Commit Graph

236 Commits

Author SHA1 Message Date
Nong Li
58631d9ce0 Fix parquet insert .test files. 2014-01-08 10:49:46 -08:00
Alex Behm
2277386d4d IMPALA-225: Compound predicate ranges on partition keys crash impalad. 2014-01-08 10:49:45 -08:00
Nong Li
563cbfa3a8 Enable parquet testing 2014-01-08 10:49:40 -08:00
Lenni Kuff
3e70594484 Fix unstable top-n test case 2014-01-08 10:49:37 -08:00
Lenni Kuff
cba9cd00dd Fix full data load build break due to constructing incorrect HDFS paths 2014-01-08 10:49:34 -08:00
Marcel Kornacker
398e725a23 make broadcast joins the default join strategy 2014-01-08 10:49:34 -08:00
Alex Behm
1b2e8280d4 Fix NULL issues. 2014-01-08 10:49:32 -08:00
Lenni Kuff
e218721386 IMPALA-198: Support setting file format, table comment in CREATE TABLE LIKE statements 2014-01-08 10:49:31 -08:00
Marcel Kornacker
d7e22f44bb Partitioned hash joins
- added PlanNode.numNodes, PlanNode.avgRowSize and PlanNode.computeStats()
- fixing up some cardinality estimates
- Planner now tries to do a cost-based decision between broadcast join and join with full repartitioning (both inputs)
- ExchangeNode now distinguishes between its input and output row descriptor: the output potentially contains more tuples
- fixed problem related to cancellation and concurrent hash table builds.

Not included:
- partitioned joins that take advantage of existing partitions of the inputs; those will have to wait for a follow-on change
2014-01-08 10:49:29 -08:00
Lenni Kuff
558d5ce755 Data loading: Exec DDL statements via Impala and don't recreate metadata if it exists 2014-01-08 10:49:28 -08:00
Nong Li
1fcfb72bc4 IMPALA-145: Fix order by limit 0 crash. 2014-01-08 10:49:27 -08:00
Nong Li
ebab23841a Add back pressure in hdfs-scan-node to prevent excessive buffer queueing. 2014-01-08 10:49:25 -08:00
Skye Wanderman-Milne
a7e15b1417 Update Parquet scanner to only scan a file if assigned the first split.
Also re-enable Parquet tests.
2014-01-08 10:49:25 -08:00
Alex Behm
673d7b97cf IMPALA-190: Insert with NULL partition keys results in SIGSEGV. 2014-01-08 10:49:22 -08:00
Lenni Kuff
15f0313283 Add analysis checks for length of RowFormat strings, fix escaping of row format values 2014-01-08 10:49:21 -08:00
Lenni Kuff
018a72bfe2 IMPALA-189: Properly support NULL partition key values in ALTER .. PARTITION statements 2014-01-08 10:49:21 -08:00
Alan Choi
4a503a4e35 IMP-808 construct runtime state in fe-support to eval now() 2014-01-08 10:49:20 -08:00
Nong Li
329763e5ab Disable parquet tests. 2014-01-08 10:49:20 -08:00
Lenni Kuff
36e9fe1c1a Run compute table stats statements using Hive CLI
This works around a problem with computing table stats via the Hive Meta Store client
API. When executing these stements via the MetaStoreClient, all tables were getting a
num_rows=0 value returned from the ANALYZE TABLE query.
2014-01-08 10:49:19 -08:00
Nong Li
20fc700002 Fix precision issue in text table writer. 2014-01-08 10:49:19 -08:00
Alex Behm
0821e2f826 IMPALA-66: Support for UNION with constant SELECT clauses. 2014-01-08 10:49:18 -08:00
Lenni Kuff
5a0b1270c4 Add support for ALTER ... PARTITION (partitionSpec) SET FILEFORMAT/LOCATION
Adds support for:
* ALTER TABLE <table> PARTITION (partitionSpec) SET FILEFORMAT
* ALTER TABLE <table> PARTITION (partitionSpec) SET LOCATION

