[KIEKER-1841] Fixing Kieker Usage and Tutorial Examples

[rju]

JIRA Issue: KIEKER-1841 Fixing Kieker Usage and Tutorial Examples Original Reporter: Reiner Jung

---

These examples must be ported to gradle 6.6/6.7. However, not all of them are still relevant and some might be better moved elsewhere. Also there are examples missing for, e.g., the new pipe and filter framework.

List of general examples

• opad-anomaly-detection (keep for now for documentation purposes, but R bridge is not working properly,, maintenance issues)
• trace-analysis (could be removed. New analysis should use the new filter. Solution porting)
• JavaEEServletContainerExample (This is actually not JavaEE, but just Java servlet. Could be moved in a separate project which would also help to avoid binaries in git. Discussion necessary)
• adaptive-monitoring (needs update, also our API needs documentation)
• custom-timesource
• probe-aspectj
• probe-manual
• probe-spring
• sampler-jvm
• sampler-sigar (must be replaced by oshi)
• OverheadEvaluationMicrobenchmark (should not be part of the examples, I think this should be part of MooBench project only. As it is a performance benchmark, maybe it should be a separate git repository or at least in a separate subproject. Discussion necessary)
• SPECjEnterprise2010 (separate project)

Userguide

• appendix-AMQP
• appendix-Kafka
• ch2--bookstore-application
• ch2--manual-instrumentation
• ch3-4--custom-components
• ch5--trace-monitoring-aspectj
• appendix-JMS
• appendix-Sigar (sigar is obsolete, port to oshi)

[rju]

author rju -- Tue, 3 Nov 2020 13:56:53 +0100

Please note that all examples still appear in the distribution, but they are unchecked.

A key issue with our examples is that they should be build after jars have been compiled and put in a local repository for gradle or maven to find. Then subsequently, each example should be build without copy tricks. This is even more an issue when we use third party libraries which we have then to provide in our distribution instead of fetching them from their respective repository.

Current model:

• special libs which we have to update by hand and sync with automatically updated libraries in the normal part of Kieker.
• libraries are copied based on regex into lib folders.
• Example is test compiled.
• Example is test executed.
• Example is cleaned.
• Example is packed into the result bundles of Kieker.

Issues with that approach:

• We have to add jars to a lib folder in our git repository which are not necessary to be there when we would use normal dependency resolving mechanisms.
• Users of the examples have to copy files when they get our bundle and want to try it out instead of resolving it via maven/gradle
• The copy and clean mechanism is complex comprises 92 tasks all doing something similar, but using different configurations and setups.

[rju]

author rju -- Thu, 5 Nov 2020 08:49:35 +0100

• opad-anomaly-detection (delete), make ticket
• userguide sigar (delete), make ticket
• moobench goes to moobench repo.