CDH5.4.5运行Phoenix导入CSV文件

1.安装phoenix

在界面上设置Phoenix的parcel包:

http://52.11.56.155:7180/cmf/settings?groupKey=config.scm.parcel.display_group&groupParent=

添加一个Remote Parcel Repository URLs url:http://archive.cloudera.com/cloudera-labs/phoenix/parcels/1.0/

CM会自动发现新的parcel，然后点击Download,Distribute and Active。重启集群

2.进入到某台服务器上，查看phoenix的安装路径

[root@ip---- ~]# cd /opt/cloudera/parcels/CLABS_PHOENIX
[root@ip---- phoenix]# ls
bin  dev  examples  lib  phoenix-4.3.-clabs-phoenix-1.0.-client.jar  phoenix-4.3.-clabs-phoenix-1.0.-server.jar  phoenix-4.3.-clabs-phoenix-1.0.-server-without-antlr.jar

bin目录下为可执行文件，examples目录下为一些样例

3.导入CSV格式的表

CSV文件为/root/ceb/cis_cust_imp_info.csv,内容如下：

,,BR01,2000.01
,,BR01,2000.02
,,BR02,2000.03

定义一个表结构的文件/root/ceb/cis_cust_imp_info.sql，内容如下，

CREATE TABLE IF NOT EXISTS cis_cust_imp_info(
statistics_dt varchar(50),
cust_id varchar(50),
open_org_id varchar(50),
assert9_bal decimal(18,2),
CONSTRAINT pk PRIMARY KEY (statistics_dt, cust_id)
); 

注意最后的分号是必须的。

运行命令，导入CSV

[root@ip---- phoenix]# bin/psql.py 172.31.25.244 /root/ceb/cis_cust_imp_info.sql /root/ceb/cis_cust_imp_info.csv
SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See http://www.slf4j.org/codes.html#StaticLoggerBinder for further details.
// :: WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable
// :: WARN impl.MetricsConfig: Cannot locate configuration: tried hadoop-metrics2-phoenix.properties,hadoop-metrics2.properties
no rows upserted
Time: 0.259 sec(s)

csv columns from database.
CSV Upsert complete.  rows upserted
Time: 0.067 sec(s)

在hbase shell中进行验证：

hbase(main)::> list
TABLE
CIS_CUST_IMP_INFO
SYSTEM.CATALOG
SYSTEM.SEQUENCE
SYSTEM.STATS
 row(s) in 0.2650 seconds

=> ["CIS_CUST_IMP_INFO", "SYSTEM.CATALOG", "SYSTEM.SEQUENCE", "SYSTEM.STATS"]
hbase(main)::> scan 'CIS_CUST_IMP_INFO'
ROW                                                COLUMN+CELL
 \x00100010001001                          column=:ASSERT9_BAL, timestamp=, value=\xC2\x15\x01\x02
 \x00100010001001                          column=:OPEN_ORG_ID, timestamp=, value=BR01
 \x00100010001001                          column=:_0, timestamp=, value=
 \x00100010001002                          column=:ASSERT9_BAL, timestamp=, value=\xC2\x15\x01\x03
 \x00100010001002                          column=:OPEN_ORG_ID, timestamp=, value=BR01
 \x00100010001002                          column=:_0, timestamp=, value=
 \x00100010001003                          column=:ASSERT9_BAL, timestamp=, value=\xC2\x15\x01\x04
 \x00100010001003                          column=:OPEN_ORG_ID, timestamp=, value=BR02
 \x00100010001003                          column=:_0, timestamp=, value=
 row(s) in 0.1840 seconds

4.以MR的方式导入大量CSV文件

[root@ip---- phoenix]# hadoop jar phoenix-4.3.-clabs-phoenix-1.0.-client.jar org.apache.phoenix.mapreduce.CsvBulkLoadTool --table cis_cust_imp_info --input /root/ceb/cis_cust_imp_info.csv --zookeeper 172.31.25.244

