Demo script.

Signed-off-by: Klaus Ma <mada3@huawei.com>
volcano-sh · Jun 19, 2019 · 04c2046 · 04c2046
1 parent e42994d
commit 04c2046
Show file tree

Hide file tree

Showing 21 changed files with 612 additions and 111 deletions.
diff --git a/docs/samples/kubecon-2019-china/deploys/controllers.yaml b/docs/samples/kubecon-2019-china/deploys/controllers.yaml
@@ -83,12 +83,19 @@ spec:
     spec:
       serviceAccount: vk-controllers
       imagePullSecrets:
-          - name: default-secret
+        - name: default-secret
       containers:
-          - name: vk-controllers
-            image: swr.cn-north-1.myhuaweicloud.com/hwstaff_z00383385/vk-controllers:latest
-            args:
-              - --alsologtostderr
-              - -v=4
-              - 2>&1
-            imagePullPolicy: "IfNotPresent"
+        - name: vk-controllers
+          image: swr.cn-north-1.myhuaweicloud.com/hwstaff_z00383385/vk-controllers:latest
+          args:
+            - --alsologtostderr
+            - -v=3
+            - 2>&1
+          imagePullPolicy: "IfNotPresent"
+          resources:
+            requests:
+              cpu: "1000m"
+              memory: "2048Mi"
+            limits:
+              cpu: "1000m"
+              memory: "2048Mi"
diff --git a/docs/samples/kubecon-2019-china/deploys/scheduler.yaml b/docs/samples/kubecon-2019-china/deploys/scheduler.yaml
@@ -4,18 +4,18 @@ metadata:
   name: vk-scheduler-configmap
   namespace: volcano
 data:
-    kube-batch.conf: |
-      actions: "enqueue, allocate, backfill"
-      tiers:
-      - plugins:
-        - name: priority
-        - name: gang
-        - name: conformance
-      - plugins:
-        - name: drf
-        - name: predicates
-        - name: proportion
-        - name: nodeorder
+  kube-batch.conf: |
+    actions: "enqueue, allocate, backfill"
+    tiers:
+    - plugins:
+      - name: priority
+      - name: gang
+      - name: conformance
+    - plugins:
+      - name: drf
+      - name: predicates
+      - name: proportion
+      - name: nodeorder
 ---
 apiVersion: v1
 kind: ServiceAccount
@@ -113,13 +113,21 @@ spec:
         - name: default-secret
       containers:
         - name: vk-scheduler
-          image: swr.cn-north-1.myhuaweicloud.com/hwstaff_z00383385/vk-scheduler:latest
+          image: swr.cn-north-1.myhuaweicloud.com/hwstaff_z00383385/vk-kube-batch:latest
           args:
             - --alsologtostderr
             - --scheduler-conf=/volcano.scheduler/kube-batch.conf
+            - --scheduler-name=volcano
             - -v=3
             - 2>&1
           imagePullPolicy: "IfNotPresent"
+          resources:
+            requests:
+              cpu: "1000m"
+              memory: "2048Mi"
+            limits:
+              cpu: "1000m"
+              memory: "2048Mi"
           volumeMounts:
             - name: scheduler-config
               mountPath: /volcano.scheduler

