201403-ICPE-Dublin

Information

The 2-hour tutorial was held on March 23, 2014 at the ` 5th ACM/SPEC International Conference on Performance Engineering <http://icpe2014.ipd.kit.edu/>`_ in Dublin, Irland.

Presenters:

1.  Andre van Hoorn, University of Stuttgart, Germany

2.  Nils Ehmke, Kiel University, Germany

For this tutorial, Kieker version 1.8 has been used.

Material provided outside this Wiki page:

• Kieker and Kieker.WebGUI release archives (binary and doc): http://kieker-monitoring.net/download/

• Kieker user guide: kieker-1.8_userguide.pdf

• Presentation slides:

  1. Original slides with overlays: 20140323-ICPE-Tutorial-slides.pdf

  2. Handout versions: 20140323-ICPE-Tutorial-handout.pdf

Schedule

• 14:30 Introduction and Overview of Approach (avh)

• 14:45 Interactive: Quick Start (nie)

• 15:00 Intro (cont’d) (avh)

• 15:05 Use Cases in Research and Practice (avh)

• 15:20 Kieker’s Monitoring Component (avh)

• 15:35 Kieker’s Analysis Component & WebGUI (nie)

• 15:50 Interactive: Java EE Monitoring (avh)

• 16:05 A Detailed Look at Selected Use Cases (nie/avh)

• 16:30 End

Instructions for the Interactive Parts

It makes sense to put the following parts to separate pages.

Preparation

1. Unzip source and binary release archives

   20140323-ICPE-Tutorial/kieker-1.8-release <master>* $ unzip kieker-1.8_binaries.zip && mv kieker-1.8 kieker-1.8-bin

   20140323-ICPE-Tutorial/kieker-1.8-release <master>* $ unzip kieker-1.8_sources.zip && mv kieker-1.8 kieker-1.8-src

1. Start Eclipse with new/empty workspace 20140323-ICPE-Tutorial/workspace-avh

Quick Start (nie)

• Preparation:

  • Create shortlink to directory 20140323-ICPE-Tutorial in Konqueror

  • Create new workspace in eclipse, located in 20140323-ICPE-Tutorial/eclipse-workspace

• Download -> s. Folie

  • Files to be found in dir: 20140323-ICPE-Tutorial/download-kieker-1.8/

  • User Guide vorstellen

  • Extract kieker-1.8_binaries.zip to 20140323-ICPE-Tutorial/

  • … and rename to kieker-1.8-binaries

• Presentation of Release Archive (Binary, Source, User Guide)

• Bookstore-Beispiel mit AspectJ aus dem User Guide

  1. Bookstore ohne Instrumentierung

     1. Import example project ch2–bookstore-application into workspace

     2. Bookstore zunächst kurz vorstellen

     3. Starten ohne Instrumentierung: BookstoreStarter.java -> run as -> Java application

  2. Instrumentierung mit AspectJ (Vollinstrumentierung ohne Annotationen)

     1. Create directory lib/

     2. Copy kieker-1.8_aspectj.jar

     3. Create directory src/META-INF

     4. Copy aop.example.xml to src/META-INF/aop.xml

     5. Select full aspect and include “*”

     6. run with -javaagent:lib/kieker-1.8_aspectj.jar

  3. Filesystem-Log

     1. Show console output

     2. Show .map and .dat file in Konqueror

  4. TraceAnalyse mit Visualisierung

     1. Add kieker-1.8_emf.jar to build path

     2. Add commons-cli-1.2.jar to build path

     3. Run *TraceAnalysisTool* (run kieker-1.8_emf.jar as “Java Application”)

     4. Run configuration with program arguments:

        • -i /tmp/kieker-20121121-150448579-UTC-pc-vanhoorn-KIEKER-SINGLETON

        • –plot-Deployment-Operation-Dependency-Graph

        • -o .

