Skip to content

AWS Load Balancer Controller: one change per PR-gate rule (14 rules) - #5

Open
pvishwa3 wants to merge 9 commits into
lbc-basefrom
lbc-pr
Open

pvishwa3 wants to merge 9 commits into
lbc-basefrom
lbc-pr

Conversation

@pvishwa3

@pvishwa3 pvishwa3 commented Nov 3, 2026

Copy link
Copy Markdown

Live test of the AWS Load Balancer Controller PR-gate rules in runtimez-com/deployment-risk-action@v1.0.9, which uses image 1.0.9.

Each file under k8s/lbc/ is one positive for one rule: 13 static/diff rules and target-annotation-overridden. The base branch lbc-base holds the "before" side of the 5 diff rules. The cluster checks read rtz-eks through the repo secrets.

Expected: all 14 wrs-addon-awslbc-* rules in the report. Cluster context should also add target-backend-service-missing (the TGB's Service does not exist) and target-readiness-gate-off (ip targets in a namespace without the readiness-gate label).

Known false positive, not fixed in 1.0.9: action-config-invalid fires on an actions.* annotation no backend references (L8).

🤖 Generated with Claude Code

https://claude.ai/code/session_014NPEGf1kUYALnqg7HFRtBH

@github-actions

github-actions Bot commented Nov 3, 2026 •

Copy link
Copy Markdown

Deployment risk: HIGH (score 61) — 51 findings

Severity Rule Resource Title
HIGH mrs-svc-selector-mismatch Service/pr3svc Service selector matches no workload — traffic black-hole
HIGH mrs-svc-selector-mismatch Service/nlb13 Service selector matches no workload — traffic black-hole
HIGH mrs-svc-selector-mismatch Service/dep15 Service selector matches no workload — traffic black-hole
HIGH mrs-svc-selector-mismatch Service/dep15-intree Service selector matches no workload — traffic black-hole
HIGH mrs-svc-selector-mismatch Service/dep16-custom Service selector matches no workload — traffic black-hole
HIGH mrs-svc-selector-mismatch Service/dep16-custom-clean Service selector matches no workload — traffic black-hole
MEDIUM wrs-spec-no-readiness Deployment/worker No readiness probe — traffic may hit unready pods
LOW wrs-spec-no-liveness Deployment/worker No liveness probe — hung pods won't auto-restart
LOW wrs-rs-nolimit Deployment/worker No CPU/memory limit — noisy-neighbour risk
LOW mrs-runasroot-unspecified Deployment/worker runAsNonRoot not set — container may run as root
LOW mrs-labels-missing-required Deployment/worker Missing required labels (app.kubernetes.io/name or /version)
LOW mrs-allow-priv-esc-unset Deployment/worker allowPrivilegeEscalation not set — defaults can allow privilege escalation
LOW mrs-readonly-rootfs Deployment/worker readOnlyRootFilesystem not enabled — writable container root filesystem
LOW mrs-seccomp-unset Deployment/worker seccompProfile not set — relies on cluster default
INFO mrs-diff-resource-added Ingress/rtzlbc/pr1 Resource added to manifest
INFO mrs-diff-resource-added Ingress/rtzlbc/pr2 Resource added to manifest
INFO mrs-diff-resource-added Ingress/rtzlbc/pr3 Resource added to manifest
INFO mrs-diff-resource-added Service/rtzlbc/pr3svc Resource added to manifest
INFO mrs-diff-resource-added Ingress/rtzlbc/pr4 Resource added to manifest
INFO mrs-diff-resource-added Ingress/rtzlbc/pr5 Resource added to manifest
INFO mrs-diff-resource-added Ingress/rtzlbc/pr6a Resource added to manifest
INFO mrs-diff-resource-added Ingress/rtzlbc/pr6b Resource added to manifest
INFO mrs-diff-resource-added TargetGroupBinding/rtzlbc/pr7 Resource added to manifest
INFO mrs-diff-resource-added IngressClassParams/pr8 Resource added to manifest
INFO mrs-diff-resource-added Ingress/rtzlbc/pr14 Resource added to manifest
INFO mrs-diff-resource-added Service/rtzlbc/dep15 Resource added to manifest
INFO mrs-diff-resource-added Service/rtzlbc/dep15-intree Resource added to manifest
INFO mrs-diff-resource-added Service/rtzlbc/dep16-custom Resource added to manifest
INFO mrs-diff-resource-added Service/rtzlbc/dep16-custom-clean Resource added to manifest

