fix: Resolve k8sd-proxy-related reconciliation errors #146
Conversation
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
claudiubelu
force-pushed
the
reconciliation
branch
3 times, most recently
from
f7fadeb to
fd24930
Compare
claudiubelu
changed the title
| notReadyErr *K8sdProxyNotReady | ||
| ) | ||
| if errors.As(err, ¬FoundErr) || errors.As(err, ¬ReadyErr) { | ||
| return ctrl.Result{RequeueAfter: 30 * time.Second}, nil |
There was a problem hiding this comment.
Why choose 30 seconds? It seems quite long. How long does the service usually take to become ready?
There was a problem hiding this comment.
To be fair, the value is a bit arbitrary. We could lower it to something like 15 seconds.
The k8sd-proxy pods can take quite a bit to spawn. Based on a test's run (in a test run that didn't have this PR), we can see in the Certificates Controller Reconciler, we can see how many k8sd-proxy related Reconciler errors there are, and how long it can take until they're no longer an issue (from: https://github.com/canonical/cluster-api-k8s/actions/runs/14885411990/job/41834498342)
2025-05-07T22:45:51Z ERROR Reconciler error {"controller": "ck8sconfig", "controllerGroup": "bootstrap.cluster.x-k8s.io", "controllerKind": "CK8sConfig", "CK8sConfig": {"name":"worker-md-0-cld5g-7dvb9","namespace":"workload-cluster-certificate-refresh-u270u9"}, "namespace": "workload-cluster-certificate-refresh-u270u9", "name": "worker-md-0-cld5g-7dvb9", "reconcileID": "59ed7295-ad38-44ee-9119-1268fc2955b4", "error": "failed to request join token: failed to create k8sd proxy: failed to get proxy pods: there isn't any k8sd-proxy pods in target cluster"}
.
.
.
2025-05-07T22:46:02Z ERROR Reconciler error {"controller": "ck8sconfig", "controllerGroup": "bootstrap.cluster.x-k8s.io", "controllerKind": "CK8sConfig", "CK8sConfig": {"name":"worker-md-0-cld5g-7dvb9","namespace":"workload-cluster-certificate-refresh-u270u9"}, "namespace": "workload-cluster-certificate-refresh-u270u9", "name": "worker-md-0-cld5g-7dvb9", "reconcileID": "1e4eade8-44be-41bd-967d-b57e581e9ea1", "error": "failed to request join token: failed to create k8sd proxy: failed to get k8sd proxy for control plane, previous errors: pod 'k8sd-proxy-w626x' is not Ready"}
.
.
.
2025-05-07T22:46:34Z ERROR Reconciler error {"controller": "machine", "controllerGroup": "cluster.x-k8s.io", "controllerKind": "Machine", "Machine": {"name":"worker-md-0-cld5g-7dvb9","namespace":"workload-cluster-certificate-refresh-u270u9"}, "namespace": "workload-cluster-certificate-refresh-u270u9", "name": "worker-md-0-cld5g-7dvb9", "reconcileID": "a78fbd92-b84c-400a-b3c2-d25e5ebd4207", "error": "failed to get certificates expiry date: failed to create k8sd proxy: missing k8sd proxy pod for node capick8s-certificate-refresh-opv64c-worker-md-0-cld5g-7dvb9"}
.
.
.
2025-05-07T22:47:37Z ERROR Reconciler error {"controller": "machine", "controllerGroup": "cluster.x-k8s.io", "controllerKind": "Machine", "Machine": {"name":"worker-md-0-cld5g-7dvb9","namespace":"workload-cluster-certificate-refresh-u270u9"}, "namespace": "workload-cluster-certificate-refresh-u270u9", "name": "worker-md-0-cld5g-7dvb9", "reconcileID": "ddb63da9-22b4-4183-a7e3-c20a47fc0f05", "error": "failed to get certificates expiry date: failed to create k8sd proxy: missing k8sd proxy pod for node capick8s-certificate-refresh-opv64c-worker-md-0-cld5g-7dvb9"}
.
.
.
2025-05-07T22:47:51Z DEBUG events Certificates refresh in progress. TTL: 1y {"type": "Normal", "object": {"kind":"Machine","namespace":"workload-cluster-certificate-refresh-u270u9","name":"worker-md-0-cld5g-7dvb9","uid":"b96ba028-7986-41da-9f52-9b9df1b6f7cc","apiVersion":"cluster.x-k8s.io/v1beta1","resourceVersion":"1879"}, "reason": "CertificatesRefreshInProgress"}
2025-05-07T22:47:51Z INFO controllers.Certificates Certificates refreshed {"namespace": "workload-cluster-certificate-refresh-u270u9", "machine": "worker-md-0-cld5g-7dvb9", "cluster": "capick8s-certificate-refresh-opv64c", "machine": "worker-md-0-cld5g-7dvb9", "expiry": "2026-05-07T22:47:51Z"}
...
So, 2 minutes, more or less.
|
Bootstrap controller manager logs without this change: https://paste.ubuntu.com/p/C4rftSV4mS/ |
claudiubelu
force-pushed
the
reconciliation
branch
2 times, most recently
from
4cdd44f to
c9617c8
Compare
Currently, the cluster-api-k8s controllers are being filled with reconciliation errors because the k8sd-proxy does not yet exist on a node, or it's not yet Ready. On error, the reconciliation request is put on an exponential backoff queue, which results in those requests being solved later and later. This can cause delays in various CAPI-related operations, such as scaling the number of nodes (requesting join tokens), certificate refreshes, and so on. In case of a k8sd-proxy related error (Pod does not yet exist, or it's not Ready), we're now deferring the request.
claudiubelu
force-pushed
the
reconciliation
branch
from
c9617c8 to
8bbcf07
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Currently, the
cluster-api-k8scontrollers are being filled with reconciliation errors because thek8sd-proxydoes not yet exist on a node, or it's not yetReady.On error, the reconciliation request is put on an exponential backoff queue, which results in those requests being solved later and later. This can cause delays in various CAPI-related operations, such as scaling the number of nodes (requesting join tokens), certificate refreshes, and so on.
In case of a
k8sd-proxyrelated error (Pod does not yet exist, or it's notReady), we're now deferring the request.