🐛 clusterctl init fails with 'failed to get cert-manager object'

[killianmuldoon]

clusterctl init is failing with an error 'failed to get cert-manager object'. This is due to cert-manager changing its repo location (from github.com/jetstack to github.com/cert-manager).

The solution for this issue is to add the new URL to the clusterctl config file, as described here, adding:

cert-manager:
  url: "https://github.com/cert-manager/cert-manager/releases/latest/cert-manager.yaml"

Alternatively users can use the clusterctl overrides folder to install the cert-manager from a yaml as described here

/kind bug

[fabriziopandini]

/milestone v1.2
This is due to the GitHub client, that uses owner and repository from the above URL but it doesn't detect the URL redirect.

Let's change the URL as a quick fix, but long term we should implement URL redirect check in clusterctl, eventually rewriting the owner/repository passed to the GitHub client.

[fabriziopandini]

(fix to be backported up to 0.4)

[killianmuldoon]

I'll open another issue around the github client once the fix is in.

[sbueringer]

Just fyi. The DownloadRelesaeAsset func of the GitHub client is actually following redirects. But I think the issue is that we are retrieving a release (including assets) from a release of the new location and then passing in the owner of the old location.

But the fix stays the same, we have to detect the change and pass the right owner/repository through.