install/kubernetes: use HTTP for agent {liveness,readiness}Probe

Use the newly introduced /healthz HTTP path exposed on localhost to check for liveness/readiness of the Cilium agent. In the long term this would also allow to exclude the cilium CLI tool from the runtime image and move it into a separate image used with an ephemeral container for debugging. Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
cilium · May 12, 2020 · d613dea · d613dea
1 parent 7e5cc77
commit d613dea
Show file tree

Hide file tree

Showing 3 changed files with 22 additions and 22 deletions.
diff --git a/daemon/cmd/agenthealth.go b/daemon/cmd/agenthealth.go
@@ -52,7 +52,7 @@ func (d *Daemon) startAgentHealthHTTPService(addr string) {
 	lc := net.ListenConfig{Control: setsockoptReuseAddrAndPort}
 	ln, err := lc.Listen(context.Background(), "tcp", addr)
 	if err != nil {
-		log.Fatal(err)
+		log.WithError(err).Fatalf("Unable to listen on %s for healthz status API server", addr)
 	}
 
 	mux := http.NewServeMux()
@@ -82,7 +82,7 @@ func (d *Daemon) startAgentHealthHTTPService(addr string) {
 		if err == http.ErrServerClosed {
 			log.Info("healthz status API server shutdown")
 		} else if err != nil {
-			log.WithError(err).Fatal("Unable to start status healthz status API server")
+			log.WithError(err).Fatal("Unable to start healthz status API server")
 		}
 	}()
 	log.Infof("Started healthz status API server on address %s", addr)

diff --git a/install/kubernetes/cilium/charts/agent/templates/daemonset.yaml b/install/kubernetes/cilium/charts/agent/templates/daemonset.yaml
@@ -55,11 +55,11 @@ spec:
         command:
         - cilium-agent
         livenessProbe:
-          exec:
-            command:
-            - cilium
-            - status
-            - --brief
+          httpGet:
+            host: '127.0.0.1'
+            path: /healthz
+            port: {{ .Values.global.agent.healthPort }}
+            scheme: HTTP
           failureThreshold: 10
           # The initial delay for the liveness probe is intentionally large to
           # avoid an endless kill & restart cycle if in the event that the initial
@@ -69,11 +69,11 @@ spec:
           successThreshold: 1
           timeoutSeconds: 5
         readinessProbe:
-          exec:
-            command:
-            - cilium
-            - status
-            - --brief
+          httpGet:
+            host: '127.0.0.1'
+            path: /healthz
+            port: {{ .Values.global.agent.healthPort }}
+            scheme: HTTP
           failureThreshold: 3
           initialDelaySeconds: 5
           periodSeconds: 30

diff --git a/install/kubernetes/quick-install.yaml b/install/kubernetes/quick-install.yaml
@@ -364,11 +364,11 @@ spec:
         command:
         - cilium-agent
         livenessProbe:
-          exec:
-            command:
-            - cilium
-            - status
-            - --brief
+          httpGet:
+            host: '127.0.0.1'
+            path: /healthz
+            port: 9876
+            scheme: HTTP
           failureThreshold: 10
           # The initial delay for the liveness probe is intentionally large to
           # avoid an endless kill & restart cycle if in the event that the initial
@@ -378,11 +378,11 @@ spec:
           successThreshold: 1
           timeoutSeconds: 5
         readinessProbe:
-          exec:
-            command:
-            - cilium
-            - status
-            - --brief
+          httpGet:
+            host: '127.0.0.1'
+            path: /healthz
+            port: 9876
+            scheme: HTTP
           failureThreshold: 3
           initialDelaySeconds: 5
           periodSeconds: 30