Merge pull request #92 from wangjiawei04/master

update ctr training yamls

doc/resource/ctr.yaml

@@ -26,7 +26,7 @@ spec:
               type: ""
             name: seqdata
           containers:
-            - image: wangjiawei1993/edldemo:v19
+            - image: hub.baidubce.com/ctr/edldemo:latest
               command:
                 - paddle_k8s
                 - start_fluid
@@ -116,7 +116,7 @@ spec:
               type: ""
             name: seqdata
           containers:
-            - image: wangjiawei1993/edldemo:v19
+            - image: hub.baidubce.com/ctr/edldemo:latest
               command:
                 - paddle_k8s
                 - start_fluid
@@ -206,7 +206,7 @@ spec:
               type: ""
             name: seqdata
           containers:
-            - image: wangjiawei1993/edldemo:v19
+            - image: hub.baidubce.com/ctr/edldemo:latest
               command:
                 - paddle_k8s
                 - start_fluid

doc/resource/cube.yaml

@@ -7,7 +7,7 @@ metadata:
 spec:
   containers:
   - name: cube-0
-    image: wangjiawei1993/cube:v11
+    image: hub.baidubce.com/ctr/cube:latest
     workingDir: /cube
     command: ['/bin/bash']
     args: ['start.sh']
@@ -28,7 +28,7 @@ metadata:
 spec:
   containers:
   - name: cube-1
-    image: wangjiawei1993/cube:v11
+    image: hub.baidubce.com/ctr/cube:latest
     workingDir: /cube
     command: ['/bin/bash']
     args: ['start.sh']

doc/resource/ftp.yaml

@@ -12,7 +12,7 @@ spec:
     name: file-home
   containers:
   - name: file-server
-    image: halverneus/static-file-server
+    image: hub.baidubce.com/ctr/file-server:latest
     ports:
     - containerPort: 8080
     volumeMounts:

doc/resource/paddle-suite.yaml

@@ -12,7 +12,7 @@ spec:
     name: file-home
   containers:
   - name: file-server
-    image: halverneus/static-file-server
+    image: hub.baidubce.com/ctr/file-server:latest
     ports:
     - containerPort: 8080
     volumeMounts:
@@ -37,7 +37,6 @@ spec:
 
 ---
 
-
 apiVersion: v1
 kind: Pod
 metadata:
@@ -47,7 +46,7 @@ metadata:
 spec:
   containers:
   - name: cube-0
-    image: wangjiawei1993/cube:v11
+    image: hub.baidubce.com/ctr/cube:latest
     workingDir: /cube
     command: ['/bin/bash']
     args: ['start.sh']
@@ -68,7 +67,7 @@ metadata:
 spec:
   containers:
   - name: cube-1
-    image: wangjiawei1993/cube:v11
+    image: hub.baidubce.com/ctr/cube:latest
     workingDir: /cube
     command: ['/bin/bash']
     args: ['start.sh']
@@ -123,7 +122,7 @@ metadata:
 spec:
   containers:
   - name: cube-transfer
-    image: wangjiawei1993/cube-transfer:v18
+    image: hub.baidubce.com/ctr/cube-transfer:latest
     workingDir: /
     env:
       - name: POD_IP
@@ -150,7 +149,7 @@ metadata:
 spec:
   containers:
   - name: paddleserving
-    image: wangjiawei1993/paddleserving:v7-debug
+    image: hub.baidubce.com/ctr/paddleserving:latest
     workingDir: /serving
     command: ['/bin/bash']
     args: ['run.sh']
@@ -183,7 +182,7 @@ metadata:
 spec:
   containers:
   - name: pdservingclient
-    image: wangjiawei1993/pdservingclient:v4
+    image: hub.baidubce.com/ctr/pdservingclient:latest
     workingDir: /
     command: ['bash']
     args: ['nonstop.sh']
@@ -196,7 +195,7 @@ kind: Job
 metadata:
   name: edl-demo
 spec:
-  minAvailable: 6
+  minAvailable: 4
   schedulerName: volcano
   policies:
   - event: PodEvicted
@@ -204,7 +203,7 @@ spec:
   - event: PodFailed
     action: RestartJob
   tasks:
-    - replicas: 3
+    - replicas: 2
       name: pserver
       template:
         metadata:
@@ -219,7 +218,7 @@ spec:
               type: ""
             name: seqdata
           containers:
-            - image: wangjiawei1993/edldemo:v19
+            - image: hub.baidubce.com/ctr/edldemo:latest
               command:
                 - paddle_k8s
                 - start_fluid
@@ -273,9 +272,9 @@ spec:
                 - name: PADDLE_IS_LOCAL
                   value: "0"
                 - name: PADDLE_TRAINERS_NUM
-                  value: "3"
+                  value: "2"
                 - name: PADDLE_PSERVERS_NUM
-                  value: "3"
+                  value: "2"
                 - name: FLAGS_rpc_deadline
                   value: "36000000"
                 - name: ENTRY
@@ -309,7 +308,7 @@ spec:
               type: ""
             name: seqdata
           containers:
-            - image: wangjiawei1993/edldemo:v19
+            - image: hub.baidubce.com/ctr/edldemo:latest
               command:
                 - paddle_k8s
                 - start_fluid
@@ -368,9 +367,9 @@ spec:
                 - name: PADDLE_PORT
                   value: "30240"
                 - name: PADDLE_PSERVERS_NUM
-                  value: "3"
+                  value: "2"
                 - name: PADDLE_TRAINERS_NUM
-                  value: "3"
+                  value: "2"
                 - name: PADDLE_TRAINING_ROLE
                   value: TRAINER
                 - name: TRAINING_ROLE
@@ -381,7 +380,7 @@ spec:
                   value: (/postprocess &) && cd /workspace/ctr && python train.py --is_local 0 --cloud_train 1
           restartPolicy: OnFailure
 
