Details
diff --git a/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/monitoring/aspects/TigrisCoordinator.java b/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/monitoring/aspects/TigrisCoordinator.java
index 4e070d3..00e1355 100644
--- a/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/monitoring/aspects/TigrisCoordinator.java
+++ b/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/monitoring/aspects/TigrisCoordinator.java
@@ -205,7 +205,7 @@ public class TigrisCoordinator implements Runnable {
try {
repository.save(logTrace);
logger.debug("New trace entry: " + logTrace);
- sampling.addSampledItem(granularity, currentTimeMillis() - monitoringStartTime);
+ sampling.addSampledItem(granularity, currentTimeMillis() - monitoringStartTime, 0);
} catch (Exception e) {
logger.debug("Couldn't trace " + signature + " due to: " + e.getMessage());
}
diff --git a/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/Sampling.java b/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/Sampling.java
index a42fa22..1919118 100644
--- a/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/Sampling.java
+++ b/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/Sampling.java
@@ -1,7 +1,6 @@
package br.ufrgs.inf.prosoft.tigris.sampling;
import com.google.common.util.concurrent.AtomicDouble;
-import org.apache.commons.lang3.RandomUtils;
import org.apache.commons.math3.distribution.BinomialDistribution;
import org.apache.commons.math3.ml.neuralnet.sofm.util.ExponentialDecayFunction;
import org.apache.commons.math3.stat.descriptive.DescriptiveStatistics;
@@ -35,7 +34,7 @@ public class Sampling {
private long startTime;
private ExponentialDecayFunction decayingPrecision;
private FrequencyDataSet population = new FrequencyDataSet(), sample = new FrequencyDataSet();
- private PerformanceBaselineDataSet performanceBaselineDataSet = new PerformanceBaselineDataSet();
+ Map<Granularity, DescriptiveStatistics> sampledMemoryDataSet = new ConcurrentHashMap<>();
private Map<Granularity, DescriptiveStatistics> sampledDataSet = new ConcurrentHashMap<>();
Logger logger = LoggerFactory.getLogger(Sampling.class);
@@ -177,12 +176,18 @@ public class Sampling {
// this.performanceBaselineDataSet.addItem(granularity, executionTime);
// }
- public void addSampledItem(Granularity granularity, long executionTime) {
+ public void addSampledItem(Granularity granularity, long executionTime, long actualMemUsed) {
sample.addItem(granularity, executionTime);
DescriptiveStatistics statistics = sampledDataSet.getOrDefault(granularity, new DescriptiveStatistics());
statistics.addValue(executionTime);
sampledDataSet.put(granularity, statistics);
+
+ if (actualMemUsed > 0) {
+ DescriptiveStatistics memoryStatistics = sampledMemoryDataSet.getOrDefault(granularity, new DescriptiveStatistics());
+ memoryStatistics.addValue(actualMemUsed);
+ sampledMemoryDataSet.put(granularity, memoryStatistics);
+ }
}
public long getMonitoringCycleTime() {
@@ -194,6 +199,7 @@ public class Sampling {
}
public double decayingConfidenceFactor(long timeInMilliseconds) {
+ if (timeInMilliseconds == 0) return 0.999999999999;
synchronized (decayingPrecisionLock) {
return new BigDecimal(decayingPrecision.value(timeInMilliseconds))
.setScale(4, BigDecimal.ROUND_FLOOR).doubleValue();
@@ -270,6 +276,7 @@ public class Sampling {
this.sample.clear();
this.population.clear();
this.sampledDataSet.clear();
+ this.sampledMemoryDataSet.clear();
this.startTime = System.currentTimeMillis();
logger.info("Monitoring is reset...");
}
@@ -281,40 +288,8 @@ public class Sampling {
return monitoringCycle;
}
- private int baselineWindow;
private Instant baselineWindowStart = Instant.now();
-// public void managePerformanceBaseline(PerformanceBaselineDataSet performanceMonitoring) {
-// if (performanceBaselineEnabled) { //is it already enabled?
