Integrate bb-common with fluentbit
General MR
Summary
Integrate bb-common with fluentbit
Umbrella Branch
integrate-fluentbit-bb-common
Relevant logs/screenshots
Before our changes
k get networkpolicies.networking.k8s.io -o wide
NAME POD-SELECTOR AGE
allow-api-egress-fluentbit app.kubernetes.io/name=fluent-bit 5m47s
allow-dns-lookups app.kubernetes.io/name=fluent-bit 5m47s
allow-fluentbit-sidecar-scraping app.kubernetes.io/name=fluent-bit 5m47s
allow-helm-test-egress-fluentbit helm-test=enabled 5m47s
allow-intra-namespace app.kubernetes.io/name=fluent-bit 5m47s
allow-loki-egress-fluentbit app.kubernetes.io/name=fluent-bit 5m47s
allow-prometheus app.kubernetes.io/name=fluent-bit 5m47s
allow-tempo-egress-fluentbit <none> 5m47s
allow-to-istiod app.kubernetes.io/name=fluent-bit 5m47s
After our changes
NAME POD-SELECTOR AGE
allow-egress-from-fluent-bit-to-kubeapi app.kubernetes.io/name=fluent-bit 39s
allow-egress-from-fluent-bit-to-ns-logging-pod-elasticsearch-tcp-port-9200 app.kubernetes.io/name=fluent-bit 39s
allow-egress-from-fluent-bit-to-ns-logging-pod-logging-loki-tcp-port-3100 app.kubernetes.io/name=fluent-bit 39s
allow-egress-from-fluent-bit-to-ns-tempo-pod-tempo-tcp-port-9411 app.kubernetes.io/name=fluent-bit 39s
allow-ingress-to-fluent-bit-tcp-port-2020-from-ns-monitoring-pod-prometheus app.kubernetes.io/name=fluent-bit 39s
default-egress-allow-all-in-ns <none> 39s
default-egress-allow-istiod <none> 39s
default-egress-allow-kube-dns <none> 39s
default-egress-deny-all <none> 39s
default-ingress-allow-all-in-ns <none> 39s
default-ingress-allow-prometheus-to-istio-sidecar <none> 39s
default-ingress-deny-all <none> 39s
Linked Issue
Upgrade Notices
Fluent Bit is now leveraging our bb-common integration for network policies and Istio-related resources. Please refer to this blog post for additional information on the integration. During this process, network policy definitions have been normalized and some keys are now in kebab-case (e.g. external-elastic, external-loki, external-fluentd), so any existing overrides using older names must be updated. Loki egress is now explicitly managed through the bb-common NetworkPolicy DSL (and enabled by default for in- cluster Loki), and AuthorizationPolicy generation (when Istio hardened is enabled) is tied to service account-qualified rules; if you rely on custom ingress/egress, ensure those rules include the correct namespace/service account identity to avoid unintended denies.



