使用Oozie添加Hive分区

Question

I'm using HPD-2.4.2 and try to add partitions to an external Hive table using an Oozie coordinator job. I created a coordinator that daily tiggers the following workflow:

<workflow-app name="addPartition" xmlns="uri:oozie:workflow:0.4">
  <start to="hive"/>
  <action name="hive">
   <hive2 xmlns="uri:oozie:hive2-action:0.1">
    <job-tracker>${jobTracker}</job-tracker>
    <name-node>${nameNode}</name-node>
    <jdbc-url>jdbc:hive2://${jdbcPath}</jdbc-url>
    <password>yarn</password>
    <script>${appPath}/addPartition.q</script>
    <param>nameNode=${nameNode}</param>
    <param>dt=${dt}</param>
    <param>path=${path}</param>
   </hive2>
   <ok to="end" />
   <error to="fail" />
  </action>
  <kill name="fail">
   <message>
    Workflow failed, error message[${wf:errorMessage(wf:lastErrorNode())}]
   </message>
  </kill>
  <end name="end" />
</workflow-app>

The executed script contains

CREATE EXTERNAL TABLE IF NOT EXISTS visits (sid BIGINT, os STRING, browser STRING, visit_time TIMESTAMP)
  PARTITIONED BY (dt STRING) 
  STORED AS PARQUET;

ALTER TABLE visits ADD PARTITION(dt = '${dt}') LOCATION   '${nameNode}/data/parquet/visitors/${path}'; 

If I run the job the table is created but no partition is added. In yarn log I find:

Beeline command arguments :
         -u
         jdbc:hive2://localhost:10000/default
         -n
         yarn
         -p
         yarn
         -d
         org.apache.hive.jdbc.HiveDriver
         --hivevar
         nameNode=hdfs://bigdata01.local:8020
         --hivevar
         dt=2016-01-05
         --hivevar
         path=2016/01/05
         -f
         addPartition.q
         -a
         delegationToken
         --hiveconf
         mapreduce.job.tags=oozie-1b3b2ee664df7ac9ee436379d784955a

Fetching child yarn jobs
tag id : oozie-1b3b2ee664df7ac9ee436379d784955a
Child yarn jobs are found - 
=================================================================

>>> Invoking Beeline command line now >>>
[...]
0: jdbc:hive2://localhost:10000/default> ALTER TABLE visits ADD    PARTITION(dt = '${dt}') LOCATION  '${nameNode}/data/parquet/visitors/${path}';

It looks as if the parameters in the ALTER TABLE are not replaced, to check this I tried to call beeline directly from the CLI:

beeline -u jdbc:hive2://localhost:10000/default -n yarn -p yarn -d org.apache.hive.jdbc.HiveDriver --hivevar nameNode=hdfs://bigdata01.local:8020 --hivevar dt="2016-01-03" --hivevar path="2016/01/03" -e "ALTER TABLE visits ADD PARTITION(dt='${dt}') LOCATION '${nameNode}/data/parquet/visitors/${path}';"

which results in an error:

Connecting to jdbc:hive2://localhost:10000/default
Connected to: Apache Hive (version 1.2.1000.2.4.2.0-258)
Driver: Hive JDBC (version 1.2.1000.2.4.2.0-258)
Transaction isolation: TRANSACTION_REPEATABLE_READ
Error: Error while processing statement: FAILED: Execution Error, return code 1 from org.apache.hadoop.hive.ql.exec.DDLTask. partition spec is invalid; field dt does not exist or is empty (state=08S01,code=1)

if I run the alter statement without parameters

beeline -u jdbc:hive2://localhost:10000/default -n yarn -p yarn -d org.apache.hive.jdbc.HiveDriver -e "ALTER TABLE visits ADD PARTITION(dt='2016-01-03') LOCATION 'hdfs://bigdata01.local:8020/data/parquet/visitors/2016/01/03';"

or open a beeline console with hivevars defined and execute the alter statement

beeline -u jdbc:hive2://localhost:10000/default -n yarn -p yarn -d org.apache.hive.jdbc.HiveDriver --hivevar nameNode=hdfs://bigdata01.local:8020 --hivevar dt="2016-01-03" --hivevar path="2016/01/03"
0: jdbc:hive2://localhost:10000/default>  ALTER TABLE visits ADD PARTITION(dt = '${dt}') LOCATION '${nameNode}/data/parquet/visitors/${path}';

the partition is created.

Where am I wrong?

Update:

The values for the parameters in the hive2 action are defined in the oozie.properties file and the coordinator.xml

<property>
  <name>nameNode</name>
  <value>${nameNode}</value>
</property> 
<property>
 <name>dt</name>                
 <value>${coord:formatTime(coord:dateOffset(coord:nominalTime(), -1,'DAY'),'yyyy-MM-dd')}</value>
</property>
<property>
  <name>path</name>                    
  <value>${coord:formatTime(coord:dateOffset(coord:nominalTime(), -1,'DAY'),'yyyy/MM/dd')}</value>
</property>

in yarn log you find

Parameters:
------------------------
nameNode=hdfs://bigdata01.local:8020
dt=2016-01-05
path=2016/01/05 

before they are set as hivevars in the beeline call from the hive2 action.

Answer 1

Thanks for your help but I give up. Instead of a hive2 action I will use a ssh action to execute beeline with a static alter statement.

<ssh xmlns="uri:oozie:ssh-action:0.1">
  <host>${sshUser}@${sshHost}</host>
  <command>"beeline"</command>
  <args>-u</args>
  <args>jdbc:hive2://localhost:10000/default</args>
  <args>-n</args>
  <args>yarn</args>
  <args>-p</args>
  <args>yarn</args>
  <args>-d</args>
  <args>org.apache.hive.jdbc.HiveDriver</args>
  <args>-e</args>
  <args>"ALTER TABLE visits ADD PARTITION(dt='${dt}') LOCATION '${nameNode}/data/raw/parquet/visitors/${path}';"</args>
  <capture-output />
</ssh>

Answer 2

Finally found the problem. You have to use double quotes instead of single quotes ;-)

$ beeline -u jdbc:hive2://localhost:10000/default -n yarn -p yarn -d org.apache.hive.jdbc.HiveDriver --hivevar foo=bar -e "SELECT '${foo}' as foo;     

+------+--+
| foo  |  
+------+--+
|      |
+------+--+  

beeline -u jdbc:hive2://localhost:10000/default -n yarn -p yarn -d org.apache.hive.jdbc.HiveDriver --hivevar foo=bar -e 'SELECT "${foo}" as foo;'  

+------+--+
| foo  |
+------+--+
| bar  |
+------+--+

beeline -u jdbc:hive2://localhost:10000/default -n yarn -p yarn -d org.apache.hive.jdbc.HiveDriver --hivevar foo=bar -f selectFoo.q

+------+--+
| foo  |
+------+--+
| bar  |
+------+--+