presto查hive报错：end index must not be greater than size 问题分析和解决

tips

presto 0.208连接hive有不少坑，请尽量不要选择这个版本。presto0.208以上的版本，jdk需要8_151+。

问题还原：

集群环境
hive 1.1.0
presto 0.208
hadoop 2.6
集群有张hive表使用hive-cli查询是OK，但是使用presto cli进行select * from table_name 同样的sql 语法查询会报错，报错核心如下：

java.lang.RuntimeException: Error fetching next at http://10.90.50.169:28080/v1/statement/20221117_120110_00027_yetvx/2 returned an invalid response: JsonResponse{statusCode=500, statusMessage=Internal Server Error, headers={content-length=[5070], content-type=[text/plain], date=[Mon, 17 Jun 2019 12:01:10 GMT]}, hasValue=false} [Error: com.fasterxml.jackson.databind.JsonMappingException: end index (4) must not be greater than size (1) (through reference chain: com.facebook.presto.client.QueryResults["data"])  at com.fasterxml.jackson.databind.JsonMappingException.wrapWithPath(JsonMappingException.java:379)  at com.fasterxml.jackson.databind.JsonMappingException.wrapWithPath(JsonMappingException.java:339)  at com.fasterxml.jackson.databind.ser.std.StdSerializer.wrapAndThrow(StdSerializer.java:343)  at com.fasterxml.jackson.databind.ser.std.BeanSerializerBase.serializeFields(BeanSerializerBase.java:698)  at com.fasterxml.jackson.databind.ser.BeanSerializer.serialize(BeanSerializer.java:155)  at com.fasterxml.jackson.databind.ser.DefaultSerializerProvider.serializeValue(DefaultSerializerProvider.java:292)  at com.fasterxml.jackson.databind.ObjectWriter$Prefetch.serialize(ObjectWriter.java:1419)  at com.fasterxml.jackson.databind.ObjectWriter.writeValue(ObjectWriter.java:940)  at io.airlift.jaxrs.JsonMapper.writeTo(JsonMapper.java:233)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor$TerminalWriterInterceptor.invokeWriteTo(WriterInterceptorExecutor.java:266)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor$TerminalWriterInterceptor.aroundWriteTo(WriterInterceptorExecutor.java:251)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor.proceed(WriterInterceptorExecutor.java:163)  at org.glassfish.jersey.server.internal.JsonWithPaddingInterceptor.aroundWriteTo(JsonWithPaddingInterceptor.java:109)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor.proceed(WriterInterceptorExecutor.java:163)  at org.glassfish.jersey.server.internal.MappableExceptionWrapperInterceptor.aroundWriteTo(MappableExceptionWrapperInterceptor.java:85)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor.proceed(WriterInterceptorExecutor.java:163)  at org.glassfish.jersey.message.internal.MessageBodyFactory.writeTo(MessageBodyFactory.java:1135)  at org.glassfish.jersey.server.ServerRuntime$Responder.writeResponse(ServerRuntime.java:662)  at org.glassfish.jersey.server.ServerRuntime$Responder.processResponse(ServerRuntime.java:395)  at org.glassfish.jersey.server.ServerRuntime$Responder.process(ServerRuntime.java:385)  at org.glassfish.jersey.server.ServerRuntime$AsyncResponder$3.run(ServerRuntime.java:884)  at org.glassfish.jersey.internal.Errors$1.call(Errors.java:272)  at org.glassfish.jersey.internal.Errors$1.call(Errors.java:268)  at org.glassfish.jersey.internal.Errors.process(Errors.java:316)  at org.glassfish.jersey.internal.Errors.process(Errors.java:298)  at org.glassfish.jersey.internal.Errors.process(Errors.java:268)  at org.glassfish.jersey.process.internal.RequestScope.runInScope(RequestScope.java:289)  at org.glassfish.jersey.server.ServerRuntime$AsyncResponder.resume(ServerRuntime.java:916)  at org.glassfish.jersey.server.ServerRuntime$AsyncResponder.resume(ServerRuntime.java:872)  at io.airlift.http.server.AsyncResponseHandler$1.onSuccess(AsyncResponseHandler.java:102)  at com.google.common.util.concurrent.Futures$CallbackListener.run(Futures.java:1347)  at io.airlift.concurrent.BoundedExecutor.drainQueue(BoundedExecutor.java:78)  at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)  at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)  at java.lang.Thread.run(Thread.java:748)

