启动Hive独立元数据存储3.0.0时PartitionExpressionProxy加载失败

在Kubernetes中部署连接启用SSL的PostgreSQL的HMS（独立版v3.0.0）时，执行schematool -dbType postgres -initSchema可成功初始化PostgreSQL中的所需表，但运行./start-metastore时出现错误：Error loading PartitionExpressionProxy: org.apache.hadoop.hive.metastore.DefaultPartitionExpressionProxy class not found。

配置文件 #

HMS Dockerfile #

FROM alpine:latest

COPY hadoop-3.1.0.tar.gz /opt/hadoop-3.1.0.tar.gz
COPY hive-standalone-metastore-3.0.0-bin.tar.gz /opt/hive-standalone-metastore-3.0.0-bin.tar.gz

RUN wget -O /etc/apk/keys/amazoncorretto.rsa.pub  https://apk.corretto.aws/amazoncorretto.rsa.pub  && \
    echo "https://apk.corretto.aws/" >> /etc/apk/repositories  && \
    apk update && apk upgrade && \
    apk add --no-cache curl bash && \
    apk add amazon-corretto-8 && \
    apk add amazon-corretto-8-jre && \
    tar -xvf /opt/hive-standalone-metastore-3.0.0-bin.tar.gz -C /opt/ && \
    tar -xvf /opt/hadoop-3.1.0.tar.gz -C /opt/  && \
    rm /opt/*.tar.gz

COPY postgresql-42.5.4.jar /opt/apache-hive-metastore-3.0.0-bin/lib

ENV HIVE_HOME=/opt/apache-hive-metastore-3.0.0-bin \
    CLASSPATH=/opt/apache-hive-metastore-3.0.0-bin/lib/ \
    HADOOP_HOME=/opt/hadoop-3.1.0 \
    JAVA_HOME=/usr/lib/jvm/java-8-amazon-corretto

metastore-site.xml #

<configuration>
        <!--    Configure MetaStore as a server. To start more servers, change port-->
        <property>
            <name>metastore.thrift.uris</name>
            <value>thrift://meta-store.default.svc.cluster.local:9083</value>
            <description>
                - Thrift URI for the remote metastore. Used by metastore client to connect to remote metastore.
                - IP address (or fully-qualified domain name) and port of the metastore host
            </description>
        </property>
        <!--    For running MetaStore in standalone mode. -->
        <property>
            <name>metastore.task.threads.always
            </name>
            <value>org.apache.hadoop.hive.metastore.events.EventCleanerTask</value>
        </property>
        <property>
            <name>metastore.expression.proxy</name>
            <value>org.apache.hadoop.hive.metastore.DefaultPartitionExpressionProxy
            </value>
        </property>
        <!--Postgres' connection details -->
        <property>
            <name>javax.jdo.option.ConnectionURL</name>
            <value>jdbc:postgresql://postgres-svc.default.svc.cluster.local:5432/hive?ssl=true&amp;sslMode=verify-ca&amp;sslcert=/etc/tls.crt&amp;sslkey=/etc/postgresql.pk8&amp;sslrootcert=/etc/root.crt&amp;sslpassword=postgres</value>
        </property>
        <property>
            <name>javax.jdo.option.ConnectionDriverName</name>
            <value>org.postgresql.Driver</value>
        </property>
        <property>
            <name>javax.jdo.option.ConnectionUserName</name>
            <value>postgres</value>
        </property>
        <property>
            <name>javax.jdo.option.ConnectionPassword</name>
            <value>postgres</value>
        </property>
        
        
        <!--URL of the default location for -->
        <!--    <property>-->
        <!--        <name>hive.metastore.warehouse.dir</name>-->
        <!--        <value>hdfs://localhost:9000/user/hive/warehouse</value>-->
        <!--        <description>-->
        <!--            URI of the default location for tables in the default catalog and database.-->
        <!--        </description>-->
        <!--    </property>-->
        
        <property>
            <name>datanucleus.schema.autoCreateAll</name>
            <value>false</value>
            <description>
                - Auto creates the necessary schema in the RDBMS at startup if one does not exist.
                - Set this to false after creating it once.
                - To enable auto create also set hive.metastore.schema.verification=false.
                - Auto creation is not recommended in production; run schematool instead.
            </description>
        </property>
        <property>
            <name>datanucleus.autoStartMechanism</name>
            <value>SchemaTable</value>
        </property>
        <property>
            <name>datanucleus.fixedDatastore</name>
            <value>true</value>
        </property>
    
    </configuration>

Kubernetes部署YAML #

apiVersion: apps/v1
kind: Deployment
metadata:
  name: hive-meta-store
  namespace: default
spec:
  replicas: 1
  selector:
    matchLabels:
      app: hive-metastore-deployment
  template:
    metadata:
      labels:
        app: hive-metastore-deployment
    spec:
      securityContext:
        runAsUser: 0
      containers:
        - name: metastore
          image: hive-metastore-v0.0.1
          imagePullPolicy: "IfNotPresent"
          ports:
            - containerPort: 9083
              name: thrift-port
#          envFrom:
#            - configMapRef:
#                name: postgres-config
          volumeMounts:
            - name: metastore-xml
              mountPath: /opt/apache-hive-metastore-3.0.0-bin/conf/metastore-site.xml
              subPath: metastore-site.xml
            - name: postgres-ssl-crt
              mountPath: /etc/tls.crt
              subPath: tls.crt
            - name: postgres-ssl-key
              mountPath: /etc/tls.key
              subPath: tls.key
            - name: postgres-ssl-root
              mountPath: /etc/root.crt
              subPath: root.crt
            - name: postgres-ssl-pk8
              mountPath: /etc/postgresql.pk8
              subPath: postgresql.pk8
          command:
            - /bin/sh
            - -c
            - "watch ps"
      volumes:
        - name: metastore-xml
          configMap:
            name: metastore-site.xml
        - name: postgres-ssl-crt
          secret:
            secretName: postgres-ssl-secret
            items:
              - key: tls.crt
                path: tls.crt
                mode: 0444
        - name: postgres-ssl-key
          secret:
            secretName: postgres-ssl-secret
            items:
              - key: tls.key
                path: tls.key
                mode: 0444
        - name: postgres-ssl-root
          secret:
            secretName: postgres-ssl-secret
            items:
              - key: root.crt
                path: root.crt
                mode: 0444
        - name: postgres-ssl-pk8
          secret:
            secretName: postgres-ssl-secret
            items:
              - key: postgresql.pk8
                path: postgresql.pk8
                mode: 0444

解决办法 #

• 修复CLASSPATH环境变量：Dockerfile中设置的CLASSPATH=/opt/apache-hive-metastore-3.0.0-bin/lib/无法加载目录下的所有JAR包，Java类路径需要明确指定JAR文件。修改为：

  CLASSPATH=$HIVE_HOME/lib/*:$HADOOP_HOME/share/hadoop/common/*:$HADOOP_HOME/share/hadoop/hdfs/*:$HADOOP_HOME/share/hadoop/common/lib/*
  

  确保HMS启动时能加载所有依赖类。

• 修正metastore-site.xml中的类名格式：metastore.expression.proxy配置项的value值存在换行，会导致类名解析错误。将其改为一行无多余空格的类名：

  <property>
      <name>metastore.expression.proxy</name>
      <value>org.apache.hadoop.hive.metastore.DefaultPartitionExpressionProxy</value>
  </property>
  

• 修正Kubernetes部署的启动命令：当前YAML中的watch ps并未启动HMS服务，替换为正确的启动命令：

  command:
    - /bin/sh
    - -c
    - "$HIVE_HOME/bin/start-metastore"
  

• 校验HMS包完整性：确认hive-standalone-metastore-3.0.0-bin.tar.gz解压后的$HIVE_HOME/lib目录中存在hive-metastore-3.0.0.jar，该JAR包包含DefaultPartitionExpressionProxy类。若缺失，重新下载完整的HMS安装包。

内容的提问来源于stack exchange，提问作者jake wong