Deploying AI-VAULT

Prerequisites

Installation preparations have been completed. For details, see Installation Preparations.

Procedure

1. Log in to the Kubernetes management node as the administrator and run the following command to check whether the AI-VAULT image and version number are correct:

   docker images | grep ai-vault

   The following is an example:

   # docker images | grep ai-vault
   ai-vault                      v3.0.0              197f4e8780ca        21 minutes ago         154MB

   • If the image and version are correct, go to the next step.
   • If either the image or version is incorrect, create and distribute the image by referring to Creating an AI-VAULT Image.
2. Create the ai-vault.yaml file in the user's private directory (unreadable and unwritable by other users) on the Kubernetes master node. The following is an example of the Kubernetes startup configuration: Ensure that only one AI-VAULT container exists in a cluster.
   

   Use the name of the image created in Creating an AI-VAULT Image. Otherwise, the deployment fails.

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     name: ai-vault
     namespace: mindx
   spec:
     replicas: 1
     selector:
       matchLabels:
         app: ai-vault
     template:
       metadata:
         labels:
           app: ai-vault
         annotations:
           seccomp.security.alpha.kubernetes.io/pod: runtime/default
       spec:
         automountServiceAccountToken: false
         containers:
           - name: ai-vault
             securityContext:
               allowPrivilegeEscalation: false
               readOnlyRootFilesystem: true
               capabilities:
                 drop: ["ALL"]
               runAsUser: 9001
               runAsGroup: 9001
             image: ai-vault:{version}
             env:
             - name: POD_IP
               valueFrom:
                 fieldRef:
                   fieldPath: status.podIP
             resources:
               requests:
                 memory: 1000Mi
                 cpu: "1"
               limits:
                 memory: 8000Mi
                 cpu: "10"
             imagePullPolicy: Never
             command: [ "/bin/bash", "-c", "--"]
             args: [ "ulimit -n 1024 && ./ai-vault run" ]
             volumeMounts:
               - name: ai-vault-data
                 mountPath:  /home/AiVault/.ai-vault
         volumes:
           - name: ai-vault-data
             hostPath:
               path: /home/AiVault/.ai-vault
               type: Directory
   ---
   apiVersion: v1
   kind: Service
   metadata:
     name: ai-vault
     namespace: mindx
     labels:
       app: ai-vault
   spec:
     type: ClusterIP
     ports:
       - name: http
         port: 8180
         targetPort: 8180
         protocol: TCP
       - name: socket
         port: 8181
         targetPort: 8181
         protocol: TCP
     selector:    
       app: ai-vault

3. Change the permission on the YAML startup configuration file to 600.

   chmod 600 ai-vault.yaml

4. (Optional) Modify the startup parameters of AI-VAULT in the YAML file as required. For details about the boot parameters, see AI-VAULT Command Description. You can run the ./ai-vault -h command to view the parameter descriptions.

   vim ai-vault.yaml

5. Run the following command to start AI-VAULT:

   kubectl apply -f ai-vault.yaml

   The following is an example:

   # kubectl apply -f ai-vault.yaml 
   deployment.apps/ai-vault created
   service/ai-vault created
   # kubectl get pod -n mindx
   NAME                                  READY   STATUS     RESTARTS   AGE
   ai-vault-44534765b-hn5mx          1/1    Running   0         11s

Log Description

• Operation log file: /home/ {userName}/.ai-vault/log/ai-vault_operate.log
  • Operation logs are backed up for a maximum of 30 days.
  • A maximum of 30 operation log files can be retained after dump.
• Run log file: /home/{userName}/.ai-vault/log/ai-vault_run.log.
  • Run logs are backed up for a maximum of seven days.
  • A maximum of 30 run log files can be retained after dump.
• In the preceding command, {userName} indicates the user name created in Creating User Accounts. For example, if the user name is AiVault, the corresponding file path is /home/AiVault/.ai-vault/log/ai-vault_run.log.
• When the size of a single log file reaches 20 MB, the automatic dump function is triggered.