diff --git a/docs/samples/kubecon-2019-china/drf/nginx-1.yaml b/docs/samples/kubecon-2019-china/drf/nginx-1.yaml
@@ -0,0 +1,26 @@
+apiVersion: apps/v1
+kind: ReplicaSet
+metadata:
+  name: nginx-1
+  labels:
+    app: nginx-1
+spec:
+  # modify replicas according to your case
+  replicas: 8
+  selector:
+    matchLabels:
+      app: nginx-1
+  template:
+    metadata:
+      labels:
+        app: nginx-1
+    spec:
+      schedulerName: volcano
+      containers:
+      - name: nginx-1
+        image: nginx
+        resources:
+          requests:
+            cpu: "1000m"
+          limits:
+            cpu: "1000m"
diff --git a/docs/samples/kubecon-2019-china/drf/nginx-2.yaml b/docs/samples/kubecon-2019-china/drf/nginx-2.yaml
@@ -0,0 +1,26 @@
+apiVersion: apps/v1
+kind: ReplicaSet
+metadata:
+  name: nginx-2
+  labels:
+    app: nginx-2
+spec:
+  # modify replicas according to your case
+  replicas: 8
+  selector:
+    matchLabels:
+      app: nginx-2
+  template:
+    metadata:
+      labels:
+        app: nginx-2
+    spec:
+      schedulerName: volcano
+      containers:
+      - name: nginx-2
+        image: nginx
+        resources:
+          requests:
+            cpu: "1000m"
+          limits:
+            cpu: "1000m"
diff --git a/docs/samples/kubecon-2019-china/drf/nginx.yaml b/docs/samples/kubecon-2019-china/drf/nginx.yaml
@@ -0,0 +1,25 @@
+apiVersion: apps/v1
+kind: ReplicaSet
+metadata:
+  name: nginx
+  labels:
+    app: nginx
+spec:
+  # modify replicas according to your case
+  replicas: 8
+  selector:
+    matchLabels:
+      app: nginx
+  template:
+    metadata:
+      labels:
+        app: nginx
+    spec:
+      containers:
+      - name: nginx
+        image: nginx
+        resources:
+          requests:
+            cpu: "1000m"
+          limits:
+            cpu: "1000m"
diff --git a/docs/samples/kubecon-2019-china/gang/mpi-example.yaml b/docs/samples/kubecon-2019-china/gang/mpi-example.yaml
@@ -0,0 +1,77 @@
+apiVersion: batch.volcano.sh/v1alpha1
+kind: Job
+metadata:
+  name: lm-mpi-job
+  labels:
+    # 根据业务需要设置作业类型
+    "volcano.sh/job-type": "MPI"
+spec:
+  # 设置最小需要的服务 (小于总replicas数)
+  minAvailable: 4
+  schedulerName: volcano
+  plugins:
+    # 提供 ssh 免密认证
+    ssh: []
+    # 提供运行作业所需要的网络信息，hosts文件，headless service等
+    svc: []
+  # 如果有pod被 杀死，重启整个作业
+  policies:
+    - event: PodEvicted
+      action: RestartJob
+  tasks:
+    - replicas: 1
+      name: mpimaster
+      # 当 mpiexec 结束，认识整个mpi作业结束
+      policies:
+        - event: TaskCompleted
+          action: CompleteJob
+      template:
+        spec:
+          # Volcano 的信息会统一放到 /etc/volcano 目录下
+          containers:
+            - command:
+                - /bin/sh
+                - -c
+                - |
+                  MPI_HOST=`cat /etc/volcano/mpiworker.host | tr "\n" ","`;
+                  mkdir -p /var/run/sshd; /usr/sbin/sshd;
+                  mpiexec --allow-run-as-root --host ${MPI_HOST} -np 3 mpi_hello_world;
+              image: swr.cn-north-1.myhuaweicloud.com/hwstaff_z00383385/example-mpi:0.0.1
+              name: mpimaster
+              ports:
+                - containerPort: 22
+                  name: mpijob-port
+              workingDir: /home
+              resources:
+                requests:
+                  cpu: "500m"
+                limits:
+                  cpu: "500m"
+          restartPolicy: OnFailure
+          imagePullSecrets:
+            - name: default-secret
+    - replicas: 3
+      name: mpiworker
+      template:
+        spec:
+          containers:
+            - command:
+                - /bin/sh
+                - -c
+                - |
+                  mkdir -p /var/run/sshd; /usr/sbin/sshd -D;
+              image: swr.cn-north-1.myhuaweicloud.com/hwstaff_z00383385/example-mpi:0.0.1
+              name: mpiworker
+              ports:
+                - containerPort: 22
+                  name: mpijob-port
+              workingDir: /home
+              resources:
+                requests:
+                  cpu: "1000m"
+                limits:
+                  cpu: "1000m"
+          restartPolicy: OnFailure
+          imagePullSecrets:
+            - name: default-secret
+
diff --git a/docs/samples/kubecon-2019-china/gang/nginx.yaml b/docs/samples/kubecon-2019-china/gang/nginx.yaml
@@ -0,0 +1,25 @@
+apiVersion: apps/v1
+kind: ReplicaSet
+metadata:
+  name: nginx
+  labels:
+    app: nginx
+spec:
+  # modify replicas according to your case
+  replicas: 6
+  selector:
+    matchLabels:
+      app: nginx
+  template:
+    metadata:
+      labels:
+        app: nginx
+    spec:
+      containers:
+      - name: nginx
+        image: nginx
+        resources:
+          requests:
+            cpu: "1000m"
+          limits:
+            cpu: "1000m"
diff --git a/docs/samples/kubecon-2019-china/horovod-sample/lm-horovod-tf-mnist-v0.5.yaml b/docs/samples/kubecon-2019-china/horovod-sample/lm-horovod-tf-mnist-v0.5.yaml
@@ -2,12 +2,18 @@ apiVersion: batch.volcano.sh/v1alpha1
 kind: Job
 metadata:
   name: lm-horovod-job
+  labels:
+    "volcano.sh/job-type": Horovod
 spec:
   minAvailable: 4
-  schedulerName: kube-batch
+  schedulerName: volcano
   plugins:
     ssh: []
     svc: []
+  # 如果有pod被 杀死，重启整个作业
+  policies:
+    - event: PodEvicted
+      action: RestartJob
   tasks:
     - replicas: 1
       name: master
@@ -56,11 +62,11 @@ spec:
                   name: job-port
               resources:
                 requests:
-                  cpu: "2048m"
-                  memory: "4096Mi"
+                  cpu: "1000m"
+                  memory: "2048Mi"
                 limits:
-                  cpu: "2048m"
-                  memory: "4096Mi"
+                  cpu: "1000m"
+                  memory: "2048Mi"
           restartPolicy: OnFailure
           imagePullSecrets:
             - name: default-secret