Add-on findings · KEDA (2)

Severity Rule Resource Title
HIGH wrs-addon-keda-target-has-other-scaledobject ScaledObject/shop/worker Workload is already scaled by another KEDA ScaledObject
HIGH wrs-addon-keda-target-has-other-scaledobject ScaledObject/shop/legacy Workload is already scaled by another KEDA ScaledObject

Add-on findings · AWS Load Balancer Controller (20)

Severity Rule Resource Title
HIGH wrs-addon-awslbc-group-name-invalid Ingress/rtzlbc/pr1 IngressGroup name the AWS Load Balancer Controller rejects
HIGH wrs-addon-awslbc-group-order-out-of-range Ingress/rtzlbc/pr2 IngressGroup order outside [-1000, 1000]
HIGH wrs-addon-awslbc-annotation-value-invalid Ingress/rtzlbc/pr3 AWS Load Balancer Controller annotation value the controller rejects
HIGH wrs-addon-awslbc-annotation-value-invalid Service/rtzlbc/pr3svc AWS Load Balancer Controller annotation value the controller rejects
HIGH wrs-addon-awslbc-action-config-invalid Ingress/rtzlbc/pr4 ALB actions / conditions / transforms annotation the controller rejects
HIGH wrs-addon-awslbc-action-annotation-missing Ingress/rtzlbc/pr5 Ingress path uses use-annotation but names no actions annotation
HIGH wrs-addon-awslbc-tgb-invalid TargetGroupBinding/rtzlbc/pr7 TargetGroupBinding the AWS Load Balancer Controller webhook refuses
HIGH wrs-addon-awslbc-icp-invalid IngressClassParams/pr8 IngressClassParams the AWS Load Balancer Controller webhook refuses
LOW wrs-addon-awslbc-deprecated-annotation Service/rtzlbc/dep15 Deprecated AWS Load Balancer Controller Service annotation
LOW wrs-addon-awslbc-deprecated-annotation Service/rtzlbc/dep16-custom Deprecated AWS Load Balancer Controller Service annotation
HIGH wrs-addon-awslbc-group-conflict Ingress/rtzlbc/pr6a Ingresses of one IngressGroup set a group-wide setting to different values
HIGH wrs-addon-awslbc-group-conflict Ingress/rtzlbc/ctl9 Ingresses of one IngressGroup set a group-wide setting to different values
LOW wrs-addon-awslbc-target-annotation-overridden Ingress/rtzlbc/pr14 Ingress annotation ignored because its IngressClassParams sets the same field
HIGH wrs-addon-awslbc-target-backend-service-missing TargetGroupBinding/rtzlbc/tgb12 Load balancer backend names a Service that does not exist
LOW wrs-addon-awslbc-target-readiness-gate-off TargetGroupBinding/rtzlbc/pr7 ip targets in a namespace without the pod readiness gate
HIGH wrs-addon-awslbc-scheme-change-replaces-lb Ingress/rtzlbc/ctl9 Load balancer scheme change deletes and re-creates the load balancer
MEDIUM wrs-addon-awslbc-target-group-replaced Ingress/rtzlbc/ctl10 Target type or backend protocol change replaces the target group
MEDIUM wrs-addon-awslbc-ingress-moves-group Ingress/rtzlbc/ctl11 Ingress changes IngressGroup and moves to another ALB
HIGH wrs-addon-awslbc-tgb-immutable-changed TargetGroupBinding/rtzlbc/tgb12 TargetGroupBinding change to a field the webhook holds immutable
HIGH wrs-addon-awslbc-loadbalancerclass-changed Service/rtzlbc/nlb13 Service loadBalancerClass change on a LoadBalancer Service
What changed and how to fix it
  • mrs-svc-selector-mismatch · Service/pr3svc — The Service's spec.selector does not match any workload's pod labels in this bundle. Traffic sent to this Service will be dropped. Check the selector labels and ensure the target Deployment/StatefulSet has matching pod template labels.
  • mrs-svc-selector-mismatch · Service/nlb13 — The Service's spec.selector does not match any workload's pod labels in this bundle. Traffic sent to this Service will be dropped. Check the selector labels and ensure the target Deployment/StatefulSet has matching pod template labels.
  • mrs-svc-selector-mismatch · Service/dep15 — The Service's spec.selector does not match any workload's pod labels in this bundle. Traffic sent to this Service will be dropped. Check the selector labels and ensure the target Deployment/StatefulSet has matching pod template labels.
  • mrs-svc-selector-mismatch · Service/dep15-intree — The Service's spec.selector does not match any workload's pod labels in this bundle. Traffic sent to this Service will be dropped. Check the selector labels and ensure the target Deployment/StatefulSet has matching pod template labels.
