Repository navigation
fix: wait for node Ready before completing bootstrap - #132
Draft
bschimke95 wants to merge 2 commits into
Draft
bschimke95 wants to merge 2 commits into
bschimke95 wants to merge 2 commits into
Conversation
bschimke95
force-pushed
the
fix/join-wait-for-bootstrap-node-registration
branch
from
October 6, 2026 16:04
b4a0f58 to
96d16b7
Compare
onBootstrapControlPlane starts the control-plane services and returns success once the kube-apiserver responds, but never waits for this node's own kubelet to register its Kubernetes Node object. That registration is asynchronous, so a node joining the cluster right after bootstrap can see zero Node objects and fail with "the cluster has no nodes - cannot determine cluster Kubernetes version" in handleNoUpgradeInProgress, even though the cluster is healthy. Wait for this node to report Ready before returning from bootstrap, so bootstrap only completes once the cluster actually has a node.
bschimke95
force-pushed
the
fix/join-wait-for-bootstrap-node-registration
branch
from
October 7, 2026 08:38
96d16b7 to
4d08e49
Compare
bschimke95
changed the title
There was a problem hiding this comment.
🟡 Changes recommended
Waiting for Ready blocks supported bootstrap configurations without managed networking.
1 open finding
What changed in this PR
Adds bootstrap synchronization to prevent joins before the initial Kubernetes node registers.
Changes:
- Waits for a Ready node after API server startup.
- Adds readiness polling and tests.
| File | Description |
|---|---|
pkg/k8sd/app/hooks_bootstrap.go |
Adds the bootstrap readiness wait. |
pkg/k8sd/app/cluster_util.go |
Implements node readiness polling. |
pkg/k8sd/app/cluster_util_test.go |
Tests successful waiting and timeout behavior. |
🧠 Review effort: Balanced
Give feedback about Copilot approvals in this survey to enter a drawing for a $150 gift card.
bschimke95
force-pushed
the
fix/join-wait-for-bootstrap-node-registration
branch
2 times, most recently
from
October 8, 2026 14:41
0adb645 to
7fe5dec
Compare
Waiting for the Ready condition required a working CNI, so bootstrap with managed networking disabled (a supported, documented configuration) would block for the full timeout and fail. It also raced NotifyFeatureController, the only thing that installs the CNI in the common case, creating an ordering dependency on an unrelated restart-recovery code path instead of a deterministic sequence. The race this closes only needs a joining node to see at least one Node object (NodeVersions() in hooks_join.go lists all Node objects), not for that node to be Ready. Wait for registration instead: it doesn't depend on CNI/NotifyFeatureController at all, so it also removes the ordering dependency. Addresses review feedback from @copilot-pull-request-reviewer on this PR.
bschimke95
force-pushed
the
fix/join-wait-for-bootstrap-node-registration
branch
from
October 8, 2026 18:37
7fe5dec to
75c9f72
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
onBootstrapControlPlanestarts the control-plane services and returns success once kube-apiserver responds, but never waits for this node's own kubelet to register its KubernetesNodeobject. That registration is asynchronous, so a node joining the cluster right after bootstrap can see zeroNodeobjects and fail inhandleNoUpgradeInProgresswith "the cluster has no nodes - cannot determine cluster Kubernetes version", even though the cluster is healthy. Seen in https://github.com/canonical/k8s-snap/actions/runs/36700416543/job/109844096943.