-// if (Duration.between(baselineWindowStart, Instant.now()).getSeconds() > baselineWindow) {
-//// logger.info("Collected performance baseline of {} traces", this.performanceBaselineDataSet.getTotalItems());
-// performanceBaselineEnabled = false;
-// this.performanceBaselineDataSet.clear();
-// }
-// return;
-// }
-//
-// double chance = new BinomialDistribution(1, 0.1d).sample();
-// if (chance == 1) {
-// baselineWindow = 2;//RandomUtils.nextInt(2, 4);
-// baselineWindowStart = Instant.now();
-//
-// logger.info("Enabling performance baseline that needs {} seconds of traces.", baselineWindow);
-// performanceBaselineEnabled = true;
-// }
-//
-//
-//
-//// double chance = new BinomialDistribution(1, 0.1d).sample();
-//// if (chance == 1) {
-//// baselineWindow = 2;//RandomUtils.nextInt(2, 4);
-//// baselineWindowStart = Instant.now();
-////
-//// logger.info("Enabling performance baseline that needs {} seconds of traces.", baselineWindow);
-//// performanceBaselineEnabled = true;
-//// }
-// }
-
public FrequencyDataSet getSample() {
return sample;
}
@@ -327,6 +302,10 @@ public class Sampling {
return sampledDataSet;
}
+ public Map<Granularity, DescriptiveStatistics> getSampledMemoryTraces() {
+ return sampledMemoryDataSet;
+ }
+
public double getSamplingRate() {
return samplingRate;
}
diff --git a/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/SamplingAspect.java b/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/SamplingAspect.java
index ee10160..1e1ccfc 100644
--- a/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/SamplingAspect.java
+++ b/tigris/src/main/java/br/ufrgs/inf/prosoft/tigris/sampling/SamplingAspect.java
@@ -1,8 +1,6 @@
package br.ufrgs.inf.prosoft.tigris.sampling;
-import br.ufrgs.inf.prosoft.tigris.configuration.annotation.TigrisConfiguration;
-import br.ufrgs.inf.prosoft.tigris.exceptions.ConfigurationException;
-import br.ufrgs.inf.prosoft.tigris.utils.ConfigurationUtils;
+import org.apache.commons.math3.stat.descriptive.DescriptiveStatistics;
import org.apache.commons.math3.stat.descriptive.SummaryStatistics;
import org.aspectj.lang.ProceedingJoinPoint;
import org.aspectj.lang.annotation.Around;
@@ -20,71 +18,59 @@ import java.util.Set;
import static java.lang.System.nanoTime;
@Aspect
-public class SamplingAspect implements Runnable {
+public class SamplingAspect {
@Pointcut(
//any execution except the own framework
"(" +
"(" +
- "execution(* org.dacapo.h2.TPCCSubmitter.runTransaction(..)) || " + //h2
- "execution(* org.dacapo.lusearch.QueryProcessor.doPagingSearch(..)) || " + //lusearch
- "execution(* org.dacapo.xalan.XalanWorker.transform(..)) || " + //xalan
-
- //tradebeans:
- "(" +
- "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doHome(..)) || " +
- "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doPortfolio(..)) || " +
- "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doQuote(..)) || " +
- "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doBuy(..)) || " +
- "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doUpdate(..)) || " +
- "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doRegister(..)) || " +
- "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doSell(..))" +
- ") || " +
- //cassandra:
- "(" +
- "execution(* site.ycsb.workloads.CoreWorkload.doTransactionRead(..)) || " +
- "execution(* site.ycsb.workloads.CoreWorkload.doTransactionUpdate(..)) || " +
- "execution(* site.ycsb.workloads.CoreWorkload.doTransactionInsert(..)) || " +
- "execution(* site.ycsb.workloads.CoreWorkload.doTransactionScan(..)) || " +
- "execution(* site.ycsb.workloads.CoreWorkload.doTransactionReadModifyWrite(..))" +
- ")" +
+ "execution(* org.dacapo.h2.TPCCSubmitter.runTransaction(..)) || " + //h2
+ "execution(* org.dacapo.lusearch.QueryProcessor.doPagingSearch(..)) || " + //lusearch
+ "execution(* org.dacapo.xalan.XalanWorker.transform(..)) || " + //xalan
+
+ //tradebeans:
+ "(" +
+ "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doHome(..)) || " +
+ "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doPortfolio(..)) || " +
+ "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doQuote(..)) || " +
