Apache spark sql SparkSQL：相同的查询返回不同的结果

我遇到了一个奇怪的问题。我想从一个数据框中获取所有数据，并将其插入到一个永久的配置单元表中，并将其索引到elasticsearch。查询很简单，即

select*from result*

，我将循环遍历每一行并插入ES。简单的

插入select*from result

，但我得到了不同的结果。为了检查，我创建了3个不同的临时表，如下所示

spark.sql("select * from QtyContribution").join(getRevenueContribution(spark,table2), "item").join(finalUniqueItem(spark), "item").registerTempTable("hola");
spark.sql("select * from QtyContribution").join(getRevenueContribution(spark,table2), "item").join(finalUniqueItem(spark), "item").registerTempTable("hola1");
spark.sql("select * from QtyContribution").join(getRevenueContribution(spark,table2), "item").join(finalUniqueItem(spark), "item").registerTempTable("hola2");

每个查询都是相同的，只是表不同。及

Dataset<Row> dframe1 = spark.sql("select * from hola");
        Row[] row1 = (Row[]) dframe1.collect();
        int q=1;
        for(Row s : row1){
            System.out.println(s.get(0)+" =======df1======= "+ q++);

        }

Dataset<Row> dframe2 = spark.sql("select * from hola1");
        Row[] row2 = (Row[]) dframe2.collect();
        int w=1;
        for(Row s : row2){
            System.out.println(s.get(0)+" =======df2======= "+ w++);

        }

Dataset<Row> dframe3 = spark.sql("select * from hola2");
        Row[] row3 = (Row[]) dframe3.collect();
        int e=1;
        for(Row s : row2){
            System.out.println(s.get(0)+" =======df3======= "+ e++);

        }

我做了些什么

Dataset<Row> dframe = spark.sql("select * from hola1");
        Row[] row = (Row[]) dframe.collect();
        int i = 1;
        for (Row r : row) {
        bulkRequest.add(client.prepareIndex("twitter1234", "use1", String.valueOf(i))
                .setSource(jsonBuilder()
                        .startObject()
                        .field("item", r.get(0))
                        .field("qty_contrib", r.get(1))
                        .field("division", r.get(2))
                        .field("rev_contrib", r.get(3))
                        .field("bp", r.get(4))
                        .endObject()
                )
        );
            System.out.println(i++ +" ==== "+r.get(0));
        }

发生了什么事

Dataset<Row> dframe = spark.sql("select * from hola1");
        Row[] row = (Row[]) dframe.collect();
        int i = 1;
        for (Row r : row) {
        bulkRequest.add(client.prepareIndex("twitter1234", "use1", String.valueOf(i))
                .setSource(jsonBuilder()
                        .startObject()
                        .field("item", r.get(0))
                        .field("qty_contrib", r.get(1))
                        .field("division", r.get(2))
                        .field("rev_contrib", r.get(3))
                        .field("bp", r.get(4))
                        .endObject()
                )
        );
            System.out.println(i++ +" ==== "+r.get(0));
        }

1534 ==== BM8942