Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Followup to #242 This adds a new integration test that verifies that pods don't get scheduled until all the networking has been setup, making sure having linkerd-cni run run first doesn't trigger scheduling. Note that for now we just consider the linkerd-cni+calico combo. Worklow: 1. Install calico and linkerd-cni. 2. Label node and then add node selectors to calico and linkerd-cni to run only on the current node. 3. Create a new node. Calico nor linkerd-cni will be triggered in it. 4. Persist an nginx pod in that node. Verify it doesn't get scheduled. 5. Add label to linkerd-cni to have it trigger in the new node. 6. Verify the nginx pod still isn't scheduled. 7. Add label to calico to have it trigger in the new node. 8. Verify the nginx pod is finally scheduled. Note that currently the test fails at step 6, but will pass as soon #242 gets merged. In order to test succesfully, you can apply this patch: ```diff diff --git a/cni-plugin/integration/manifests/calico/linkerd-cni.yaml b/cni-plugin/integration/manifests/calico/linkerd-cni.yaml index 05f110b..419abdd 100644 --- a/cni-plugin/integration/manifests/calico/linkerd-cni.yaml +++ b/cni-plugin/integration/manifests/calico/linkerd-cni.yaml @@ -123,7 +123,7 @@ spec: # script copies the files into place and then sleeps so # that Kubernetes doesn't keep trying to restart it. - name: install-cni - image: test.l5d.io/linkerd/cni-plugin:test + image: ghcr.io/alpeb/cni-plugin:v1.2.8 #image: cr.l5d.io/linkerd/cni-plugin:edge-22.12.1 env: - name: DEST_CNI_NET_DIR ```
- Loading branch information