+ "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doBuy(..)) || " +
+ "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doUpdate(..)) || " +
+ "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doRegister(..)) || " +
+ "execution(* org.apache.geronimo.daytrader.javaee6.dacapo.DaCapoTrader.doSell(..))" +
+ ") || " +
+ //cassandra:
+ "(" +
+ "execution(* site.ycsb.workloads.CoreWorkload.doTransactionRead(..)) || " +
+ "execution(* site.ycsb.workloads.CoreWorkload.doTransactionUpdate(..)) || " +
+ "execution(* site.ycsb.workloads.CoreWorkload.doTransactionInsert(..)) || " +
+ "execution(* site.ycsb.workloads.CoreWorkload.doTransactionScan(..)) || " +
+ "execution(* site.ycsb.workloads.CoreWorkload.doTransactionReadModifyWrite(..))" +
+ ")" +
") " +
-// "(execution(* *(..)) && !within(br.ufrgs.inf.prosoft..*) " +
- //avoid calls from repository while serializing objects, it is necessary if a hash could not be used
- //"&& !cflow(call(* br.ufrgs.inf.prosoft.tigris.monitoring.storage..*(..))) " +
- //conditional to enable and disable at runtime
"&& if())"
)
public static boolean anyCall() {
return enabled;
}
- //tigris config
- private final TigrisConfiguration tigrisConfiguration;
private static Sampling sampling;
- public SamplingAspect() throws ClassNotFoundException {
- Class<?> configClass = ConfigurationUtils.getAvailableConfigurationClass(TigrisConfiguration.class);
- if (configClass == null) {
- //workaround for cassandra
- configClass = Class.forName("site.ycsb.ApplicationInitializer");
- }
+ public SamplingAspect() {
+ final String sampling_rate = System.getProperty("sampling_rate");
+ final String sampling_cycle_time = System.getProperty("sampling_cycle_time");
+ final String sampling_adaptive = System.getProperty("sampling_adaptive");
+ final String sampling_enabled = System.getProperty("sampling_enabled");
+ final String sampling_inversely = System.getProperty("sampling_inversely");
- if (configClass == null) {
- logger.info("Tigris tracing disabled, there is no annotations.");
- throw new ConfigurationException("Tigris tracing disabled, there is no annotations.");
- }
+ logger.info("Loaded properties. sampling_rate: {}, sampling_cycle_time: {}, sampling_adaptive: {}, sampling_enabled: {}, sampling_inversely: {}",
+ sampling_rate, sampling_cycle_time, sampling_adaptive, sampling_enabled, sampling_inversely);
- tigrisConfiguration = configClass.getAnnotation(TigrisConfiguration.class);
+ samplingEnabled = Boolean.parseBoolean(sampling_enabled);
+ adaptSamplingRateInverselyProportionalOps = Boolean.parseBoolean(sampling_inversely);
- logger.info("@TigrisConfiguration found.");
-
- sampling = new Sampling(tigrisConfiguration.samplingPercentage(), tigrisConfiguration.cycleTimeInMilliseconds(), tigrisConfiguration.adaptiveSamplingRate());
-
-// adaptiveSamplingExecutor.scheduleWithFixedDelay(
-// this::run,
-// 5, 1, TimeUnit.SECONDS);
+ sampling = new Sampling(Double.parseDouble(sampling_rate), Long.parseLong(sampling_cycle_time),
+ Boolean.parseBoolean(sampling_adaptive));
}
static Logger logger = LoggerFactory.getLogger(SamplingAspect.class);
@@ -93,19 +79,17 @@ public class SamplingAspect implements Runnable {
public static boolean samplingEnabled = true;
-// private final ScheduledExecutorService adaptiveSamplingExecutor = Executors.newSingleThreadScheduledExecutor(new NamedThreads(
-// "adaptive-sampling",
-// "readiness evaluation, pba trigger and sampling adaptation"
-// ));
-
@Around("anyCall()")
public Object aroundMethods(ProceedingJoinPoint joinPoint) throws Throwable {
long startTime = nanoTime();
+ long beforeUsedMem = Runtime.getRuntime().totalMemory() - Runtime.getRuntime().freeMemory();
Object result = joinPoint.proceed();
long endTime = nanoTime();
+ long afterUsedMem = Runtime.getRuntime().totalMemory() - Runtime.getRuntime().freeMemory();
long totalTime = endTime - startTime;
+ long actualMemUsed = afterUsedMem - beforeUsedMem;
- String signature = joinPoint.getSignature().toString() + joinPoint.getArgs()[0].toString(); //TODO this is to distinguish traces in H2 or lusearch / also run with Xalan / cassandra / tradebeans?
