I'm trying to run a service exposed via port 80 and 443. The SSL termination happens on the pod. I specified only port 80 for liveness probe but for some reasons kubernates is probing https (443) as well. Why is that and how can I stop it probing 443?
Kubernates config
apiVersion: v1
kind: Secret
metadata:
name: myregistrykey
namespace: default
data:
.dockerconfigjson: xxx==
type: kubernetes.io/dockerconfigjson
---
apiVersion: apps/v1beta1
kind: Deployment
metadata:
name: example-com
spec:
replicas: 0
strategy:
rollingUpdate:
maxSurge: 1
maxUnavailable: 50%
minReadySeconds: 30
template:
metadata:
labels:
app: example-com
spec:
imagePullSecrets:
- name: myregistrykey
containers:
- name: example-com
image: DOCKER_HOST/DOCKER_IMAGE_VERSION
imagePullPolicy: Always
ports:
- containerPort: 80
protocol: TCP
name: http
- containerPort: 443
protocol: TCP
name: https
livenessProbe:
httpGet:
scheme: "HTTP"
path: "/_ah/health"
port: 80
httpHeaders:
- name: Host
value: example.com
initialDelaySeconds: 35
periodSeconds: 35
readinessProbe:
httpGet:
scheme: "HTTP"
path: "/_ah/health"
port: 80
httpHeaders:
- name: Host
value: example.com
initialDelaySeconds: 35
periodSeconds: 35
resources:
requests:
cpu: 250m
limits:
cpu: 500m
---
apiVersion: v1
kind: Service
metadata:
name: example-com
spec:
type: LoadBalancer
ports:
- port: 80
protocol: TCP
targetPort: 80
nodePort: 0
name: http
- port: 443
protocol: TCP
targetPort: 443
nodePort: 0
name: https
selector:
app: example-com
The error/logs on pods clearly indicate that kubernates is trying to access the service via https.
kubectl describe pod example-com-86876875c7-b75hr
Name: example-com-86876875c7-b75hr
Namespace: default
Priority: 0
PriorityClassName: <none>
Node: aks-agentpool-37281605-0/10.240.0.4
Start Time: Sat, 17 Nov 2018 19:58:30 +0200
Labels: app=example-com
pod-template-hash=4243243173
Annotations: <none>
Status: Running
IP: 10.244.0.65
Controlled By: ReplicaSet/example-com-86876875c7
Containers:
example-com:
Container ID: docker://c5eeb03558adda435725a0df3cc2d15943966c3df53e9462e964108969c8317a
Image: example-com.azurecr.io/example-com:2018-11-17_19-58-05
Image ID: docker-pullable://example-com.azurecr.io/example-com@sha256:5d425187b8663ecfc5d6cc78f6c5dd29f1559d3687ba9d4c0421fd0ad109743e
Ports: 80/TCP, 443/TCP
Host Ports: 0/TCP, 0/TCP
State: Running
Started: Sat, 17 Nov 2018 20:07:59 +0200
Last State: Terminated
Reason: Error
Exit Code: 2
Started: Sat, 17 Nov 2018 20:05:39 +0200
Finished: Sat, 17 Nov 2018 20:07:55 +0200
Ready: False
Restart Count: 3
Limits:
cpu: 500m
Requests:
cpu: 250m
Liveness: http-get http://:80/_ah/health delay=35s timeout=1s period=35s #success=1 #failure=3
Readiness: http-get http://:80/_ah/health delay=35s timeout=1s period=35s #success=1 #failure=3
Environment:
NABU: nabu
KUBERNETES_PORT_443_TCP_ADDR: agile-kube-b3e5753f.hcp.westeurope.azmk8s.io
KUBERNETES_PORT: tcp://agile-kube-b3e5753f.hcp.westeurope.azmk8s.io:443
KUBERNETES_PORT_443_TCP: tcp://agile-kube-b3e5753f.hcp.westeurope.azmk8s.io:443
KUBERNETES_SERVICE_HOST: agile-kube-b3e5753f.hcp.westeurope.azmk8s.io
Mounts:
/var/run/secrets/kubernetes.io/serviceaccount from default-token-rcr7c (ro)
Conditions:
Type Status
Initialized True
Ready False
ContainersReady False
PodScheduled True
Volumes:
default-token-rcr7c:
Type: Secret (a volume populated by a Secret)
SecretName: default-token-rcr7c
Optional: false
QoS Class: Burstable
Node-Selectors: <none>
Tolerations: node.kubernetes.io/not-ready:NoExecute for 300s
node.kubernetes.io/unreachable:NoExecute for300s
Events:
Type Reason Age From Message
---- ------ ---- ---- -------
Normal Scheduled 10m default-scheduler Successfully assigned default/example-com-86876875c7-b75hr to aks-agentpool-37281605-0
Warning Unhealthy 3m46s (x6 over 7m16s) kubelet, aks-agentpool-37281605-0 Liveness probe failed: Get https://example.com/_ah/health: net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
Normal Pulling 3m45s (x3 over 10m) kubelet, aks-agentpool-37281605-0 pulling image "example-com.azurecr.io/example-com:2018-11-17_19-58-05"
Normal Killing 3m45s (x2 over 6m5s) kubelet, aks-agentpool-37281605-0 Killing container with id docker://example-com:Container failed liveness probe.. Container will be killed andrecreated.
Normal Pulled 3m44s (x3 over 10m) kubelet, aks-agentpool-37281605-0 Successfully pulled image "example-com.azurecr.io/example-com:2018-11-17_19-58-05"
Normal Created 3m42s (x3 over 10m) kubelet, aks-agentpool-37281605-0 Created container
Normal Started 3m42s (x3 over 10m) kubelet, aks-agentpool-37281605-0 Started container
Warning Unhealthy 39s (x9 over 7m4s) kubelet, aks-agentpool-37281605-0 Readiness probe failed: Get https://example.com/_ah/health: net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
As per your comments, you are doing an HTTP to HTTPS redirect in the pod and basically, the probe cannot connect to it. If you still want to serve a probe on port 80 you should consider using TCP probes. For example:
apiVersion: apps/v1beta1
kind: Deployment
metadata:
name: example-com
spec:
...
minReadySeconds: 30
template:
metadata:
labels:
app: example-com
spec:
imagePullSecrets:
- name: myregistrykey
containers:
- name: example-com
...
livenessProbe:
httpGet:
scheme: "HTTP"
path: "/_ah/health"
port: 80
httpHeaders:
- name: Host
value: example.com
initialDelaySeconds: 35
periodSeconds: 35
readinessProbe:
tcpSocket:
port: 80
initialDelaySeconds: 35
periodSeconds: 35
...
Or you can ignore some redirects in your application depending on the URL, just like mentioned in @night-gold's answer.
The problem doesn't come from Kubernetes but from your web server. Kubernetes is doing exactly what you are asking, probing the http url but your server is redirecting it to https, that is causing the error.
If you are using apache, you should look here Apache https block redirect or there if you use nginx nginx https block redirect