  • mrs-svc-selector-mismatch · Service/dep16-custom — The Service's spec.selector does not match any workload's pod labels in this bundle. Traffic sent to this Service will be dropped. Check the selector labels and ensure the target Deployment/StatefulSet has matching pod template labels.
  • mrs-svc-selector-mismatch · Service/dep16-custom-clean — The Service's spec.selector does not match any workload's pod labels in this bundle. Traffic sent to this Service will be dropped. Check the selector labels and ensure the target Deployment/StatefulSet has matching pod template labels.
  • wrs-spec-no-readiness · Deployment/worker — Add a readinessProbe to ensure traffic is only sent to pods that are ready to serve requests.
  • wrs-spec-no-liveness · Deployment/worker — Add a livenessProbe so Kubernetes can detect and restart stuck or deadlocked pods.
  • wrs-rs-nolimit · Deployment/worker — Set resource limits to prevent this workload from consuming excessive node resources.
  • mrs-runasroot-unspecified · Deployment/worker — No securityContext.runAsNonRoot:true is set; the container may run as root depending on the image. Set runAsNonRoot:true (and a non-zero runAsUser) to guarantee non-root.
  • mrs-labels-missing-required · Deployment/worker — Add app.kubernetes.io/name and app.kubernetes.io/version labels to the workload metadata. These labels are required for workload tracking and are used by observability, cost-allocation, and upgrade-intelligence tools.
  • mrs-allow-priv-esc-unset · Deployment/worker — Explicitly set securityContext.allowPrivilegeEscalation: false. When unset, the effective value can permit privilege escalation depending on other settings.
  • mrs-readonly-rootfs · Deployment/worker — Set securityContext.readOnlyRootFilesystem: true and mount writable paths as explicit emptyDir/volumes. A read-only root limits tampering and persistence after a compromise.
  • mrs-seccomp-unset · Deployment/worker — Set seccompProfile.type: RuntimeDefault at pod or container level so syscall filtering is explicit and not dependent on cluster defaults.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr1 — Ingress rtzlbc/pr1 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr2 — Ingress rtzlbc/pr2 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr3 — Ingress rtzlbc/pr3 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Service/rtzlbc/pr3svc — Service rtzlbc/pr3svc was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr4 — Ingress rtzlbc/pr4 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr5 — Ingress rtzlbc/pr5 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr6a — Ingress rtzlbc/pr6a was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr6b — Ingress rtzlbc/pr6b was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · TargetGroupBinding/rtzlbc/pr7 — TargetGroupBinding rtzlbc/pr7 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · IngressClassParams/pr8 — IngressClassParams pr8 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Ingress/rtzlbc/pr14 — Ingress rtzlbc/pr14 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Service/rtzlbc/dep15 — Service rtzlbc/dep15 was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Service/rtzlbc/dep15-intree — Service rtzlbc/dep15-intree was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Service/rtzlbc/dep16-custom — Service rtzlbc/dep16-custom was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • mrs-diff-resource-added · Service/rtzlbc/dep16-custom-clean — Service rtzlbc/dep16-custom-clean was added to the manifest.
    Fix: A new resource is present in the head manifest that was not in the base. Informational: verify the addition is intentional.
  • wrs-addon-keda-target-has-other-scaledobject · ScaledObject/shop/worker — ScaledObject shop/worker: Deployment worker is already scaled by ScaledObject legacy (also in this change).
    Fix: Merge the triggers into the existing ScaledObject, or remove it in an earlier change: deleting it in the same change does not help, because Argo CD, Helm and kubectl --prune apply the new ScaledObject before they delete the old one. A workload can have one ScaledObject: KEDA's admission webhook refuses the second, and without the webhook their HPAs fight over the replica count.
  • wrs-addon-keda-target-has-other-scaledobject · ScaledObject/shop/legacy — ScaledObject shop/legacy: Deployment worker is already scaled by ScaledObject worker (also in this change).