后面进一步排查，发现presto-cli中，查询的sql只要指定部分int/bigint类型的字段就会报错，对其他string类型的字段不报错。

排查思路

step 1:
hive-cli中执行show create table table_name,得到表结构，并在自己的测试环境创建同样的表test_table；

step 2:
从hive表对应的hdfs文件出导出部分文件，在hive-cli中执行load data local 语句，导入抽样数据到test_table中；

step 3:
hive-cli和presto-cli中分别执行查询语句看问题是否复现
（hive-cli中执行 select * from test_table limit 1，返回结果正常， presto-cli中执行报错，问题复现，同时发现只有查询表中的int/bigint类型字段会出问题）；
step 4:
step1的时候发现表内容的存储格式是orc ,所以执行 hive –orcfiledump /user/hive/warehouse/test_table/part-00000-4ae99c5d-9fec-47ab-a19c-c970f4ec40be-c000.snappy.orc 看orc文件的统计信息；

step4发现异常：table中定义为int,bigint类型的字段在步骤4的类型统计部分显示全部都是string类型，所以就怀疑是orc文件的文件部分字段类型和hive table中定义的不一致造成presto 在读取相关orc文件的时候报错。至于为什么hive-sql读取不报错，可我是hive对orc文件的处理和presto不一样吧。
抱着这样的猜想，决定去进行一次实验，去构建一个表结构定义和orc文件字段定义类型不一致的场景，实验过程如下：

错误复现实验

创建测试表test_string

tips:这里创建test_string表是为了将表中的orc文件拷贝到另外的test_int

#建表测试CREATE TABLE `test_string`(`server` string,`uid` string)ROW FORMAT SERDE    'org.apache.hadoop.hive.ql.io.orc.OrcSerde'STORED AS INPUTFORMAT    'org.apache.hadoop.hive.ql.io.orc.OrcInputFormat'OUTPUTFORMAT    'org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat'
/* 或者可以简写为如下    CREATE TABLE `test_string`(        `server` string,        `uid` string)        STORED AS orc*/

#hive-cli中执行show create table test_string输出如下
CREATE TABLE `test_string`(`server` string,`uid` string)ROW FORMAT SERDE'org.apache.hadoop.hive.ql.io.orc.OrcSerde'STORED AS INPUTFORMAT'org.apache.hadoop.hive.ql.io.orc.OrcInputFormat'OUTPUTFORMAT'org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat'LOCATION'hdfs://Ucluster/user/hive/warehouse/test_string'TBLPROPERTIES ('last_modified_by'='work','last_modified_time'='1560497239','transient_lastDdlTime'='1560772601')
#这里看到hive表的location在hdfs的/user/hive/warehouse/test_string目录

#向test_string表导入部分数据
insert into table test_string values ('server1', '1'),('server2', '2');

#退出hive-cli,查看test_string在hdfs上的orc文件信息概览
hadoop fs -ls /user/hive/warehouse/test_string (发现其下有一个000000_0,这是test_string的orc存储文件,输出如下）
Found 1 items-rwxrwxrwt   3 root supergroup        323 2022-11-17 21:14 /user/hive/warehouse/test_string/000000_0

hive –orcfiledump /user/hive/warehouse/test_string/000000_0 （执行这个命令查看orc文件的概览信息, 输出如下）：