     5. Show output

        1. HTML-Systemmodell

        2. Abhaengigkeitsgraph (Deployment-Operation + ggf. mehr)

        3. Call-Tree-Varianten

        4. Sequenzdiagramm -> “if possible”

Monitoring (avh)

1. Import Kieker source project (20140323-ICPE-Tutorial/kieker-1.8-release/kieker-1.8-src/) into the workspace (“Import…”->”Import existing project into workspace”)

2. Show example record CPUUtilizationRecord

3. Show example probe CPUsDetailedPercSampler

Analysis (nie)

1. …

WebGUI (nie)

1. Mem/swap example without CPU (to be prepared)

2. Open analysis editor

3. Explain example

4. Add CPU filter

5. Start analysis

6. Cockpit

JavaEE (avh)

Quick start

1. Change dir to prepared Jetty 20140323-ICPE-Tutorial/kieker-1.8-release/kieker-1.8-bin/examples/JavaEEServletContainerExample/jetty-hightide-jpetstore

2. Start instrumented JPetStore

   jetty-hightide-jpetstore <master>* $ java -jar start.jar

3. Explain console output and tail -f on monitoring log

4. Access JPetStore http://localhost:8080/jpetstore/ and click around

5. Create and show plots

   jetty-hightide-jpetstore <master>* $ mkdir plots

   jetty-hightide-jpetstore <master>* $ ../../../bin/trace-analysis.sh -i /tmp/kieker-20140319-150803890-UTC-avh-ThinkPad-RSS-KIEKER-EXAMPLE-JAVAEE/ -o plots/ –plot-Deployment-Component-Dependency-Graph –plot-Assembl

   y-Component-Dependency-Graph –plot-Deployment-Operation-Dependency-Graph responseTimes –plot-Assembly-Operation-Dependency-Graph responseTimes –print-System-Model

   jetty-hightide-jpetstore <master>* $ ../../../bin/dotPic-fileConverter.sh plots/ pdf

   jetty-hightide-jpetstore <master>* $ acroread plots/*.pdf&

Advanced

1. Explain instrumentation (Spring, Servlet)

2. Use a custom Kieker configuration:

   1. Copy META-INF from binary release to JavaEE example

      jetty-hightide-jpetstore <master>* $ cp -R ../../../META-INF/ .

   2. Edit kieker.monitoring.properties:

      1. hostname=ICPE14-SRV

      2. jmx=true

      3. adaptiveMonitoring.enabled=true

      4. kieker.monitoring.writer.filesystem.AsyncFsWriter.customStoragePath=kieker-logs

   3. Create output dir

      jetty-hightide-jpetstore <master>* $ mkdir kieker-logs

   4. Activate configuration in start.ini

      • –exec

      • -Dkieker.monitoring.configuration=META-INF/kieker.monitoring.properties

3. Sigar Sampler for CPU and MEM

   1. Copy Sigar Jar and {dll|so|…} to webapps/WEB-INF/lib/

      jetty-hightide-jpetstore <master> $ cp ../../../lib/sigar-1.6.4.jar ../../../lib/sigar-native-libs/* webapps/jpetstore/WEB-INF/lib/

   2. Activate kieker.monitoring.probe.servlet.CPUMemUsageServletContextListener in webapps/jpetstore/WEB-INF/web.xml

1. Restart Jetty

2. Click around a bit

3. Show log

1. Attach to Monitoring Controller via JConsole:

   /usr/lib/jvm/sun-jdk1.6.0_38/bin/jconsole &

   1. toString

   2. Demonstrate adaptive Monitoring

      1. Disable public void kieker.monitoring.probe.servlet.SessionAndTraceRegistrationFilter.doFilter(javax.servlet.ServletRequest, javax.servlet.ServletResponse, javax.servlet.FilterChain)

   3. enable/disable/terminate

   4. Restart Jetty

   5. Click around

Bonus:

1. Sigar analysis from User Guide

   1. Fix project

   2. Add kieker.jar and sigar.jar to build path

   3. run

2. AspectJ-based instrumentation