Closed diamonwiggins closed 5 months ago
More examples in the docs would be very helpful!
The only thing we're planning to use clusterResource
for so far is to verify that resources exist at all - I don't think there's a default way to do that (at least not a documented one), so we're doing weird things like...
- clusterResource:
checkName: knime-postgres-cluster
kind: Service
namespace: knime
name: knime-postgres-cluster
yamlPath: "metadata.name"
regex: knime-postgres-cluster
outcomes:
- fail:
when: "false"
message: knime-postgres-cluster service is missing.
- pass:
when: "true"
message: knime-postgres-cluster service is present.
...which essentially just checks that the .metadata.name
property matches the resource.
The main goal is to simply ensure that expected resources are present in the cluster. In some cases, particularly if a helm chart fails to install altogether, the support bundle will not detect that a resource is missing from the cluster.
Hey @diamonwiggins,
Could you help me to review this PR #1547? Thx!
Describe the rationale for the suggested feature.
Documentation Improvements
Currently, the ClusterResource analyzer only has a single example which is based on a
pvc
. More examples with different types of objects could help those using the analyzer. Also, for thekind
property it isn't documented which resources theClusterResource
analyzer supports. Lastly, for the resources that are supported, the naming isn't consistent. Some are lowercase and others are upper case emphasizing how important it is to have better documentation.https://troubleshoot.sh/docs/analyze/cluster-resource/
Code Improvements
The naming for the resources shouldn't be case sensitive