Processing data file /user/hive/warehouse/test_string/000000_0 [length: 323]Structure for /user/hive/warehouse/test_string/000000_0File Version: 0.12 with ORC_135Rows: 2Compression: ZLIBCompression size: 262144Type: struct<server:string,uid:string> (这里可以看到orc文件每行有两个字段，server和uid，且都是string类型)
Stripe Statistics:  Stripe 1:    Column 0: count: 2 hasNull: false    Column 1: count: 2 hasNull: false min: server1 max: server2 sum: 14    Column 2: count: 2 hasNull: false min: 1 max: 2 sum: 2
File Statistics:  Column 0: count: 2 hasNull: false  Column 1: count: 2 hasNull: false min: server1 max: server2 sum: 14  Column 2: count: 2 hasNull: false min: 1 max: 2 sum: 2
Stripes:  Stripe: offset: 3 data: 32 rows: 2 tail: 51 index: 72    Stream: column 0 section ROW_INDEX start: 3 length 11    Stream: column 1 section ROW_INDEX start: 14 length 34    Stream: column 2 section ROW_INDEX start: 48 length 27    Stream: column 1 section DATA start: 75 length 15    Stream: column 1 section LENGTH start: 90 length 6    Stream: column 2 section DATA start: 96 length 5    Stream: column 2 section LENGTH start: 101 length 6    Encoding column 0: DIRECT    Encoding column 1: DIRECT_V2    Encoding column 2: DIRECT_V2
File length: 323 bytesPadding length: 0 bytesPadding ratio: 0%

创建测试表test_int

# create table (注意，这里创建表的uid字段是int类型)CREATE TABLE `test_int`(`server` string,`uid` int) STORED AS orc

此时，hive-cli中执行show create table test_int,输出内容内容如下：

CREATE TABLE `test_int`(`server` string,`uid` bigint)ROW FORMAT SERDE'org.apache.hadoop.hive.ql.io.orc.OrcSerde'STORED AS INPUTFORMAT'org.apache.hadoop.hive.ql.io.orc.OrcInputFormat'OUTPUTFORMAT'org.apache.hadoop.hive.ql.io.orc.OrcOutputFormat'LOCATION'hdfs://Ucluster/user/hive/warehouse/test_int'

试着将test_string表的hdfs orc文件移动到test_int表的location下（这里是hdfs://Ucluster/user/hive/warehouse/test_int）

load data inpath 'hdfs://Ucluster/user/hive/warehouse/test_string/000000_0' overwrite into table test_int;

此时退出hive-cli ，查看test_int表的orc文件概览信息

hive –orcfiledump /user/hive/warehouse/test_int/000000_0

Processing data file /user/hive/warehouse/test_int/000000_0 [length: 323]Structure for /user/hive/warehouse/test_int/000000_0File Version: 0.12 with ORC_135Rows: 2Compression: ZLIBCompression size: 262144Type: struct<server:string,uid:string>
Stripe Statistics:Stripe 1:    Column 0: count: 2 hasNull: false    Column 1: count: 2 hasNull: false min: server1 max: server2 sum: 14    Column 2: count: 2 hasNull: false min: 1 max: 2 sum: 2
File Statistics:Column 0: count: 2 hasNull: falseColumn 1: count: 2 hasNull: false min: server1 max: server2 sum: 14Column 2: count: 2 hasNull: false min: 1 max: 2 sum: 2
Stripes:Stripe: offset: 3 data: 32 rows: 2 tail: 51 index: 72    Stream: column 0 section ROW_INDEX start: 3 length 11    Stream: column 1 section ROW_INDEX start: 14 length 34    Stream: column 2 section ROW_INDEX start: 48 length 27    Stream: column 1 section DATA start: 75 length 15    Stream: column 1 section LENGTH start: 90 length 6    Stream: column 2 section DATA start: 96 length 5    Stream: column 2 section LENGTH start: 101 length 6    Encoding column 0: DIRECT    Encoding column 1: DIRECT_V2    Encoding column 2: DIRECT_V2
File length: 323 bytesPadding length: 0 bytesPadding ratio: 0%

hive/presto查询结果对比

经过上面两个建表的步骤，我们就得到了一个符合预期的test_int表（表的定义中uid为int类型，表的hdfs orc文件中的uid的类型为string。

# hive测试如下hive> select * from test_int;OKserver1  1server2  2Time taken: 4.186 seconds, Fetched: 2 row(s)
hive> select uid  from test_int;OK12Time taken: 0.178 seconds, Fetched: 2 row(s)

