval);

            createChartExportButton("<@spring.message "perfTest.report.exportImg.button"/>", "<@spring.message "perfTest.report.exportImg.title"/>");
        };
        ajaxObj.call();
    }
    function drawChart(id, data, yFormat, interval) {
        return new Chart(id, data, interval, {yAxisFormatter: yFormat}).plot();
    }

    function drawOptionalChart(id, data, interval, labels) {
        var result = drawChart(id, data, interval, labels);
        if (result.isEmpty()) {
            $("#" + id).hide();
            $("#" + id + "_header").hide();
        }
    }
    getMonitorDataAndDraw(${id}, "${targetIP}");


script>

3.结果展示

对工程进行打包：

hugangdeMacBook-Pro:CPC hugang$ mvn clean -Dmaven.test.skip=true package

部署新war包，服务正常启动后；需重新下载监控包，因为依赖的ngrinder-core.jar包有改动，重新启动监控服务后，执行性能测试任务。

系统监控数据文件：/root/.ngrinder/perftest/0_999/${任务id}/report下 monitor_system_${ip}.data新增load、read、write和memUsedPercentage。

前端展示：