Add e2e tests (#1)

README.md

@@ -1,6 +1,6 @@
 ## About
 
-Memcache server discovery based on Kubernetes endpoints for [Go memcache client](https://github.com/bradfitz/gomemcache).
+Memcache server discovery based on Kubernetes endpoints for [Go memcache client](https://github.com/bradfitz/gomemcache)
 
 How it works:
 * Watches endpoints containing memcache pod ips using [tiny k8s client](https://github.com/castai/k8s-client-go)
@@ -10,8 +10,8 @@ How it works:
 
 In dynamic environments like Kubernetes both clients (pods calling memcache) and memcache pods could change.
 There are few issues:
-1. Calling memcache using clusterIP service is quite useless (unless you have one replica) as kernel will perform round robbin across service endpoints.
-2. You can make clusterIP service better by adding sessionAffinity. This will ensure that client pod always goes to the same memcache instance. But what happens if pods are redeployed?
+1. Calling memcache using clusterIP service is quite useless (unless you have one replica) as kernel will perform round-robin across service endpoints.
+2. You can enable sessionAffinity. This will ensure that client pod always goes to the same memcache instance. But what happens if pods are redeployed?
 3. Using StatefulSet with headless services and passing each endpoint is kind of static too.
 
 ## Installing
@@ -62,7 +62,7 @@ metadata:
   name: endpoints-watcher
 rules:
   - apiGroups: [ "" ]
-    resources: [ "pods" ]
+    resources: [ "endpoints" ]
     verbs: [ "get", "watch" ]
 ---
 kind: RoleBinding

e2e/e2e.yaml

@@ -8,15 +8,17 @@ kind: Role
 apiVersion: rbac.authorization.k8s.io/v1
 metadata:
   name: e2e
+  namespace: e2e
 rules:
   - apiGroups: [ "" ]
-    resources: [ "pods" ]
+    resources: [ "endpoints" ]
     verbs: [ "get", "watch" ]
 ---
 kind: RoleBinding
 apiVersion: rbac.authorization.k8s.io/v1
 metadata:
   name: e2e
+  namespace: e2e
 roleRef:
   apiGroup: rbac.authorization.k8s.io
   kind: Role
@@ -40,6 +42,6 @@ spec:
       serviceAccount: e2e
       containers:
         - name: e2e
-          image: replace-img
+          image: ghcr.io/castai/k8s-memcache-selector/e2e:08c1a77b1161295ac24a83d6f0c558560374b7da
       restartPolicy: Never
   backoffLimit: 0

e2e/memcache.yaml

@@ -32,15 +32,6 @@ spec:
       labels:
         app.kubernetes.io/name: memcache
     spec:
-      affinity:
-        podAntiAffinity:
-          preferredDuringSchedulingIgnoredDuringExecution:
-              - podAffinityTerm:
-                labelSelector:
-                  matchLabels:
-                    app.kubernetes.io/name: memcache
-                topologyKey: kubernetes.io/hostname
-                weight: 1
       securityContext:
         fsGroup: 1001
       containers:

e2e/run.sh

@@ -29,3 +29,7 @@ kubectl wait pods -l app.kubernetes.io/name=memcache --for condition=Ready --tim
 
 kubectl apply -f e2e.yaml --dry-run=client -oyaml | sed "s/replace-img/$(echo "$img" | sed 's/\//\\\//g')/" | kubectl apply -f - -n e2e
 kubectl wait job/e2e --for=condition=complete --timeout=15s  -n e2e
+
+echo "Test completed"
+echo "Logs:"
+kubectl logs -l app=e2e -n e2e