Closed glrf closed 2 years ago
First investigation on our instance on lieutenant-dev
Lieutenant-dev was deployed with image tag latest
, which pointed to https://hub.docker.com/layers/projectsyn/lieutenant-operator/latest/images/sha256-23de5bff707464e9068404179f247de4411441b22409e23f84e5179a7bb34b9d?context=explore. This image was built before the Operator SDK upgrade merged in #175 . So the bug report refers to a version that should have been tested with the master
image tag (which gets pushed with every merge commit to master).
When deploying with the master
tag, the mentioned bug doesn't appear. However, there are a number of other errors logged in the operator. Such errors are like "can't find Vault Secret" or "finalizer not removed".
It's currently unclear whether the Dev environment is simply outdated and should be cleaned manually or if the new version cannot handle the old data. Both cases do not give a good gut feeling to deploy the operator to Int or production environments.
It's currently a mess. I will try to isolate the errors and open new issues. It's currently not possible to reproduce this one.
When deleting a cluster the operator panics because of a nil pointer deference, while trying to remove the steward secret.
Steps to Reproduce the Problem
It is unclear which steps exactly lead to this. The general steps where:
Actual Behavior
The operator panics while trying to handle the deletion of a cluster.
Cluster Resource
Vault Secret
The original secret should still be present on
vault-int
.Log
Expected Behavior
The operator is able to handle the cluster deletion or return an error without crashing if the cluster resource is in an inconsistent state.