    Fix: Merge the triggers into the existing ScaledObject, or remove it in an earlier change: deleting it in the same change does not help, because Argo CD, Helm and kubectl --prune apply the new ScaledObject before they delete the old one. A workload can have one ScaledObject: KEDA's admission webhook refuses the second, and without the webhook their HPAs fight over the replica count.
  • wrs-addon-awslbc-group-name-invalid · Ingress/rtzlbc/pr1 — Ingress rtzlbc/pr1: alb.ingress.kubernetes.io/group.name "Public_Web" is not lower-case alphanumerics, '-' or '.', starting and ending with an alphanumeric — the controller does not load the IngressGroup
    Fix: Use a group name of lower-case letters, digits, - and . that starts and ends with a letter or digit, 63 characters at most (alb.ingress.kubernetes.io/group.name, or IngressClassParams spec.group.name). The controller refuses to load the group, so the Ingress gets no load balancer and an existing one stops receiving its changes.
  • wrs-addon-awslbc-group-order-out-of-range · Ingress/rtzlbc/pr2 — Ingress rtzlbc/pr2: alb.ingress.kubernetes.io/group.order "1001" is not an integer in [-1000, 1000] — the whole IngressGroup fails to load
    Fix: Set alb.ingress.kubernetes.io/group.order to an integer between -1000 and 1000. The controller fails to load the whole IngressGroup, so no member Ingress on that ALB is reconciled until it is fixed.
  • wrs-addon-awslbc-annotation-value-invalid · Ingress/rtzlbc/pr3 — Ingress rtzlbc/pr3: alb.ingress.kubernetes.io/backend-protocol-version="HTTP3" (one of GRPC | HTTP1 | HTTP2) — the model build fails for this Ingress's whole group
    Fix: Use a value the controller accepts (scheme internal | internet-facing; target-type instance | ip; backend-protocol HTTP | HTTPS; backend-protocol-version HTTP1 | HTTP2 | GRPC; listen-ports a JSON list of {"HTTP"|"HTTPS": 1-65535}; ALPN and IP lists in their documented form). An unknown value fails the model build: the load balancer is not created or stops updating, and on an Ingress the failure stops every Ingress of its group.
  • wrs-addon-awslbc-annotation-value-invalid · Service/rtzlbc/pr3svc — Service rtzlbc/pr3svc: service.beta.kubernetes.io/aws-load-balancer-scheme="public" (one of internal | internet-facing) — the controller fails to build this Service's load balancer
    Fix: Use a value the controller accepts (scheme internal | internet-facing; target-type instance | ip; backend-protocol HTTP | HTTPS; backend-protocol-version HTTP1 | HTTP2 | GRPC; listen-ports a JSON list of {"HTTP"|"HTTPS": 1-65535}; ALPN and IP lists in their documented form). An unknown value fails the model build: the load balancer is not created or stops updating, and on an Ingress the failure stops every Ingress of its group.
  • wrs-addon-awslbc-action-config-invalid · Ingress/rtzlbc/pr4 — Ingress rtzlbc/pr4: alb.ingress.kubernetes.io/actions.ssl-redirect: redirect without redirectConfig.statusCode
    Fix: Make the annotation valid JSON of the documented shape: an action needs a known type (forward, redirect, fixed-response), a fixed-response needs a statusCode, and a forward to several target groups needs a weight on each. The validating webhook refuses a bad conditions annotation; a bad action fails the Ingress group's model build.
  • wrs-addon-awslbc-action-annotation-missing · Ingress/rtzlbc/pr5 — Ingress rtzlbc/pr5: backend ssl-redirect use port.name use-annotation with no alb.ingress.kubernetes.io/actions. — the path serves a fixed 503 (default --tolerate-non-existent-backend-action=true)
    Fix: Add alb.ingress.kubernetes.io/actions. for the path backend whose port.name is use-annotation. Without it the controller serves that path a fixed 503 (the default --tolerate-non-existent-backend-action=true), or fails the whole IngressGroup when the flag is false.
  • wrs-addon-awslbc-tgb-invalid · TargetGroupBinding/rtzlbc/pr7 — TargetGroupBinding rtzlbc/pr7: neither spec.targetGroupARN nor spec.targetGroupName is set
    Fix: Set spec.targetGroupARN or spec.targetGroupName; drop spec.nodeSelector when targetType is ip; set iamRoleArnToAssume when registerTargetsWithPodAvailabilityZone is true, and use targetType ip with an assumed role; write vpcID as vpc- plus 8, 17 or 32 lower-case hex characters. The validating webhook refuses the apply.
