> On May 10, 2018, noon, Sahil Takiar wrote:
> > itests/util/src/main/java/org/apache/hadoop/hive/cli/control/CliConfigs.java
> > Lines 674 (patched)
> > <https://reviews.apache.org/r/67023/diff/1/?file=2018287#file2018287line674>
> >
> >     is this necessary since you set the cluster type to mr above?
> 
> Andrew Sherman wrote:
>     Ha good question. Yes it is necessary as setClusterType() does not always 
> set the cluster type :-( - it allows the cluster type to overridden with 
> -Dclustermode=xxx
> 
> Sahil Takiar wrote:
>     interesting, should we handle other cluster types like Spark or MR too?

looks like it does already


> On May 10, 2018, noon, Sahil Takiar wrote:
> > ql/src/java/org/apache/hadoop/hive/ql/session/SessionState.java
> > Lines 545 (patched)
> > <https://reviews.apache.org/r/67023/diff/1/?file=2018292#file2018292line545>
> >
> >     whats the cache for? can q-tests even specify custom URIs? whats the 
> > use case for support multiple fs URIs?
> 
> Andrew Sherman wrote:
>     OK I admit I copied this code from the way hdfsEncryptionShims works 
> without fully understanding it.
> 
> Sahil Takiar wrote:
>     can we delete it then? i didn't realize this would require modifying code 
> outside of itests, so I think we should make any changes to core Hive as 
> minimal as possible

deleted


> On May 10, 2018, noon, Sahil Takiar wrote:
> > shims/common/src/main/java/org/apache/hadoop/hive/shims/HadoopShims.java
> > Lines 699 (patched)
> > <https://reviews.apache.org/r/67023/diff/1/?file=2018300#file2018300line699>
> >
> >     whats the use case for this dummy class? so we can run ec tests on 
> > hadoop versions that don't support ec? wouldn't be just disable the 
> > clidriver entirely for versions that don't support ec?
> 
> Andrew Sherman wrote:
>     I'm imagining wanting to run a test on bith EC and non-EC
> 
> Sahil Takiar wrote:
>     i thought the `NoopHdfsErasureCodingShim` was used when "the hadoop 
> version does not support hdfs Erasure Coding". u can still run on EC and 
> non-EC folders without this, right?

Yes that is the use of NoopHdfsErasureCodingShim. I think you always need 2 
implementations of the interface


> On May 10, 2018, noon, Sahil Takiar wrote:
> > shims/common/src/main/java/org/apache/hadoop/hive/shims/HadoopShims.java
> > Lines 741 (patched)
> > <https://reviews.apache.org/r/67023/diff/1/?file=2018300#file2018300line741>
> >
> >     since we have a cache anyway, would it make more sense to just remove 
> > this and make it a loading cache?
> 
> Andrew Sherman wrote:
>     I don't know, maybe. There is some advantage in keeping the same 
> structure as the HdfsEncryptionShim.
> 
> Sahil Takiar wrote:
>     see comment above about removing the cache

OK I have removed the cache and it does make the code simpler


- Andrew


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/67023/#review202831
-----------------------------------------------------------


On May 11, 2018, 11:38 p.m., Andrew Sherman wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/67023/
> -----------------------------------------------------------
> 
> (Updated May 11, 2018, 11:38 p.m.)
> 
> 
> Review request for hive and Sahil Takiar.
> 
> 
> Repository: hive-git
> 
> 
> Description
> -------
> 
> TestErasureCodingHDFSCliDriver uses a test-only CommandProcessor 
> "ErasureProcessor"
> which allows .q files to contain Erasure Coding commands similar to those 
> provided
> by the hdfs ec command
> https://hadoop.apache.org/docs/current/hadoop-project-dist/hadoop-hdfs/HDFSErasureCoding.html.
> The Erasure Coding functionality is exposed through a new shim 
> "HdfsFileErasureCodingPolicy".
> At this stage there are two .q files:
> erasure_commnds.q (a simple test to show ERASURE commands can run on local fs 
> via
> TestCliDriver or on hdfs via TestErasureCodingHDFSCliDriver), and
> erasure_simple.q (which does some trivial queries to demonstrate basic 
> functionality).
> More tests will come in future commits.
> 
> 
> Diffs
> -----
> 
>   
> itests/qtest/src/test/java/org/apache/hadoop/hive/cli/TestErasureCodingHDFSCliDriver.java
>  PRE-CREATION 
>   itests/src/test/resources/testconfiguration.properties 
> cf6d19a5937c3f4a82e4ffe09201af8a79da2e3d 
>   
> itests/util/src/main/java/org/apache/hadoop/hive/cli/control/CliConfigs.java 
> 6628336807b06cab49063673be0d8e9c5b5a7101 
>   itests/util/src/main/java/org/apache/hadoop/hive/ql/QTestUtil.java 
> 750fc69c5f210ca8f7bfe81b82ee9e3001fc07ba 
>   
> ql/src/java/org/apache/hadoop/hive/ql/processors/CommandProcessorFactory.java 
> 3d47991b603c94c8da2106e67192c8513ef783a7 
>   ql/src/java/org/apache/hadoop/hive/ql/processors/ErasureProcessor.java 
> PRE-CREATION 
>   ql/src/java/org/apache/hadoop/hive/ql/processors/HiveCommand.java 
> 56c7516ecfaf2421b0f3d3a188d05f38715b25b2 
>   ql/src/java/org/apache/hadoop/hive/ql/session/SessionState.java 
> 9f65a771f95a7c0bd3fdb4e56e47c0fc70235850 
>   
> ql/src/test/org/apache/hadoop/hive/ql/processors/TestCommandProcessorFactory.java
>  de43c2866f64e2ed5c74eab450de28f1a79248dc 
>   ql/src/test/queries/clientpositive/erasure_commands.q PRE-CREATION 
>   ql/src/test/queries/clientpositive/erasure_simple.q PRE-CREATION 
>   ql/src/test/results/clientpositive/erasure_commands.q.out PRE-CREATION 
>   ql/src/test/results/clientpositive/erasurecoding/erasure_commands.q.out 
> PRE-CREATION 
>   ql/src/test/results/clientpositive/erasurecoding/erasure_simple.q.out 
> PRE-CREATION 
>   shims/0.23/src/main/java/org/apache/hadoop/hive/shims/Hadoop23Shims.java 
> ec06a88dc21346473bec6589c703167d50e3b367 
>   shims/common/src/main/java/org/apache/hadoop/hive/shims/HadoopShims.java 
> b89081761780bf1f305d0196bb94bb0b54f7184f 
>   testutils/ptest2/conf/deployed/master-mr2.properties 
> 7edc307f85744d60d322ad8087164625677fc230 
> 
> 
> Diff: https://reviews.apache.org/r/67023/diff/2/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Andrew Sherman
> 
>

Reply via email to