+ String signature = joinPoint.getSignature().toString() + joinPoint.getArgs()[0].toString();
Granularity granularity = new Granularity(GranularityType.METHOD, signature);
if (samplingEnabled) {
@@ -117,7 +101,7 @@ public class SamplingAspect implements Runnable {
boolean decision = sampling.samplingDecision(granularity, totalTime);
if (decision)
- sampling.addSampledItem(granularity, nanoTime() - startTime);
+ sampling.addSampledItem(granularity, nanoTime() - startTime, actualMemUsed);
}
monitoring.addMonitoringItem(granularity, totalTime);
@@ -126,21 +110,19 @@ public class SamplingAspect implements Runnable {
public static boolean adaptSamplingRateInverselyProportionalOps = false;
- public static void computeCycle (int sec, Set<Integer> markers) {
+ public static void computeCycle(int sec, Set<Integer> markers) {
if (!sampling.isAdaptiveSamplingRate() && markers != null && markers.contains(sec)) {
- if (samplingEnabled) { //uniform or inversely proportional
- final MonitoringCycle monitoringCycle = sampling.endMonitoringCycle();
- mseMap.put(sec, monitoringCycle.getAverageProcTimesSample());
- } else { //no monitoring
- OptionalDouble mean = monitoring.currentMonitoring.values().stream().mapToDouble(SummaryStatistics::getMean)
+ if (samplingEnabled) { //uniform, full or inversely proportional
+ final OptionalDouble mean = sampling.sampledMemoryDataSet.values().stream().mapToDouble(DescriptiveStatistics::getMean)
.average();
- monitoring.currentMonitoring.clear();
+ sampling.sampledMemoryDataSet.clear();
+
mseMap.put(sec, mean.getAsDouble());
}
}
}
- public static void addOperationsPerSecondAndAdapt (int operationsPerSecond, int sec, Set<Integer> markers) {
+ public static void addOperationsPerSecondAndAdapt(int operationsPerSecond, int sec, Set<Integer> markers) {
computeCycle(sec, markers);
if (adaptSamplingRateInverselyProportionalOps) {
@@ -150,8 +132,6 @@ public class SamplingAspect implements Runnable {
if (sampling.isAdaptiveSamplingRate()) {
currentSamplingRate = sampling.getSamplingRate();
- currentNormalBehavior = monitoring.baselinesPerSecondToNormal.getPercentile(50);
- currentNormalMonitoring = monitoring.monitoringPerSecondToNormal.getPercentile(50);
// response times are used only to increase/decrease the sampling rate
// if keep the response times under the error margin (compared to the baseline times) -> increase
@@ -164,7 +144,9 @@ public class SamplingAspect implements Runnable {
if (!sampling.isPerformanceBaselineEnabled() && sampling.isReady()) {
logger.info("Sample is ready, releasing for analysis and resetting...");
- mseMap.put(sec, sampling.getSample().getMeanExecutionTime());
+ final OptionalDouble mean = sampling.sampledMemoryDataSet.values().stream().mapToDouble(DescriptiveStatistics::getMean)
+ .average();
+ mseMap.put(sec, mean.getAsDouble());
cycleMarkers.add(sec);
cycle = sampling.endMonitoringCycle();
}
@@ -177,45 +159,12 @@ public class SamplingAspect implements Runnable {
public static PerformanceBaselineDataSet monitoring = new PerformanceBaselineDataSet();
static double currentSamplingRate;
- static double currentNormalMonitoring;
- static double currentNormalBehavior;
static MonitoringCycle cycle = new MonitoringCycle();
- @Override
- public void run() {
-// if (adaptSamplingRateInverselyProportionalOps) {
-// currentSamplingRate = sampling.adaptSamplingRateInverselyProportional(currentOps);