diff --git a/docs/samples/kubecon-2019-china/mpi-sample/mpi-example.yaml b/docs/samples/kubecon-2019-china/mpi-sample/mpi-example.yaml
@@ -2,28 +2,40 @@ apiVersion: batch.volcano.sh/v1alpha1
 kind: Job
 metadata:
   name: lm-mpi-job
+  labels:
+    # 根据业务需要设置作业类型
+    "volcano.sh/job-type": "MPI"
 spec:
-  minAvailable: 4
-  schedulerName: kube-batch
+  # 设置最小需要的服务 (小于总replicas数)
+  minAvailable: 3
+  schedulerName: volcano
   plugins:
+    # 提供 ssh 免密认证
     ssh: []
+    # 提供运行作业所需要的网络信息，hosts文件，headless service等
     svc: []
+  # 如果有pod被 杀死，重启整个作业
+  policies:
+    - event: PodEvicted
+      action: RestartJob
   tasks:
     - replicas: 1
       name: mpimaster
+      # 当 mpiexec 结束，认识整个mpi作业结束
       policies:
         - event: TaskCompleted
           action: CompleteJob
       template:
         spec:
+          # Volcano 的信息会统一放到 /etc/volcano 目录下
           containers:
             - command:
                 - /bin/sh
                 - -c
                 - |
                   MPI_HOST=`cat /etc/volcano/mpiworker.host | tr "\n" ","`;
                   mkdir -p /var/run/sshd; /usr/sbin/sshd;
-                  mpiexec --allow-run-as-root --host ${MPI_HOST} -np 3 mpi_hello_world;
+                  mpiexec --allow-run-as-root --host ${MPI_HOST} -np 2 mpi_hello_world;
               image: swr.cn-north-1.myhuaweicloud.com/hwstaff_z00383385/example-mpi:0.0.1
               name: mpimaster
               ports:
@@ -40,7 +52,7 @@ spec:
           restartPolicy: OnFailure
           imagePullSecrets:
             - name: default-secret
-    - replicas: 3
+    - replicas: 2
       name: mpiworker
       template:
         spec:
@@ -58,12 +70,12 @@ spec:
               workingDir: /home
               resources:
                 requests:
-                  cpu: "2048m"
-                  memory: "4096Mi"
+                  cpu: "1024m"
+                  memory: "2048Mi"
                 limits:
-                  cpu: "2048m"
-                  memory: "4096Mi"
+                  cpu: "1024m"
+                  memory: "2048Mi"
           restartPolicy: OnFailure
           imagePullSecrets:
             - name: default-secret
----
+