fix: remove startup probe

In some cases the Redis pod isn't reachable by Connaisseur, which results in Connaisseur never starting up even though it could perform its task even without the Redis pod. Thus the startup probe for Connaisseur, trying to reach the Redis pod was removed.

charts/connaisseur/templates/deployment.yaml

@@ -42,13 +42,6 @@ spec:
               path: /ready
               port: https
               scheme: HTTPS
-          startupProbe:
-            httpGet:
-              path: /start
-              port: https
-              scheme: HTTPS
-            periodSeconds: 5
-            failureThreshold: 30
           securityContext:
             {{- toYaml .Values.kubernetes.deployment.securityContext | nindent 12 }}
           resources:

cmd/connaisseur/main.go

@@ -36,7 +36,6 @@ func startServer(config *config.Config, alerting *alerting.Config) {
 	mux.HandleFunc("/health", handler.HandleHealth)
 	mux.HandleFunc("/ready", handler.HandleHealth)
 	mux.HandleFunc("/mutate", handler.HandleMutate)
-	mux.HandleFunc("/start", handler.HandleStart)
 	mux.HandleFunc("/metrics", promhttp.Handler().ServeHTTP)
 
 	cache := caching.NewCacher()

internal/handler/validation/skip.go

@@ -8,6 +8,7 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"strings"
 
 	"github.com/sirupsen/logrus"
 )
@@ -106,6 +107,9 @@ func getCachedDigest(
 
 	val, err := cache.Get(ctx, img.OriginalString())
 	if err != nil {
+		if strings.Contains(err.Error(), "dial tcp") {
+			logrus.Warnf("error connecting to cache: %s", err)
+		}
 		return "", nil, fmt.Errorf("cache miss for image %s: %s", img.OriginalString(), err)
 	}