发生错误：

java.util.concurrent.ExecutionException: java.lang.IllegalAccessError: class com.google.protobuf.HBaseZeroCopyByteString cannot access its superclass com.google.protobuf.LiteralByteString
    at java.util.concurrent.FutureTask.report(FutureTask.java:)
    at java.util.concurrent.FutureTask.get(FutureTask.java:)
    at org.apache.hadoop.hbase.client.HTable.coprocessorService(HTable.java:)
    at org.apache.hadoop.hbase.client.HTable.coprocessorService(HTable.java:)
    at org.apache.phoenix.query.ConnectionQueryServicesImpl.metaDataCoprocessorExec(ConnectionQueryServicesImpl.java:)
    at org.apache.phoenix.query.ConnectionQueryServicesImpl.getTable(ConnectionQueryServicesImpl.java:)
    at org.apache.phoenix.schema.MetaDataClient.updateCache(MetaDataClient.java:)
    at org.apache.phoenix.schema.MetaDataClient.updateCache(MetaDataClient.java:)
    at org.apache.phoenix.schema.MetaDataClient.getCurrentTime(MetaDataClient.java:)
    at org.apache.phoenix.compile.StatementContext.getCurrentTime(StatementContext.java:)
    at org.apache.phoenix.execute.BaseQueryPlan.iterator(BaseQueryPlan.java:)
    at org.apache.phoenix.execute.BaseQueryPlan.iterator(BaseQueryPlan.java:)
    at org.apache.phoenix.jdbc.PhoenixStatement$.call(PhoenixStatement.java:)
    at org.apache.phoenix.jdbc.PhoenixStatement$.call(PhoenixStatement.java:)
    at org.apache.phoenix.call.CallRunner.run(CallRunner.java:)
    at org.apache.phoenix.jdbc.PhoenixStatement.executeQuery(PhoenixStatement.java:)
    at org.apache.phoenix.jdbc.PhoenixStatement.executeQuery(PhoenixStatement.java:)
    at org.apache.phoenix.jdbc.PhoenixDatabaseMetaData.getColumns(PhoenixDatabaseMetaData.java:)
    at org.apache.phoenix.util.CSVCommonsLoader.generateColumnInfo(CSVCommonsLoader.java:)
    at org.apache.phoenix.mapreduce.CsvBulkLoadTool.buildImportColumns(CsvBulkLoadTool.java:)
    at org.apache.phoenix.mapreduce.CsvBulkLoadTool.loadData(CsvBulkLoadTool.java:)
    at org.apache.phoenix.mapreduce.CsvBulkLoadTool.run(CsvBulkLoadTool.java:)
    at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:)
    at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:)
    at org.apache.phoenix.mapreduce.CsvBulkLoadTool.main(CsvBulkLoadTool.java:)
    at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
    at sun.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:)
    at java.lang.reflect.Method.invoke(Method.java:)
    at org.apache.hadoop.util.RunJar.run(RunJar.java:)
    at org.apache.hadoop.util.RunJar.main(RunJar.java:)
Caused by: java.lang.IllegalAccessError: class com.google.protobuf.HBaseZeroCopyByteString cannot access its superclass com.google.protobuf.LiteralByteString

网上搜索，发现是由于HBASE的一个bug，解决方法是:

[root@ip---- phoenix]# cd /opt/cloudera/parcels/CDH/lib/hadoop
ln -s /opt/cloudera/parcels/CDH-5.4.-.cdh5.4.5.p0./lib/hbase/lib/hbase-protocol-1.0.-cdh5.4.5.jar hbase-protocol-1.0.-cdh5.4.5.jar

重新运行导入命令，发现如下错误:

// :: WARN security.UserGroupInformation: PriviledgedActionException as:root (auth:SIMPLE) cause:org.apache.hadoop.security.AccessControlException: Permission denied: user=root, access=WRITE, inode="/user":hdfs:supergroup:drwxr-xr-x
    at org.apache.hadoop.hdfs.server.namenode.DefaultAuthorizationProvider.checkFsPermission(DefaultAuthorizationProvider.java:)
    at org.apache.hadoop.hdfs.server.namenode.DefaultAuthorizationProvider.check(DefaultAuthorizationProvider.java:)

原因是/user目录权限问题，用hdfs用户重新跑一遍，发生错误。用chmod 修改/user为777

sudo -u hdfs hdfs dfs -chmod  /user

sudo -u hdfs hadoop jar phoenix-4.3.-clabs-phoenix-1.0.-client.jar org.apache.phoenix.mapreduce.CsvBulkLoadTool --table cis_cust_imp_info --input /root/ceb/cis_cust_imp_info.csv --zookeeper 172.31.25.244

15/09/04 11:06:05 ERROR mapreduce.CsvBulkLoadTool: Import job on table=CIS_CUST_IMP_INFO failed due to exception:org.apache.hadoop.mapreduce.lib.input.InvalidInputException: Input path does not exist: hdfs://ip-172-31-25-243.us-west-2.compute.internal:8020/root/ceb/cis_cust_imp_info.csv
15/09/04 11:06:05 INFO client.ConnectionManager$HConnectionImplementation: Closing zookeeper sessionid=0x14f97b7df1400a4

原来用MR模式跑，文件需要放到HDFS上

这时MR运行可以顺利完成，HFile顺利生产，但是在loadIncremental环境卡住了。原因在于load到hbase中的表属于hbase:hbase，但生产的HFile文件属于当前用户和组。所以以hbase用户运行

sudo -u hbase hadoop jar phoenix-4.3.-clabs-phoenix-1.0.-client.jar org.apache.phoenix.mapreduce.CsvBulkLoadTool --table cis_cust_imp_info --input /root/ceb/cis_cust_imp_info.csv --zookeeper 172.31.25.244

顺利搞定！