-// return;
-// }
-//
-// if (adaptSamplingRateInverselyProportionalUsers) {
-// currentSamplingRate = sampling.adaptSamplingRateInverselyProportional(currentUsers);
-// return;
-// }
-//
-// if (SamplingAspect.enabled && sampling.isAdaptiveSamplingRate()) {
-// sampling.managePerformanceBaseline();
-// currentSamplingRate = sampling.getSamplingRate();
-//
-// if (!sampling.isPerformanceBaselineEnabled() && sampling.isReady()) {
-// logger.info("Sample is ready, releasing for analysis and resetting...");
-// cycle = sampling.endMonitoringCycle();
-// }
-// }
- }
-
public static double getCurrentSamplingRate() {
return currentSamplingRate;
}
- public static double getCurrentNormalMonitoring() {
- return currentNormalMonitoring;
- }
-
- public static double getCurrentNormalBehavior() {
- return currentNormalBehavior;
- }
-
public static MonitoringCycle getCycle() {
return cycle;
}
diff --git a/tigris/src/test/java/br/ufrgs/inf/prosoft/tigris/SamplingTest.java b/tigris/src/test/java/br/ufrgs/inf/prosoft/tigris/SamplingTest.java
index 0f9fb3a..c808797 100644
--- a/tigris/src/test/java/br/ufrgs/inf/prosoft/tigris/SamplingTest.java
+++ b/tigris/src/test/java/br/ufrgs/inf/prosoft/tigris/SamplingTest.java
@@ -3,9 +3,7 @@ package br.ufrgs.inf.prosoft.tigris;
import br.ufrgs.inf.prosoft.tigris.sampling.Granularity;
import br.ufrgs.inf.prosoft.tigris.sampling.GranularityType;
import br.ufrgs.inf.prosoft.tigris.sampling.Sampling;
-import org.apache.commons.collections4.CollectionUtils;
import org.apache.commons.math3.ml.neuralnet.sofm.util.ExponentialDecayFunction;
-import org.apache.commons.math3.stat.inference.TestUtils;
import org.junit.Assert;
import org.junit.Test;
@@ -32,7 +30,7 @@ public class SamplingTest {
Granularity granularity = new Granularity(GranularityType.METHOD, "function");
for (int i = 0; i < 2000; i++) {
Assert.assertTrue(sampling.samplingDecision(granularity, 0));
- sampling.addSampledItem(granularity, 0);
+ sampling.addSampledItem(granularity, 0, 0);
}
Assert.assertEquals(2000, sampling.getSample().getTotalItems());
Assert.assertEquals(2000, sampling.getPopulation().getTotalItems());
@@ -44,7 +42,7 @@ public class SamplingTest {
Granularity granularity = new Granularity(GranularityType.METHOD, "function");
for (int i = 0; i < 2000; i++) {
Assert.assertTrue(sampling.samplingDecision(granularity, 0));
- sampling.addSampledItem(granularity, 0);
+ sampling.addSampledItem(granularity, 0, 0);
}
Assert.assertTrue(sampling.isReady());
}
@@ -55,17 +53,17 @@ public class SamplingTest {
for (int i = 0; i < 1000; i++) {
Granularity granularity = new Granularity(GranularityType.METHOD, "function");
Assert.assertTrue(sampling.samplingDecision(granularity, i));
- sampling.addSampledItem(granularity, 0);
+ sampling.addSampledItem(granularity, 0, 0);
}
for (int i = 0; i < 1000; i++) {
Granularity granularity = new Granularity(GranularityType.METHOD, "function2");
Assert.assertTrue(sampling.samplingDecision(granularity, i));
- sampling.addSampledItem(granularity, 0);
+ sampling.addSampledItem(granularity, 0, 0);
}
for (int i = 0; i < 1000; i++) {
Granularity granularity = new Granularity(GranularityType.METHOD, "function3");
Assert.assertTrue(sampling.samplingDecision(granularity, i));
- sampling.addSampledItem(granularity, 0);
+ sampling.addSampledItem(granularity, 0, 0);
}
Assert.assertTrue(sampling.isReady());
}