# presto测试如下这里的10.90.50.169:28080为presto coordinator的主机名和监听端口，database使用默认的default库， 从结果发现，查询test_int表的string 类型字段是ok的，查询int类型的uid字段会报错。./presto-cli –debug –server 10.90.50.169:28080 –catalog hive –schema default 
# presto 查询string 类型的字段okpresto:default> select server  from test_int; server--------- server1 server2(2 rows

测试重点来了！！！

# 查询int类型的字段报错，复现之前的问题。presto:default> select uid from test_int;java.lang.RuntimeException: Error fetching next at http://10.90.50.169:28080/v1/statement/20221117_133934_00034_yetvx/2 returned an invalid response: JsonResponse{statusCode=500, statusMessage=Internal Server Error, headers={content-length=[5070], content-type=[text/plain], date=[Mon, 17 Jun 2019 13:39:34 GMT]}, hasValue=false} [Error: com.fasterxml.jackson.databind.JsonMappingException: end index (8) must not be greater than size (1) (through reference chain: com.facebook.presto.client.QueryResults["data"])  at com.fasterxml.jackson.databind.JsonMappingException.wrapWithPath(JsonMappingException.java:379)  at com.fasterxml.jackson.databind.JsonMappingException.wrapWithPath(JsonMappingException.java:339)  at com.fasterxml.jackson.databind.ser.std.StdSerializer.wrapAndThrow(StdSerializer.java:343)  at com.fasterxml.jackson.databind.ser.std.BeanSerializerBase.serializeFields(BeanSerializerBase.java:698)  at com.fasterxml.jackson.databind.ser.BeanSerializer.serialize(BeanSerializer.java:155)  at com.fasterxml.jackson.databind.ser.DefaultSerializerProvider.serializeValue(DefaultSerializerProvider.java:292)  at com.fasterxml.jackson.databind.ObjectWriter$Prefetch.serialize(ObjectWriter.java:1419)  at com.fasterxml.jackson.databind.ObjectWriter.writeValue(ObjectWriter.java:940)  at io.airlift.jaxrs.JsonMapper.writeTo(JsonMapper.java:233)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor$TerminalWriterInterceptor.invokeWriteTo(WriterInterceptorExecutor.java:266)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor$TerminalWriterInterceptor.aroundWriteTo(WriterInterceptorExecutor.java:251)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor.proceed(WriterInterceptorExecutor.java:163)  at org.glassfish.jersey.server.internal.JsonWithPaddingInterceptor.aroundWriteTo(JsonWithPaddingInterceptor.java:109)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor.proceed(WriterInterceptorExecutor.java:163)  at org.glassfish.jersey.server.internal.MappableExceptionWrapperInterceptor.aroundWriteTo(MappableExceptionWrapperInterceptor.java:85)  at org.glassfish.jersey.message.internal.WriterInterceptorExecutor.proceed(WriterInterceptorExecutor.java:163)  at org.glassfish.jersey.message.internal.MessageBodyFactory.writeTo(MessageBodyFactory.java:1135)  at org.glassfish.jersey.server.ServerRuntime$Responder.writeResponse(ServerRuntime.java:662)  at org.glassfish.jersey.server.ServerRuntime$Responder.processResponse(ServerRuntime.java:395)  at org.glassfish.jersey.server.ServerRuntime$Responder.process(ServerRuntime.java:385)  at org.glassfish.jersey.server.ServerRuntime$AsyncResponder$3.run(ServerRuntime.java:884)  at org.glassfish.jersey.internal.Errors$1.call(Errors.java:272)  at org.glassfish.jersey.internal.Errors$1.call(Errors.java:268)  at org.glassfish.jersey.internal.Errors.process(Errors.java:316)  at org.glassfish.jersey.internal.Errors.process(Errors.java:298)  at org.glassfish.jersey.internal.Errors.process(Errors.java:268)  