Fix flux e2e test

[beni0888]

Signed-off-by: Jesús Benito Calzada bjesus@vmware.com

Description of the change

The Flux e2e test is currently failing while selecting version 6.2.0 of the podInfo package to install. The reason of
the error is that the target version of the package to install is no longer available in the version selector. In this PR we
remove the part of the test that picks a specific application version, so it install the latest version available.

Benefits

• The Flux e2e test no longer fails.

Possible drawbacks

None

Applicable issues

N/A

Additional information

• Failing pipeline
• Recorded Video

[netlify]

✅ Deploy Preview for kubeapps-dev canceled.

Built without sensitive environment variables

Name	Link
🔨 Latest commit	6c85c43
🔍 Latest deploy log	https://app.netlify.com/sites/kubeapps-dev/deploys/636de6b75f609b00098903f3

[gfichtenholt]

ooops my bad. This is happening because we deploy a repo https://raw.githubusercontent.com/fluxcd/source-controller/main/config/samples/source_v1beta2_helmrepository.yaml, which points to https://stefanprodan.github.io/podinfo and apparently they periodically add new versions of the sample app. You can see it here There was no technical reason to point to 6.2.0, it just was what as out there at the time. So, your fix will work for now. I think a longer term fix would be to either:

• get rid of the 4 lines in the test that select a specific version to deploy (kind of like equivalent helm test does it with the apache package) OR
• not to depend on that helm repo and set up one of our own, say on ghcr.io. That way we are fully in charge of the contents of the repo. I have done something similar for flux integration tests in the past.

Easiest is probably option (1). Feel free to do that or I can do it

[beni0888]

ooops my bad. This is happening because we deploy a repo https://raw.githubusercontent.com/fluxcd/source-controller/main/config/samples/source_v1beta2_helmrepository.yaml, which points to https://stefanprodan.github.io/podinfo and apparently they periodically add new versions of the sample app. You can see it here There was no technical reason to point to 6.2.0, it just was what as out there at the time. So, your fix will work for now. I think a longer term fix would be to either:

• get rid of the 4 lines in the test that select a specific version to deploy (kind of like equivalent helm test does it with the apache package) OR
• not to depend on that helm repo and set up one of our own, say on ghcr.io. That way we are fully in charge of the contents of the repo. I have done something similar for flux integration tests in the past.

Easiest is probably option (1). Feel free to do that or I can do it

Ok, thanks for the explanation! I will go ahead with the first solution you proposed as it is way simpler.