fix: wait namespace is actually deleted (#413) #422
Conversation
|
How can we help here? Without this fix it's causing test failures on our side. See kyverno/kyverno#5413. |
|
@chipzoller im waiting for the CI PR to merge so that I can investigate better. |
|
I wonder if this happens because k8s components are not deployed in the cluster, therefore namespace finalizers are never processed. |
|
yeah... the integration-tests are just against etcd and api-server (no controllers). We will need to look into a solution around that. The e2e tests should be against a kind cluster but don't appear to be... need to look into why on that as well. I really like this PR and will work to work it in. |
Co-authored-by: Erik Godding Boye <egboye@gmail.com> Signed-off-by: Charles-Edouard Brétéché <charles.edouard@nirmata.com>
kensipe
force-pushed
the
ken/ns-delete2
branch
from
e40ded7
to
be31307
Compare
Signed-off-by: Ken Sipe <kensipe@gmail.com>
|
@eddycharly as suspected... when running against testenv, we can't delete namespaces and need to skip-delete. Testing all looked good. plan to merge after CI. Thanks for this great work! |
Originally on #413 was merged and caused failures. We have issues with contributor PR being tested on circle ci. This is the same PR. The squashed commit was reverted and this PR is a cherry-pick of that commit
Co-authored-by: Erik Godding Boye egboye@gmail.com
Signed-off-by: Charles-Edouard Brétéché charles.edouard@nirmata.com
What this PR does / why we need it:
Fixes #