test: add e2e readme

Makefile

@@ -89,11 +89,6 @@ test: ginkgo ## Run tests.
 	go tool cover -func=./server/http/coverage.out 
 	go tool cover -func=./common/gpuallocator/coverage.out
 
-.PHONY: e2e-test
-e2e-test: ## Run e2e tests.
-	go test -timeout 20m  -v ./e2e -shared-volumes-dir /data/nfs/ding/cartpole --kubeconfig ~/.kube/config
-	go tool cover -func=./e2e/coverage.out 
-
 ##@ Build
 
 build: generate  ## Build di-operator binary.

config/samples/dijob-cartpole.yaml

@@ -169,11 +169,11 @@ spec:
             value: "1"
           resources:
             requests:
-              cpu: 8
-              memory: "10Gi"
+              cpu: 16
+              memory: "5Gi"
             limits:
-              cpu: 8
-              memory: "10Gi"
+              cpu: 16
+              memory: "5Gi"
           command: ["/bin/bash", "-c",]
           args:
           - |
@@ -196,11 +196,11 @@ spec:
             value: "1"
           resources:
             requests:
-              cpu: 8
-              memory: "20Gi"
+              cpu: 16
+              memory: "5Gi"
             limits:
-              cpu: 8
-              memory: "20Gi"
+              cpu: 16
+              memory: "5Gi"
           command: ["/bin/bash", "-c",]
           args:
           - |

config/samples/dijob-qbert.yaml

@@ -177,10 +177,10 @@ spec:
             value: "1"
           resources:
             requests:
-              cpu: 8
+              cpu: 16
               memory: "10Gi"
             limits:
-              cpu: 8
+              cpu: 16
               memory: "10Gi"
           command: ["/bin/bash", "-c",]
           args:
@@ -207,13 +207,11 @@ spec:
             value: "1"
           resources:
             requests:
-              cpu: 3
-              nvidia.com/gpu: 2
-              memory: "30Gi"
+              cpu: 16
+              memory: "20Gi"
             limits:
-              cpu: 3
-              nvidia.com/gpu: 2
-              memory: "30Gi"
+              cpu: 16
+              memory: "20Gi"
           command: ["/bin/bash", "-c",]
           args:
           - |

e2e/README.md

+# E2E Tests
+Through the e2e test, we can test the robustness of DI-engine, ensuring that DIJobs can tolerate common exceptions.
+
+## Run
+```bash
+go test -timeout 20m -cover -v ./e2e --ginkgo.v --shared-volumes-dir /data/nfs/ding --kubeconfig ~/.kube/config
+```
+- `shared-volumes-dir` represents the shared volumes directory for DI-engine modules (coordinator, collector, etc.) to exchange data and models. Different jobs's shared volumes are placed under this directory. Default `/data/nfs/ding`.
+- `kubeconfig` represents path to kubeconfig file to access kubernetes cluster. Default `$HOME/.kube/config`.
+- `timeout` can be set according to how long the test will last.

e2e/config/dijob.yaml

@@ -26,11 +26,11 @@ spec:
             value: "1"
           resources:
             requests:
-              cpu: 3
-              memory: "10Gi"
+              cpu: 2
+              memory: "5Gi"
             limits:
-              cpu: 3
-              memory: "10Gi"
+              cpu: 2
+              memory: "5Gi"
           command: ["/bin/bash", "-c",]
           args:
           - |
@@ -169,10 +169,10 @@ spec:
             value: "1"
           resources:
             requests:
-              cpu: 16
+              cpu: 8
               memory: "5Gi"
             limits:
-              cpu: 16
+              cpu: 8
               memory: "5Gi"
           command: ["/bin/bash", "-c",]
           args:
@@ -196,10 +196,10 @@ spec:
             value: "1"
           resources:
             requests:
-              cpu: 16
+              cpu: 5
               memory: "5Gi"
             limits:
-              cpu: 16
+              cpu: 5
               memory: "5Gi"
           command: ["/bin/bash", "-c",]
           args: