You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@pig.apache.org by lulynn_2008 <lu...@163.com> on 2014/07/07 05:56:20 UTC
Please help review patch for PIG-4047: Replace pig-withouthadoop
jar with pig-core jar and pig core dependencies
Hi Daniel,
PIG-4047 is created for tracing this discussion and patch is available for review. Could you help to review it?
The patch file is based on trunk branch. I assume site is meant to trunk branch. Please correct me if I was wrong.
Thanks
At 2014-07-02 08:33:21, "Daniel Dai" <da...@hortonworks.com> wrote:
>I see what's happening. JarManager finds the enclosing jar from
>classpath and wrap those into job.jar. Originally I want to ship
>dependent jars separately through distributed cache, so we don't have
>to create job.jar every time, and those jars will get reused due to
>PIG-2672, and for Pig on tez, there are more chance we can get
>session/container reuse since we don't rely on job.jar which differs
>job by job. This will solve the pig-withouthadoop.jar eventually. But
>now I realize that pig-withouthadoop.jar issue can be solved
>independently by just putting dependent jars on lib and use
>pig-core.jar instead of pig-withouthadoop.jar. That's doable, and even
>if I go for distributed cache, I still need to put pig-core.jar in the
>place of pig-withouthadoop.jar and dependent jars in lib, there is no
>conflict. So go ahead do that.
>
>Thanks,
>Daniel
>
>On Mon, Jun 30, 2014 at 11:34 PM, lulynn_2008 <lu...@163.com> wrote:
>> Hi Deniel,
>> with new struntured pig package, the scripts run succeeded with pig-0.12.0.
>> I just did following:
>> 1. devided without hadoop jar into pig core and pig core dependencies.
>> 2. save jars in 1# in lib directory
>> 3. in pig script, always add all the jars in lib directory into classpath and add pig core jar into classpath.
>>
>> Here is the test case and result:
>> [pig@hostname bin]$ ./pig
>> grunt> a = load 'studenttab10k' as (name:chararray, age:int, gpa:double);
>> grunt> b = filter a by name matches '.*or.*';
>> grunt> dump b;
>> INFO [JobControl] org.apache.hadoop.mapreduce.lib.input.FileInputFormat - Total input paths to process : 1
>> INFO [main] org.apache.hadoop.mapreduce.lib.input.FileInputFormat - Total input paths to process : 1
>> (.hor.h,11,8.0)
>> grunt> dump a;
>> INFO [JobControl] org.apache.hadoop.mapreduce.lib.input.FileInputFormat - Total input paths to process : 1
>> INFO [main] org.apache.hadoop.mapreduce.lib.input.FileInputFormat - Total input paths to process : 1
>> (lynn,18,4.5)
>> (lulu,12,4.4)
>> (fat,12,5.0)
>> (.hor.h,11,8.0)
>> Please forgive me if I was wrong. I think they should be the same to backend, if we add without hadoop into classpath or add pig core and core dependencies one by one. Pig should be OK once we added all the dependencies into classpath. Not matter in which way.
>>
>> Do you have any suggestion on other test or comment?
>>
>> Thanks
>>
>>
>>
>>
>>
>>
>>
>> At 2014-06-25 11:51:24, "Daniel Dai" <da...@hortonworks.com> wrote:
>>>Pig will not fail every time, but you will hit it depending on your
>>>script. Try the following script:
>>>
>>>a = load 'studenttab10k' as (name:chararray, age:int, gpa:double);
>>>b = filter a by name matches '.*or.*';
>>>dump b;
>>>
>>>Thanks,
>>>Daniel
>>>
>>>On Tue, Jun 24, 2014 at 7:23 PM, lulynn_2008 <lu...@163.com> wrote:
>>>> Hi Deniel,
>>>> Thanks for your detail.
>>>>
>>>> For your following reply:
>>>> " Yes, that's a better solution. However, those jars are also needed in the
>>>> backend and we need to ship those jars into distributed cache. There will be
>>>> some code change to make it happen. Just change pig script is not
>>>> sufficient."
>>>> To distributed cache part, could you provide an example? To show how thoese
>>>> jars are shipped. Thanks
>>>>
>>>> Since release 0.10.0, I am using pig with the structure I mentioned at 2#.
>>>> That is
>>>> - deviding withouthadoop jar into pig core and dependencies.
>>>> - case all the dependencies into lib directory
>>>> - add pig core and these dependencies into classpath in pig script instead
>>>> of withouthadoop jar.
>>>> I also used pig-0.11.1, 0.12.0, and alwayse running pig with pig grunt. Did
>>>> not encounter issues/errors till now. I run avro/hbase/hive/hadoop related
>>>> things. Did I miss anything to test this part? Please give your suggestions.
>>>> Thanks
>>>>
>>>>
>>>>
>>>>
>>>> At 2014-06-25 02:34:03, "Daniel Dai" <da...@hortonworks.com> wrote:
>>>>>On Mon, Jun 23, 2014 at 8:13 PM, lulynn_2008 <lu...@163.com> wrote:
>>>>>> Hi All,
>>>>>> In build.xml in branch-0.13,
>>>>>> 1. following jars are included into lib directory during packaging:
>>>>>> <copy todir="${tar.dist.dir}/lib">
>>>>>> <fileset dir="${ivy.lib.dir}" includes="jython-*.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="jruby-*.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="groovy-*.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="js-*.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="hbase-*.jar"
>>>>>> excludes="hbase-*tests.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}"
>>>>>> includes="protobuf-java-*.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="zookeeper-*.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="accumulo-*.jar"
>>>>>> excludes="accumulo-minicluster*.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="avro-*.jar"
>>>>>> excludes="avro-*tests.jar"/>
>>>>>> <fileset dir="${ivy.lib.dir}" includes="json-simple-*.jar"/>
>>>>>> </copy>
>>>>>> 2. following jars are included in without hadoop jar file:
>>>>>> <fileset dir="${ivy.lib.dir}"
>>>>>> id="runtime.dependencies-withouthadoop.jar">
>>>>>> <patternset id="pattern.runtime.dependencies-withouthadoop.jar">
>>>>>> <include name="antlr-runtime-${antlr.version}.jar"/>
>>>>>> <include name="ST4-${stringtemplate.version}.jar"/>
>>>>>> <include name="jline-${jline.version}.jar"/>
>>>>>> <include name="jackson-mapper-asl-${jackson.version}.jar"/>
>>>>>> <include name="jackson-core-asl-${jackson.version}.jar"/>
>>>>>> <include name="joda-time-${joda-time.version}.jar"/>
>>>>>> <include name="guava-${guava.version}.jar"/>
>>>>>> <include name="automaton-${automaton.version}.jar"/>
>>>>>> <include name="jansi-${jansi.version}.jar"/>
>>>>>> <include name="avro-${avro.version}.jar"/>
>>>>>> <include name="avro-mapred-${avro.version}.jar"/>
>>>>>> <include name="trevni-core-${avro.version}.jar"/>
>>>>>> <include name="trevni-avro-${avro.version}.jar"/>
>>>>>> <include name="snappy-java-${snappy.
>>>> version}.jar"/>
>>>>>> <include name="asm*.jar"/>
>>>>>> </patternset>
>>>>>> </fileset>
>>>>>> Questions:
>>>>>> 1. Could you tell what the jars in 1# and 2# are used for? What are the
>>>>>> differences between them?
>>>>>> 1. Seems all the jars in 1# and 2# will be necessary when pig running at
>>>>>> hadoop cluster. Correct?
>>>>>
>>>>>#2 are core dependency Pig depends on, those will be wrapped into
>>>>>pig-withouthadoop.jar. #1 are the convenience jar used only when user
>>>>>are using specific loader, such as HBaseStorage/AvroStorage, those are
>>>>>not needed if you don't use such loader, and will not be wrapped into
>>>>>pig-withouthadoop.jar
>>>>>
>>>>>> 2. without hadoop jar file will be invoked during pig running in hadoop
>>>>>> cluster, and this jar file includes pig core and some dependencies. Could we
>>>>>> just generate pig core jar and move all the dependencies into lib directory?
>>>>>> Then in pig script, we could alwayse add pig core into classpath add the
>>>>>> dependencies into classpath in "if [ -n "$HADOOP_BIN" ]; then" option. If
>>>>>> this:
>>>>>> - it would be clear for user to check dependencies version
>>>>>> - it would be easiler to mantain dependencies, Like version update.
>>>>>> Sometime, I would like to check whether pig could work with difference
>>>>>> version dependencies.
>>>>>
>>>>>Yes, that's a better solution. However, those jars are also needed in
>>>>>the backend and we need to ship those jars into distributed cache.
>>>>>There will be some code change to make it happen. Just change pig
>>>>>script is not sufficient.
>>>>>
>>>>>> Do you have any concern?
>>>>>>
>>>>>> Thanks
>>>>>>
>>>>>
>>>>>--
>>>>>CONFIDENTIALITY NOTICE
>>>>>NOTICE: This message is intended for the use of the individual or entity to
>>>>>which it is addressed and may contain information that is confidential,
>>>>>privileged and exempt from disclosure under applicable law. If the reader
>>>>>of this message is not the intended recipient, you are hereby notified that
>>>>>any printing, copying, dissemination, distribution, disclosure or
>>>>>forwarding of this communication is strictly prohibited. If you have
>>>>>received this communication in error, please contact the sender immediately
>>>>>and delete it from your system. Thank You.
>>>>
>>>>
>>>>
>>>
>>>--
>>>CONFIDENTIALITY NOTICE
>>>NOTICE: This message is intended for the use of the individual or entity to
>>>which it is addressed and may contain information that is confidential,
>>>privileged and exempt from disclosure under applicable law. If the reader
>>>of this message is not the intended recipient, you are hereby notified that
>>>any printing, copying, dissemination, distribution, disclosure or
>>>forwarding of this communication is strictly prohibited. If you have
>>>received this communication in error, please contact the sender immediately
>>>and delete it from your system. Thank You.
>
>--
>CONFIDENTIALITY NOTICE
>NOTICE: This message is intended for the use of the individual or entity to
>which it is addressed and may contain information that is confidential,
>privileged and exempt from disclosure under applicable law. If the reader
>of this message is not the intended recipient, you are hereby notified that
>any printing, copying, dissemination, distribution, disclosure or
>forwarding of this communication is strictly prohibited. If you have
>received this communication in error, please contact the sender immediately
>and delete it from your system. Thank You.