kubectl rabbitmq perf test improvements

[MirahImage]

Summary Of Changes

Deploys perf-test and stream-perf-test as Kubernetes Jobs, rather than running pods directly.

Additional Context

This is designed to make clean-up easier.

[mkuratczyk]

For me, the plugin stopped working with these changes:

$ kubectl rabbitmq create foo
error: Unexpected args: []
See 'kubectl apply -h' for help and examples

# after deploying an instance with a version from main
$ kubectl rabbitmq perf-test foo
error: the server doesn't have a resource type ""

[Zerpet]

Works for me

[coro]

A couple questions from me:

What's the expected method of sourcing prometheus metrics from perf test in this case? Previously there was a service created - do you get the same output from scraping the pods directly?

Can we also hide the output from the pushd/popd as well?
New output:

$ ./bin/kubectl-rabbitmq perf-test billing --rate 100
/var/folders/gm/mgfz1ccs4t7_tbh12qfkprxm0000gp/T/tmp.11LqVZJN ~/GolandProjects/cluster-operator
job.batch/perf-test created
~/GolandProjects/cluster-operator

compared with old:

$ k rabbitmq perf-test billing --rate 100
service/perf-test created
pod/perf-test created

The perf-test command works fine, but the stream one doesn't out of the box. I've checked on different versions of RMQ - it seems the one on main doesn't hit this issue, but this branch does:

k logs stream-perf-test-zpwfd
java.lang.IllegalArgumentException: Wrong scheme in rabbitmq-stream URI: stream. Should be rabbitmq-stream or rabbitmq-stream+tls
	at com.rabbitmq.stream.impl.StreamEnvironmentBuilder.toUri(StreamEnvironmentBuilder.java:74)
	at java.base/java.util.stream.ReferencePipeline$3$1.accept(ReferencePipeline.java:197)
	at java.base/java.util.ArrayList$ArrayListSpliterator.forEachRemaining(ArrayList.java:1625)
	at java.base/java.util.stream.AbstractPipeline.copyInto(AbstractPipeline.java:509)
	at java.base/java.util.stream.AbstractPipeline.wrapAndCopyInto(AbstractPipeline.java:499)
	at java.base/java.util.stream.ReduceOps$ReduceOp.evaluateSequential(ReduceOps.java:921)
	at java.base/java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:234)
	at java.base/java.util.stream.ReferencePipeline.collect(ReferencePipeline.java:682)
	at com.rabbitmq.stream.impl.StreamEnvironmentBuilder.uris(StreamEnvironmentBuilder.java:98)
	at com.rabbitmq.stream.impl.StreamEnvironmentBuilder.uris(StreamEnvironmentBuilder.java:44)
	at com.rabbitmq.stream.perf.StreamPerfTest.call(StreamPerfTest.java:546)
	at com.rabbitmq.stream.perf.StreamPerfTest.call(StreamPerfTest.java:86)
	at picocli.CommandLine.executeUserObject(CommandLine.java:1953)
	at picocli.CommandLine.executeHelpRequest(CommandLine.java:1929)
	at picocli.CommandLine.executeHelpRequest(CommandLine.java:1909)
	at picocli.CommandLine$AbstractParseResultHandler.execute(CommandLine.java:2176)
	at picocli.CommandLine.execute(CommandLine.java:2078)
	at com.rabbitmq.stream.perf.StreamPerfTest.run(StreamPerfTest.java:382)
	at com.rabbitmq.stream.perf.StreamPerfTest.main(StreamPerfTest.java:362)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:77)
	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.base/java.lang.reflect.Method.invoke(Method.java:568)
	at org.springframework.boot.loader.MainMethodRunner.run(MainMethodRunner.java:49)
	at org.springframework.boot.loader.Launcher.launch(Launcher.java:108)
	at org.springframework.boot.loader.Launcher.launch(Launcher.java:58)
	at org.springframework.boot.loader.JarLauncher.main(JarLauncher.java:88)

using cluster-operator 1.10.0, example rmq:

apiVersion: rabbitmq.com/v1beta1
kind: RabbitmqCluster
metadata:
  name: billing
spec:
  replicas: 3
  image: rabbitmq:3.9-management
  rabbitmq:
    additionalPlugins:
    - rabbitmq_stream

Otherwise works well for me, and I like the change, thanks!

[MirahImage]

What's the expected method of sourcing prometheus metrics from perf test in this case? Previously there was a service created - do you get the same output from scraping the pods directly?

Unfortunately, creating a service would defeat the purpose of using a Job, namely, allowing the perf-test command to clean up after itself. Thankfully, it's possible to monitor the job using a podMonitor instead of a serviceMonitor, so I have updated the instructions with the relevant podMonitor.

[coro]

Sorry to be a pain; the stream perf test command now works but the pushd/popd output is still shown, and the echo statements are now also shown:

$ ./bin/kubectl-rabbitmq perf-test billing --rate 100
/var/folders/gm/mgfz1ccs4t7_tbh12qfkprxm0000gp/T/tmp.lzJyhNZe ~/GolandProjects/cluster-operator
+ echo 'apiVersion: batch/v1'
+ echo 'kind: Job'
+ echo metadata:
+ echo '  name: perf-test'
+ echo '  labels:'
...

[MirahImage]

I've removed the output, at the cost of more technically incorrect bash. Unfortunately, it seems that due to the nature of kubectl plugins, it's not possible to hide pushd/popd output, so we're forced to use cd, which should generally be avoided in scripts.