Question

我正在尝试使用Hive配置水槽以使用Hive Sink类型将水槽输出保存到蜂巢表。我有单节点集群。我使用mapr hadoop发布。

这是我的flume.conf

agent1.sources = source1
agent1.channels = channel1
agent1.sinks = sink1

agent1.sources.source1.type = exec
agent1.sources.source1.command = cat /home/andrey/flume_test.data

agent1.sinks.sink1.type = hive
agent1.sinks.sink1.channel = channel1
agent1.sinks.sink1.hive.metastore = thrift://127.0.0.1:9083
agent1.sinks.sink1.hive.database = default
agent1.sinks.sink1.hive.table = flume_test
agent1.sinks.sink1.useLocalTimeStamp = false
agent1.sinks.sink1.round = true
agent1.sinks.sink1.roundValue = 10
agent1.sinks.sink1.roundUnit = minute
agent1.sinks.sink1.serializer = DELIMITED
agent1.sinks.sink1.serializer.delimiter = "," 
agent1.sinks.sink1.serializer.serdeSeparator = ','
agent1.sinks.sink1.serializer.fieldnames = id,message

agent1.channels.channel1.type = FILE
agent1.channels.channel1.transactionCapacity = 1000000
agent1.channels.channel1.checkpointInterval 30000
agent1.channels.channel1.maxFileSize = 2146435071
agent1.channels.channel1.capacity 10000000
agent1.sources.source1.channels = channel1

我的数据flume_test.data

1,AAAAAAAA
2,BBBBBBB
3,CCCCCCCC
4,DDDDDD
5,EEEEEEE
6,FFFFFFFFFFF
7,GGGGGG
8,HHHHHHH
9,IIIIII
10,JJJJJJ
11,KKKKKK
12,LLLLLLLL
13,MMMMMMMMM
14,NNNNNNNNN
15,OOOOOOOO
16,PPPPPPPPPP
17,QQQQQQQ
18,RRRRRRR
19,SSSSSSSS

这就是我如何在Hive中创建我的表

create table flume_test(id string, message string)
clustered by (message) into 1 buckets
STORED AS ORC tblproperties ("orc.compress"="NONE");

当我只使用1个存储桶时，在hive shell中选择* from flume_test命令只返回OK状态，没有数据。如果我使用超过1个存储桶，则会返回错误消息。

错误，例如在hive表选择后使用5个桶：

hive> select * from flume_test;
OK
2015-06-18 10:04:57,6909 ERROR Client fs/client/fileclient/cc/client.cc:1385 Thread: 10141 Open failed for file /user/hive/warehouse/flume_test/delta_0004401_0004500/bucket_00, LookupFid error No such file or directory(2)
2015-06-18 10:04:57,6941 ERROR Client fs/client/fileclient/cc/client.cc:1385 Thread: 10141 Open failed for file /user/hive/warehouse/flume_test/delta_0004401_0004500/bucket_00, LookupFid error No such file or directory(2)
2015-06-18 10:04:57,6976 ERROR Client fs/client/fileclient/cc/client.cc:1385 Thread: 10141 Open failed for file /user/hive/warehouse/flume_test/delta_0004401_0004500/bucket_00, LookupFid error No such file or directory(2)
2015-06-18 10:04:57,7044 ERROR Client fs/client/fileclient/cc/client.cc:1385 Thread: 10141 Open failed for file /user/hive/warehouse/flume_test/delta_0004401_0004500/bucket_00, LookupFid error No such file or directory(2)
Time taken: 0.914 seconds

Hive表数据保存在/ user / hive / warehouse / flume_test目录中，并且不是空的。

-rwxr-xr-x   3 andrey andrey          4 2015-06-17 16:28 /user/hive/warehouse/flume_test/_orc_acid_version
drwxr-xr-x   - andrey andrey          2 2015-06-17 16:28 /user/hive/warehouse/flume_test/delta_0004301_0004400

delta目录包含

-rw-r--r--   3 andrey andrey        991 2015-06-17 16:28 /user/hive/warehouse/flume_test/delta_0004301_0004400/bucket_00000
-rwxr-xr-x   3 andrey andrey          8 2015-06-17 16:28 /user/hive/warehouse/flume_test/delta_0004301_0004400/bucket_00000_flush_length

即使是猪，我也无法阅读/ user / hive / warehouse / flume_test / delta_0004301_0004400 / bucket_00000 orc文件。

此外，我尝试在hive中创建表格后设置此变量，但这并没有给出结果。

set hive.txn.manager = org.apache.hadoop.hive.ql.lockmgr.DbTxnManager;
set hive.compactor.initiator.on = true;
set hive.compactor.worker.threads = 2;

我在互联网上找到了一些例子，但它们并不完整，我不熟悉，因此我无法理解它们。

Answer 1

将这两行添加到我的配置解决了我的问题，但是当从hive读取表时我仍然有错误。我可以读表，它返回正确的结果，但有错误

agent1.sinks.sink1.hive.txnsPerBatchAsk = 2
agent1.sinks.sink1.batchSize = 10

Answer 2

您似乎没有生成avsc文件。您似乎正在使用AVRO文件创建HIVE表，因此出现错误。

使用Hive Sink将水槽输出保存到蜂巢表

2 个答案: