-
Notifications
You must be signed in to change notification settings - Fork 220
pkg/checkpoint: Try kubelet secureClient, fallback to read-only #1027
pkg/checkpoint: Try kubelet secureClient, fallback to read-only #1027
Conversation
Can one of the admins verify this patch? |
2 similar comments
Can one of the admins verify this patch? |
Can one of the admins verify this patch? |
/assign @rphillips |
Is there a default timeout on the secure kubeclient? |
I don't see one set in client-go |
* Change pod-checkpointer to first try to list pods via the Kubelet secure (10250) endpoint, then fallback to the Kubelet read-only (10255) endpoint, before defaulting to an empty list of parent pods
/ok-to-test |
I've been testing with an image On a typical bootkube cluster today (allows kubelet read-only, no ClusterRoleBinding for pod-checkpointer), the checkpointer falls back to the insecure client as usual.
On a cluster with
With this pod-checkpointer change:
I think this shows the change doesn't regess any current behavior, but adds a way to support disabling the Kubelet read-only port. I can have #1025 be a followup when a pod-checkpointer image is ready. |
coreosbot run e2e checkpointer |
/approve |
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: rphillips The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
/lgtm |
Pushed the new checkpoint image to quay (tag: |
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since poseidon/typhoon#324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since #324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since poseidon/typhoon#324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since poseidon/typhoon#324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since poseidon/typhoon#324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since poseidon/typhoon#324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since poseidon#324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
* Updates pod-checkpointer to prefer the Kubelet secure API (before falling back to the Kubelet read-only API that is disabled on Typhoon clusters since poseidon/typhoon#324) * Previously, pod-checkpointer checkpointed an initial set of pods during bootstrapping so recovery from power cycling clusters was unaffected, but logs were noisy * kubernetes-retired/bootkube#1027 * kubernetes-retired/bootkube#1025
Change pod-checkpointer to first try to list pods via the Kubeletsecure (10250) endpoint, then fallback to the Kubelet read-only (10255) endpoint, before defaulting to an empty list of parent pods
Related: #1025