This enables setting the location and fileformat of specific partitions.
2014-01-08 10:49:17 -08:00
Lenni Kuff
f4a5c0628f Cleanup HDFS directories before and after running ALTER TABLE tests 2014-01-08 10:49:17 -08:00
Skye Wanderman-Milne
8ef36831f6 Update DataErrorsTest to reflect LZO_MAX_BLOCK_SIZE rename 2014-01-08 10:49:16 -08:00
Lenni Kuff
1fb72fbc73 IMPALA-156: Support core 'ALTER TABLE' DDL command
This patch adds support for
- ALTER TABLE ADD|REPLACE COLUMNS
- ALTER TABLE DROP COLUMN
- ALTER TABLE ADD/DROP PARTITION
- ALTER TABLE SET FILEFORMAT
- ALTER TABLE SET LOCATION
- ALTER TABLE RENAME
2014-01-08 10:49:14 -08:00
Elliott Clark
0e0c02b6bd Add the ability to Select into HBase table.
* Changed frontend analysis for HBase tables
* Changed Thrift messages to allow HBase as a sink type.
* JNI Wrapper around htable
* Create hbase-table-sink
* Create hbase-table-writer
* Static init lots of JNI related code for HBase.
* Cleaned up some cpplint issues.
* Changed junit analysis tests
* Create a new HBase test table.
* Added functional tests for HBase inserts.
2014-01-08 10:49:06 -08:00
Lenni Kuff
993da8fcba Fix bug in how insert tables are generated 2014-01-08 10:49:05 -08:00
ishaan
15658f384b Include targeted performance tests in experiments and add a new query 2014-01-08 10:49:02 -08:00
Lenni Kuff
5f81becd84 Create tables used by insert tests in a supported insert format 2014-01-08 10:49:00 -08:00
Marcel Kornacker
0c36c7f327 Partitioned merge aggregation. 2014-01-08 10:48:59 -08:00
Alan Choi
57c2f828e0 IMP-791 Fix full outer join hang
In full or right outer join, the hash-join-node does not release
the io buffer when calling get next, causing deadlock.
2014-01-08 10:48:58 -08:00
Lenni Kuff
831ee529be Fixed data loading bugs, moved most tables out of load-dependent-tables 2014-01-08 10:48:56 -08:00
Lenni Kuff
7584312540 IMPALA-167: Impala should gracefully handle unsupported Hive table types 2014-01-08 10:48:56 -08:00
Lenni Kuff
ca0d23a844 IMPALA-157: Support CREATE TABLE LIKE DDL 2014-01-08 10:48:55 -08:00
Henry Robinson
8d87972695 Improve parser coverage
This patch adds support for the following SQL constructs

  - Unary + operator
  - The ALL keyword, in SELECT ALL and SELECT aggregate_func(ALL *)
  - REAL and INTEGER as type synonyms for DOUBLE and INT respectively
  - The AS keyword after a table spec. e.g. SELECT * FROM tbl AS t0
2014-01-08 10:48:54 -08:00
Elliott Clark
ade4453a17 Allow impala hbase serdproperties to have newlines 2014-01-08 10:48:54 -08:00
Alex Behm
be03e6c21c IMPALA-138: Error messages for unknown column types are particularly bad. 2014-01-08 10:48:53 -08:00
Alex Behm
a01573af63 IMPALA-65: Add MySQL-style string literals with escaping. 2014-01-08 10:48:51 -08:00
Skye Wanderman-Milne
811d5dd00b Create Avro schema directory in test warehouse 2014-01-08 10:48:50 -08:00
Nong Li
0df9476be1 Parquet data loading. 2014-01-08 10:48:48 -08:00
Lenni Kuff
e0a7b7cb55 Compute column stats on tables used by Planner tests 2014-01-08 10:48:48 -08:00
Nong Li
edaadc5091 Fix old rcfile table 2014-01-08 10:48:47 -08:00
ishaan
5ed84d7f65 IMP-739 Results for show queries should check for subset, not equality. 2014-01-08 10:48:46 -08:00
Skye Wanderman-Milne
461a48df2b Refactor testing framework to generate Avro tables. 2014-01-08 10:48:45 -08:00
Nong Li
18dfcbd361 Fix build break from bad merge in generate-schema-statements. 2014-01-08 10:48:45 -08:00
Nong Li
6e293090e6 Parquet writer.
Change-Id: I7117b545e3d3a7803a219234ad992040a6c7c4ec
2014-01-08 10:48:44 -08:00
Alexander Behm
39e443407b IMPALA-136: GROUP BY float/double. 2014-01-08 10:48:43 -08:00
Nong Li
0385d14d69 Fix pre-hive 9 rc file scanner. 2014-01-08 10:48:41 -08:00
Marcel Kornacker
d7bfe6c68d IMPALA-144: partition pruning for arbitrary predicates that are fully bound by partition columns
This makes partition pruning more effective by extending it to predicates that are fully bound by the partition column,
e.g., '<col> IN (1, 2, 3)' will also be used to prune partitions, in addition to equality and binary comparisons.
2014-01-08 10:48:41 -08:00
Lenni Kuff
328ceed4e7 Add support for generating lzo compressed text files and running tests against lzo 2014-01-08 10:48:38 -08:00