at org.glassfish.jersey.process.internal.RequestScope.runInScope(RequestScope.java:289)  at org.glassfish.jersey.server.ServerRuntime$AsyncResponder.resume(ServerRuntime.java:916)  at org.glassfish.jersey.server.ServerRuntime$AsyncResponder.resume(ServerRuntime.java:872)  at io.airlift.http.server.AsyncResponseHandler$1.onSuccess(AsyncResponseHandler.java:102)  at com.google.common.util.concurrent.Futures$CallbackListener.run(Futures.java:1347)  at io.airlift.concurrent.BoundedExecutor.drainQueue(BoundedExecutor.java:78)  at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)  at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)  at java.lang.Thread.run(Thread.java:748)Caused by: java.lang.IndexOutOfBoundsException: end index (8) must not be greater than size (1)  at io.airlift.slice.Preconditions.checkPositionIndexes(Preconditions.java:81)  at io.airlift.slice.Slice.checkIndexLength(Slice.java:1267)  at io.airlift.slice.Slice.getLong(Slice.java:479)  at com.facebook.presto.spi.block.AbstractVariableWidthBlock.getLong(AbstractVariableWidthBlock.java:63)  at com.facebook.presto.spi.type.BigintType.getObjectValue(BigintType.java:38)  at com.facebook.presto.server.protocol.RowIterable$RowIterator.computeNext(RowIterable.java:77)  at com.facebook.presto.server.protocol.RowIterable$RowIterator.computeNext(RowIterable.java:50)  at com.google.common.collect.AbstractIterator.tryToComputeNext(AbstractIterator.java:141)  at com.google.common.collect.AbstractIterator.hasNext(AbstractIterator.java:136)  at com.google.common.collect.Iterators$ConcatenatedIterator.hasNext(Iterators.java:1331)  at com.google.common.collect.Iterators$1.hasNext(Iterators.java:136)  at com.fasterxml.jackson.databind.ser.std.IterableSerializer.serializeContents(IterableSerializer.java:83)  at com.fasterxml.jackson.databind.ser.std.IterableSerializer.serialize(IterableSerializer.java:74)  at com.fasterxml.jackson.databind.ser.std.IterableSerializer.serialize(IterableSerializer.java:12)  at com.fasterxml.jackson.databind.ser.std.StdDelegatingSerializer.serialize(StdDelegatingSerializer.java:170)  at com.fasterxml.jackson.databind.ser.BeanPropertyWriter.serializeAsField(BeanPropertyWriter.java:693)  at com.fasterxml.jackson.databind.ser.std.BeanSerializerBase.serializeFields(BeanSerializerBase.java:690)  ... 31 more]  at com.facebook.presto.client.StatementClientV1.requestFailedException(StatementClientV1.java:436)  at com.facebook.presto.client.StatementClientV1.advance(StatementClientV1.java:383)  at com.facebook.presto.cli.StatusPrinter.printInitialStatusUpdates(StatusPrinter.java:123)  at com.facebook.presto.cli.Query.renderQueryOutput(Query.java:137)  at com.facebook.presto.cli.Query.renderOutput(Query.java:121)  at com.facebook.presto.cli.Console.process(Console.java:312)  at com.facebook.presto.cli.Console.runConsole(Console.java:256)  at com.facebook.presto.cli.Console.run(Console.java:146)  at com.facebook.presto.cli.Presto.main(Presto.java:31)Query is gone (server restarted?)

测试总结

如果想让hive以orc格式存储的表能正常在presto中查询，最好保证orc文件的字段类型定义和create table时的定义一致，否则，有可能会造成hive-cli中查询正常，在presto-cli 中查询异常的情况。特别是在从别处拷贝orc文件过来的时候，需要先做一下类型比对。
ps: 为何hive-cli中查询可以兼容字段类型不一致，而presto中确不行，就需要看presto的源码了。之前从网上看文档，presto在读取hdfs orc文件是进行了一些优化的，有了解细节的大大可以告诉我一声哈。另，测试是发现，test_int表在hive中查询uid字段，如果有非法字段，会显示为null。