-    - replicas: 2
+    - replicas: 1
       policies:
       - event: TaskCompleted
         action: CompleteJob
@@ -399,7 +398,7 @@ spec:
               type: ""
             name: seqdata
           containers:
-            - image: wangjiawei1993/edldemo:v19
+            - image: hub.baidubce.com/ctr/edldemo:latest
               command:
                 - paddle_k8s
                 - start_fluid
@@ -458,9 +457,9 @@ spec:
                 - name: PADDLE_PORT
                   value: "30240"
                 - name: PADDLE_PSERVERS_NUM
-                  value: "3"
+                  value: "2"
                 - name: PADDLE_TRAINERS_NUM
-                  value: "3"
+                  value: "2"
                 - name: PADDLE_TRAINING_ROLE
                   value: TRAINER
                 - name: TRAINING_ROLE
@@ -470,3 +469,4 @@ spec:
                 - name: ENTRY
                   value: (/postprocess &) && cd /workspace/ctr && python train.py --is_local 0 --cloud_train 1
           restartPolicy: OnFailure
+

doc/resource/pdclient.yaml

@@ -7,7 +7,7 @@ metadata:
 spec:
   containers:
   - name: pdservingclient
-    image: wangjiawei1993/pdservingclient:v4
+    image: hub.baidubce.com/ctr/pdservingclient:latest
     workingDir: /
     command: ['bash']
     args: ['nonstop.sh']

doc/resource/pdserving.yaml

@@ -7,7 +7,7 @@ metadata:
 spec:
   containers:
   - name: paddleserving
-    image: wangjiawei1993/paddleserving:v7-debug
+    image: hub.baidubce.com/ctr/paddleserving:latest 
     workingDir: /serving
     command: ['/bin/bash']
     args: ['run.sh']

doc/resource/transfer.yaml

@@ -7,7 +7,7 @@ metadata:
 spec:
   containers:
   - name: cube-transfer
-    image: wangjiawei1993/cube-transfer:v18
+    image: hub.baidubce.com/ctr/cube-transfer:latest
     workingDir: /
     env:
       - name: POD_IP