  • wrs-addon-awslbc-icp-invalid · IngressClassParams/pr8 — IngressClassParams pr8: spec.subnets sets both ids and tags
    Fix: Give spec.subnets either ids or tags (not both, ids without duplicates, at least one tag key), and write each spec.inboundCIDRs entry as a CIDR with no host bits set (10.0.0.0/8, not 10.0.0.1/8 or a bare address). The validating webhook refuses the apply.
  • wrs-addon-awslbc-deprecated-annotation · Service/rtzlbc/dep15 — Service rtzlbc/dep15: service.beta.kubernetes.io/aws-load-balancer-cross-zone-load-balancing-enabled (use aws-load-balancer-attributes load_balancing.cross_zone.enabled); service.beta.kubernetes.io/aws-load-balancer-internal (use aws-load-balancer-scheme: internal for "true", internet-facing for "false") — deprecated, still supported at v3.6.2
    Fix: Move the access-log and cross-zone settings into service.beta.kubernetes.io/aws-load-balancer-attributes (access_logs.s3.enabled, access_logs.s3.bucket, access_logs.s3.prefix, load_balancing.cross_zone.enabled), and replace aws-load-balancer-internal with aws-load-balancer-scheme at the SAME value ("true" is internal, "false" internet-facing; a different scheme re-creates the NLB). All of them still work at v3.6.2: the access-log and cross-zone keys win a tie against aws-load-balancer-attributes (access-log-enabled only when true), while aws-load-balancer-scheme already wins a tie against aws-load-balancer-internal. Clean-up, not a blocker.
  • wrs-addon-awslbc-deprecated-annotation · Service/rtzlbc/dep16-custom — Service rtzlbc/dep16-custom: service.beta.kubernetes.io/aws-load-balancer-cross-zone-load-balancing-enabled (use aws-load-balancer-attributes load_balancing.cross_zone.enabled); service.beta.kubernetes.io/aws-load-balancer-internal (use aws-load-balancer-scheme: internal for "true", internet-facing for "false") — deprecated, still supported at v3.6.2 (loadBalancerClass rtz.example/custom: applies when the controller runs with --load-balancer-class=rtz.example/custom)
    Fix: Move the access-log and cross-zone settings into service.beta.kubernetes.io/aws-load-balancer-attributes (access_logs.s3.enabled, access_logs.s3.bucket, access_logs.s3.prefix, load_balancing.cross_zone.enabled), and replace aws-load-balancer-internal with aws-load-balancer-scheme at the SAME value ("true" is internal, "false" internet-facing; a different scheme re-creates the NLB). All of them still work at v3.6.2: the access-log and cross-zone keys win a tie against aws-load-balancer-attributes (access-log-enabled only when true), while aws-load-balancer-scheme already wins a tie against aws-load-balancer-internal. Clean-up, not a blocker.
  • wrs-addon-awslbc-group-conflict · Ingress/rtzlbc/pr6a — Ingress rtzlbc/pr6a: IngressGroup "rtzpr-g" (rtzlbc/pr6a, rtzlbc/pr6b): conflicting alb.ingress.kubernetes.io/scheme {rtzlbc/pr6a=internal, rtzlbc/pr6b=internet-facing} — the controller fails the whole group's model build, so no Ingress of the group (and no app behind its ALB) receives changes
    Fix: Make every Ingress of the group agree on the setting, or set it on one member only. The controller merges a group into one ALB and fails the whole model build on a conflict (conflicting scheme, subnets, security groups, WAF, tags, attributes, listen-port protocol...), so every app behind that ALB stops receiving changes until it is fixed.
  • wrs-addon-awslbc-group-conflict · Ingress/rtzlbc/ctl9 — Ingress rtzlbc/ctl9: IngressGroup "rtzlbc-ok" (rtzlbc/ctl9, rtzlbc/ctl10): conflicting alb.ingress.kubernetes.io/scheme {rtzlbc/ctl9=internet-facing, rtzlbc/ctl10=internal} — the controller fails the whole group's model build, so no Ingress of the group (and no app behind its ALB) receives changes
    Fix: Make every Ingress of the group agree on the setting, or set it on one member only. The controller merges a group into one ALB and fails the whole model build on a conflict (conflicting scheme, subnets, security groups, WAF, tags, attributes, listen-port protocol...), so every app behind that ALB stops receiving changes until it is fixed.
  • wrs-addon-awslbc-target-annotation-overridden · Ingress/rtzlbc/pr14 — Ingress rtzlbc/pr14: IngressClass alb-params's IngressClassParams override alb.ingress.kubernetes.io/scheme (params scheme) — those annotations have no effect
