Resilience Probe evaluation fail when source is configured

[rogeriofbrito]

What happened: When a cmd probe with a source is configured, the probe evaluation fails.

What you expected to happen: When a cmd probe with a source mode is evaluated, the command is executed inside specified container in source and pass or fail (according the command output).

Where can this issue be corrected? (optional): probably in Chaos Engine creation.

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

1. Create a new resilience probe of type CMD

2. In Properties section, set:

   • Timeout: 60s
   • Interval: 3s

3. In Probe Details section, set:
   • Command: k6 version
   • Data Comparison
     • Type: String
     • Comparison Criteria: contains
     • Value: 0.51.0
   • Source:

     image: grafana/k6:0.51.0
     imagePullPolicy: Always
     privileged: true
     hostNetwork: false

4. Run any experiment with this probe attached. In my scenario, I ran a pod-delete experiment.

Expected probe saved in MongoDB:

Expected result after run experiment:

Expected error in pod experiment runner logs:

time="2024-06-05T15:22:40Z" level=info msg="Experiment Name: pod-delete"
time="2024-06-05T15:22:40Z" level=info msg="[PreReq]: Getting the ENV for the pod-delete experiment"
time="2024-06-05T15:22:42Z" level=info msg="[PreReq]: Updating the chaos result of pod-delete experiment (SOT)"
time="2024-06-05T15:22:44Z" level=info msg="The application information is as follows" Targets="[{namespace: order-api-app, kind: deployment, labels: [app=order-api-app]}]" Chaos Duration=15
time="2024-06-05T15:22:46Z" level=info msg="[Probe]: The cmd probe information is as follows" Mode=Edge Phase=PreChaos Name=smoke-test-order-api-native Command="k6 version" Comparator="{string contains v0.51.0}" Source="<nil>" Run Properties="{60s 3s 0 0  0    false}"
time="2024-06-05T15:22:46Z" level=info msg="name: smoke-test-order-api-native, err: {\"errorCode\":\"CMD_PROBE_ERROR\",\"reason\":\"unable to run command: \",\"target\":\"{name: smoke-test-order-api-native}\"}"
time="2024-06-05T15:22:46Z" level=error msg="Probe Failed, err: probes failed\n --- at /litmus-go/pkg/probe/probe.go:339 (execute) ---\nCaused by: {\"errorCode\":\"CMD_PROBE_ERROR\",\"reason\":\"unable to run command: \",\"target\":\"{name: smoke-test-order-api-native}\"}"

Anything else we need to know?:

• In this thread there is a discussion with more details about the problem.

• Probabaly the problem is ocurring due to an error when the Chaos Engine is created without the probe source information. The image bellow shows the Chaos Engine created to a pod-delete experiment with a probe configured with a source. Note that source information is missing.

• All tests was execured in Litmus 3.6.0 and Litmus Agent 3.6.0, both installed using helm charts in a local kind k8s cluster. The docker environment was launched by rancher-desktop/qemu on Macbook M2 pro.