Enable monitoring and high-availability
This commit is contained in:
1 parent
c4a808a565
commit
6f4bb339af
4 files changed
+52
-2
No files matched your search
@@ -8,7 +8,10 @@ cilium-cli upgrade --version v1.20.0 --dry-run-helm-values
|
||||
```
|
||||
|
||||
They preserve the Kubernetes API endpoint, VXLAN tunnel routing,
|
||||
kube-proxy replacement, one operator replica, and Hubble Relay/UI.
|
||||
kube-proxy replacement, two operator replicas, and Hubble Relay/UI. The
|
||||
operator replicas are restricted to Odroid workers, use required pod
|
||||
anti-affinity so they run on different nodes, and have a disruption budget
|
||||
that keeps at least one available during voluntary disruptions.
|
||||
|
||||
## Hubble TLS
|
||||
|
||||
|
||||
@@ -0,0 +1,35 @@
|
||||
apiVersion: monitoring.coreos.com/v1
|
||||
kind: PrometheusRule
|
||||
metadata:
|
||||
name: cilium-availability
|
||||
namespace: prometheus
|
||||
spec:
|
||||
groups:
|
||||
- name: cilium.availability
|
||||
rules:
|
||||
- alert: CiliumAgentDown
|
||||
expr: |
|
||||
kube_daemonset_status_desired_number_scheduled{namespace="kube-system", daemonset="cilium"}
|
||||
-
|
||||
kube_daemonset_status_number_ready{namespace="kube-system", daemonset="cilium"}
|
||||
> 0
|
||||
for: 5m
|
||||
labels:
|
||||
namespace: kube-system
|
||||
severity: critical
|
||||
annotations:
|
||||
summary: One or more Cilium agents are unavailable
|
||||
description: >-
|
||||
Only {{ $value }} fewer Cilium agents are ready than desired.
|
||||
Affected nodes may lose pod networking or service connectivity.
|
||||
- alert: CiliumOperatorDown
|
||||
expr: |
|
||||
kube_deployment_status_replicas_available{namespace="kube-system", deployment="cilium-operator"} < 1
|
||||
for: 5m
|
||||
labels:
|
||||
namespace: kube-system
|
||||
severity: critical
|
||||
annotations:
|
||||
summary: The Cilium operator is unavailable
|
||||
description: >-
|
||||
No Cilium operator replica has been available for at least five minutes.
|
||||
@@ -73,3 +73,9 @@ kubectl rollout status deployment/hubble-relay \
|
||||
--namespace "$NAMESPACE" --timeout=5m
|
||||
|
||||
cilium-cli status --wait
|
||||
|
||||
# Install availability alerts when Prometheus Operator is present. This keeps
|
||||
# a fresh Cilium-only bootstrap independent from the monitoring install order.
|
||||
if kubectl get customresourcedefinition prometheusrules.monitoring.coreos.com >/dev/null 2>&1; then
|
||||
kubectl apply --filename "$SCRIPT_DIR/cilium-alerts.yaml"
|
||||
fi
|
||||
+7
-1
@@ -22,7 +22,13 @@ k8sServicePort: 6443
|
||||
kubeProxyReplacement: true
|
||||
|
||||
operator:
|
||||
replicas: 1
|
||||
replicas: 2
|
||||
nodeSelector:
|
||||
kubernetes.io/os: linux
|
||||
odroid-h3: enabled
|
||||
podDisruptionBudget:
|
||||
enabled: true
|
||||
maxUnavailable: 1
|
||||
|
||||
routingMode: tunnel
|
||||
tunnelProtocol: vxlan
|
||||
Reference in new issue
Block a user