Unable to Execute In-Container JVM/Kafka Experiments in Latest Versions of ChaosBlade

[ramgiteng]

Issue

Unable to Execute In-Container JVM/Kafka Experiments in Latest Versions of ChaosBlade

Issue Description

With the latest versions of Chaosblade, I am unable to execute any Chaos experiments on a Java process that is running inside another Docker container. This applies to Kafka and JVM experiments.

Type: Bug report

Describe what happened (or what feature you want)

Setup 1: Chaosblade executable existing in a running Docker container
Setup 2: Java process running in a separate Docker container

After executing the following command in Container 1,
./blade create cri jvm --chaosblade-override true --chaosblade-release <pathToChaosBladeTarGZ> cpufullload --process java --container-name javaappcontainer --timeout 30s

Error: {"code":63067,"success":false,"error":"execContainer: container exec failed, err: exit status 1"}
Logs:
time="2024-07-07 07:37:25.976539673 EDT" level=info msg="exec container cmd: /home/<user>/chaosblade/chaosblade-1.7.4/bin/nsexec -t 3565 -p -m -n -- /bin/sh -c [ -e /opt/chaosblade/blade ] && echo True || echo False" location="/go/pkg/mod/github.com/chaosblade-io/[email protected]/exec/container/container_linux.go:93" uid=d0d348dda8a5c0ec

time="2024-07-07 07:37:25.97919496 EDT" level=info msg="run copy cmd: /home/<user>/chaosblade/chaosblade-1.7.4/bin/nsexec -t 3565 -p -m -- /bin/sh -c cat > /opt/chaosblade-1.7.4-linux-amd64.tar.gz" location="/go/pkg/mod/github.com/chaosblade-io/[email protected]/exec/container/container_linux.go:40" uid=d0d348dda8a5c0ec

time="2024-07-07 07:37:25.980804861 EDT" level=error msg="DeployChaosBlade err: exit status 2" location="/go/pkg/mod/github.com/chaosblade-io/[email protected]/exec/executor_execin.go:104" uid=d0d348dda8a5c0ec

This issue has been observed in versions after v1.5.0. I tested with the newly released v1.7.4 and still faced the same issue.

Using Chaosblade v1.5.0, I am able to execute in-container experiments on JVM and Kafka successfully.

Describe what you expected to happen

To successfully perform Cri (Docker) experiments, including JVM and Kafka, on JVMs running inside another Docker container and respond with a success "true" message.

Example: {"code":200,"success":true,"result":"55e49e9a472103b5"}

How to reproduce it (as minimally and precisely as possible)

1. Java app (Any JDK type/version)
2. Docker
3. Chaosblade versions: v1.7.2/v1.7.3/v1.7.4

Tell us your environment

1. Java app (Any JDK type/version)
2. Docker
3. Chaosblade versions: 1.7.2/1.7.3/1.7.4

Anything else we need to know?

• I believe that the issue seems to be related to the binary files (in /bin folder) that blade executable uses for copying the blade tar file to target Docker container running the Java process

• In Chaosblade v1.7.4, the bin folder contains the following:
  

• In Chaosblade v1.5.0, the bin folder contains the following:
  

Could you please help troubleshoot this issue?

Please let me know if you require any additional details and information. I will be happy to provide.

Thank you for the support.