doc/resource/volcano-ctr-demo-baiduyun.yaml

-apiVersion: batch.volcano.sh/v1alpha1
-kind: Job
-metadata:
-  name: edl-demo
-spec:
-  minAvailable: 4
-  schedulerName: volcano
-  policies:
-  - event: PodEvicted
-    action: RestartJob
-  - event: PodFailed
-    action: RestartJob
-  tasks:
-    - replicas: 2
-      name: pserver
-      template:
-        metadata:
-          labels:
-            paddle-job-pserver: fluid-ctr
-        spec:
-          imagePullSecrets:
-            - name: default-secret
-          volumes:
-          - hostPath:
-              path: /home/work/
-              type: ""
-            name: seqdata
-          containers:
-            - image: wangjiawei1993/edldemo:v19
-              command:
-                - paddle_k8s
-                - start_fluid
-              imagePullPolicy: IfNotPresent
-              name: pserver
-              volumeMounts:
-              - mountPath: /data
-                name: seqdata
-              resources:
-                limits:
-                  cpu: 10
-                  memory: 30Gi
-                  ephemeral-storage: 10Gi
-                requests:
-                  cpu: 1
-                  memory: 100M
-                  ephemeral-storage: 1Gi
-              env:
-                - name: GLOG_v
-                  value: "0"
-                - name: GLOG_logtostderr
-                  value: "1"
-                - name: TOPOLOGY
-                  value: ""
-                - name: TRAINER_PACKAGE
-                  value: /workspace
-                - name: PADDLE_INIT_NICS
-                  value: eth2
-                - name: NAMESPACE
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: metadata.namespace
-                - name: POD_IP
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: status.podIP
-                - name: POD_NAME
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: metadata.name
-                - name: PADDLE_CURRENT_IP
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: status.podIP
-                - name: PADDLE_JOB_NAME
-                  value: fluid-ctr
-                - name: PADDLE_IS_LOCAL
-                  value: "0"
-                - name: PADDLE_TRAINERS_NUM
-                  value: "2"
-                - name: PADDLE_PSERVERS_NUM
-                  value: "2"
-                - name: FLAGS_rpc_deadline
-                  value: "36000000"
-                - name: ENTRY
-                  value: cd /workspace/ctr && python train.py --is_local 0 --cloud_train 1
-                - name: PADDLE_PORT
-                  value: "30240"
-                - name: LD_LIBRARY_PATH
-                  value: /usr/local/lib:/usr/local/nvidia/lib64:/usr/local/rdma/lib64:/usr/lib64/mlnx_ofed/valgrind
-                - name: PADDLE_TRAINING_ROLE
-                  value: PSERVER
-                - name: TRAINING_ROLE
-                  value: PSERVER
-          restartPolicy: OnFailure 
-    - replicas: 2
-      policies:
-      - event: TaskCompleted
-        action: CompleteJob
-      name: trainer
-      template:
-        metadata:
-          labels:
-            paddle-job: fluid-ctr
-        spec:
-          nodeSelector:
-            nodeType: model
-          imagePullSecrets:
-            - name: default-secret
-          volumes:
-          - hostPath:
-              path: /home/work/data
-              type: ""
-            name: seqdata
-          containers:
-            - image: wangjiawei1993/edldemo:v19
-              command:
-                - paddle_k8s
-                - start_fluid
-              imagePullPolicy: IfNotPresent
-              name: trainer
-              volumeMounts:
-              - mountPath: /data
-                name: seqdata
-              resources:
-                limits:
-                  cpu: 10
-                  memory: 30Gi
-                  ephemeral-storage: 10Gi
-                requests:
-                  cpu: 1
-                  memory: 100M
-                  ephemeral-storage: 10Gi
-              env:
-                - name: GLOG_v
-                  value: "0"
-                - name: GLOG_logtostderr
-                  value: "1"
-                - name: TOPOLOGY
-                - name: TRAINER_PACKAGE
-                  value: /workspace
-                - name: PADDLE_INIT_NICS
-                  value: eth2
-                - name: CPU_NUM
-                  value: "2"
-                - name: NAMESPACE
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: metadata.namespace
-                - name: POD_IP
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: status.podIP
-                - name: POD_NAME
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: metadata.name
-                - name: PADDLE_CURRENT_IP
-                  valueFrom:
-                    fieldRef:
-                      apiVersion: v1
-                      fieldPath: status.podIP
-                - name: PADDLE_JOB_NAME
-                  value: fluid-ctr
-                - name: PADDLE_IS_LOCAL
-                  value: "0"
-                - name: FLAGS_rpc_deadline
-                  value: "36000000"
-                - name: PADDLE_PORT
-                  value: "30240"
-                - name: PADDLE_PSERVERS_NUM
-                  value: "2"
-                - name: PADDLE_TRAINERS_NUM
-                  value: "2"
-                - name: PADDLE_TRAINING_ROLE
-                  value: TRAINER
-                - name: TRAINING_ROLE
-                  value: TRAINER
-                - name: LD_LIBRARY_PATH
-                  value: /usr/local/lib:/usr/local/nvidia/lib64:/usr/local/rdma/lib64:/usr/lib64/mlnx_ofed/valgrind
-                - name: ENTRY
-                  value: (/postprocess &) && cd /workspace/ctr && python train.py --is_local 0 --cloud_train 1
-          restartPolicy: OnFailure