    Fix: Remove the annotation, or change the value on the IngressClassParams instead: a field the class's params set (group, scheme, ip-address-type, subnets, inbound CIDRs, SSL policy, target type...) overrides the Ingress annotation, so the annotation has no effect.
  • wrs-addon-awslbc-target-backend-service-missing · TargetGroupBinding/rtzlbc/tgb12 — TargetGroupBinding rtzlbc/tgb12: Service(s) nosuch exist neither in this change nor in namespace rtzlbc — the binding reports BackendNotFound and registers no target, so its target group serves nothing
    Fix: Create the Service in the same change, or point the backend at a Service that exists in that namespace. With the default --tolerate-non-existent-backend-service=true the ALB answers that path with a 503; a TargetGroupBinding reports BackendNotFound and registers no target, so the target group serves nothing.
  • wrs-addon-awslbc-target-readiness-gate-off · TargetGroupBinding/rtzlbc/pr7 — TargetGroupBinding rtzlbc/pr7: TargetGroupBinding pr7, Ingress ctl9, Ingress ctl10, Ingress ctl11 register pod IPs as targets, and namespace rtzlbc lacks the label elbv2.k8s.aws/pod-readiness-gate-inject=enabled — a rolling update can stop an old pod before the load balancer sends traffic to its replacement
    Fix: Label the namespace elbv2.k8s.aws/pod-readiness-gate-inject=enabled. With ip targets the load balancer sends traffic to pod IPs; without the readiness gate a rolling update can count a new pod ready and stop an old one before the target group has registered the new one, which drops requests during deploys.
  • wrs-addon-awslbc-scheme-change-replaces-lb · Ingress/rtzlbc/ctl9 — Ingress rtzlbc/ctl9: alb.ingress.kubernetes.io/scheme internal → internet-facing — the controller deletes the ALB and creates a new one with a new DNS name; IngressGroup "rtzlbc-ok" shares it, so every member moves (rtzlbc/ctl10 in this change)
    Fix: Plan the change as a migration: the controller cannot change a load balancer's scheme in place, so it deletes it and creates a new one with a new DNS name. Traffic fails until clients and DNS records follow the new name; on an Ingress the whole IngressGroup moves. Create the new load balancer beside the old one (another group or Service) and cut over instead.
  • wrs-addon-awslbc-target-group-replaced · Ingress/rtzlbc/ctl10 — Ingress rtzlbc/ctl10: alb.ingress.kubernetes.io/backend-protocol HTTP → HTTPS — every target group of this Ingress is replaced and its targets re-register
    Fix: Expect a short window of errors: a target type, backend protocol or protocol version change makes the controller create a new target group, register the targets and delete the old one. Deploy it outside peak, with the readiness gate enabled on the namespace so pods are registered before they are counted ready.
  • wrs-addon-awslbc-ingress-moves-group · Ingress/rtzlbc/ctl11 — Ingress rtzlbc/ctl11: moves from IngressGroup "rtzlbc-ok" to IngressGroup "rtzlbc-ok2" — its hosts are served by another load balancer with another DNS name
    Fix: Move the DNS records for this Ingress's hosts to the new ALB's address: changing alb.ingress.kubernetes.io/group.name (adding, removing or renaming it) moves the rules to another load balancer with another DNS name, and the old one keeps serving only until the controller removes the rules.
  • wrs-addon-awslbc-tgb-immutable-changed · TargetGroupBinding/rtzlbc/tgb12 — TargetGroupBinding rtzlbc/tgb12: spec.targetType changed — the webhook refuses the update
    Fix: Delete and re-create the TargetGroupBinding instead of editing it: the validating webhook refuses any change to targetGroupARN, targetType, ipAddressType or vpcID, so the apply fails and the old binding keeps running.
  • wrs-addon-awslbc-loadbalancerclass-changed · Service/rtzlbc/nlb13 — Service rtzlbc/nlb13: spec.loadBalancerClass service.k8s.aws/nlb → example.com/lb on a Service that stays type LoadBalancer — the API server refuses the update
    Fix: Re-create the Service (or create a new one) to move it to another load balancer controller: the API server refuses a change to spec.loadBalancerClass while the Service stays type LoadBalancer, so the apply fails.

Cluster-aware checks: on — 50 runtimez lookup(s) answered.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant