Question

我正在使用Kundera（3.13）在Java（8）中为Cassandra（3.11）数据库建模，并且被卡在使用复合分区键的表上。键空间/表的定义是这样的：

CREATE KEYSPACE test
WITH replication = {'class':'SimpleStrategy', 'replication_factor' : 3};

CREATE TABLE test.timeseriesdata (
  asset_id                  TEXT,
  time_series_definition_id TEXT,
  business_date_year        INT,
  business_date             DATE,
  system_time               TIMESTAMP,
  values_blob               MAP<TEXT, BLOB>,
  values_boolean            MAP<TEXT, BOOLEAN>,
  values_double             MAP<TEXT, DOUBLE>,
  values_int                MAP<TEXT, INT>,
  values_text               MAP<TEXT, TEXT>,
  values_timestamp          MAP<TEXT, TIMESTAMP>,
  PRIMARY KEY ((asset_id, time_series_definition_id, business_date_year), business_date, system_time)
) WITH CLUSTERING ORDER BY (business_date DESC, system_time DESC);

Kundera实体看起来像这样：

@Embeddable
public class TimeSeriesDataPartitionKey implements Serializable {

    @Column(name = "asset_id")
    private String assetId;

    @Column(name = "time_series_definition_id")
    private String timeSeriesDefinitionId;

    @Column(name = "business_date_year")
    private Integer businessDateYear;
}

@Embeddable
public class TimeSeriesDataPrimaryKey implements Serializable {

    @Embedded
    private TimeSeriesDataPartitionKey key;

    @Column(name = "business_date")
    @Temporal(TemporalType.DATE)
    private Date businessDate;

    @Column(name = "system_time")
    @Temporal(TemporalType.TIMESTAMP)
    private Date systemTime;
}

@Entity
@Table(name = "timeseriesdata", schema = "test@cassandra_pu")
public class TimeSeriesDataEntity {

    @EmbeddedId
    private TimeSeriesDataPrimaryKey key;

    @ElementCollection
    @Column(name = "values_blob")
    private Map<String, byte[]> valuesBlob = new HashMap<>();

    @ElementCollection
    @Column(name = "values_boolean")
    private Map<String, Boolean> valuesBoolean = new HashMap<>();

    @ElementCollection
    @Column(name = "values_double")
    private Map<String, Double> valuesDouble = new HashMap<>();

    @ElementCollection
    @Column(name = "values_int")
    private Map<String, Integer> valuesInt = new HashMap<>();

    @ElementCollection
    @Column(name = "values_text")
    private Map<String, String> valuesText = new HashMap<>();

    @ElementCollection
    @Column(name = "values_timestamp")
    @Temporal(TemporalType.TIMESTAMP)
    private Map<String, Date> valuesTimestamp = new HashMap<>();
}

我正在使用<property name="kundera.ddl.auto.prepare" value="validate"/> 在persistence.xml文件中，以确保实体建模正确。当我运行一些代码尝试添加一些数据时，它失败并显示消息

com.impetus.kundera.configure.schema.SchemaGenerationException：com.impetus.kundera.configure.schema.SchemaGenerationException：模式不匹配！验证失败。参见上表以了解不匹配情况

这里a link到diffchecker表中，在该表中粘贴了我收到的架构不匹配消息。另外，在像我一样对Kundera实体建模时，我尝试使用here，here和here中的资源。我只是无法理解如何做，我给我的老师发了邮件，他告诉我展平模型，因为制作复合钥匙是不可能的，而现在我的头脑更加模糊。

编辑

如果我使用<property name="kundera.ddl.auto.prepare" value="create"/> 然后代码运行，如果我生成了sql备份，它看起来像这样

CREATE TABLE test.timeseriesdata (
  asset_id                  TEXT,
  time_series_definition_id TEXT,
  business_date_year        INT,
  business_date             TIMESTAMP,
  system_time               TIMESTAMP,
  values_blob               MAP<TEXT, BLOB>,
  values_boolean            MAP<TEXT, BOOLEAN>,
  values_double             MAP<TEXT, DOUBLE>,
  values_int                MAP<TEXT, INT>,
  values_text               MAP<TEXT, TEXT>,
  values_timestamp          MAP<TEXT, TIMESTAMP>,
  PRIMARY KEY ((asset_id, time_series_definition_id, business_date_year), business_date, system_time)
);

所以区别是business_date type is TIMESTAMP not DATE和WITH CLUSTERING ORDER BY丢失了。

所以现在的问题是：cassandra DATE应该使用哪种Java类型？为什么缺少聚类顺序？

使用Kundera对Cassandra复合分区键进行建模

0 个答案: