PREHOOK: query: create table hive_test_smb_bucket1 (key int, value string) partitioned by (ds string) clustered by (key) sorted by (key) into 2 buckets PREHOOK: type: CREATETABLE PREHOOK: Output: database:default PREHOOK: Output: default@hive_test_smb_bucket1 POSTHOOK: query: create table hive_test_smb_bucket1 (key int, value string) partitioned by (ds string) clustered by (key) sorted by (key) into 2 buckets POSTHOOK: type: CREATETABLE POSTHOOK: Output: database:default POSTHOOK: Output: default@hive_test_smb_bucket1 PREHOOK: query: create table hive_test_smb_bucket2 (key int, value string) partitioned by (ds string) clustered by (key) sorted by (key) into 2 buckets PREHOOK: type: CREATETABLE PREHOOK: Output: database:default PREHOOK: Output: default@hive_test_smb_bucket2 POSTHOOK: query: create table hive_test_smb_bucket2 (key int, value string) partitioned by (ds string) clustered by (key) sorted by (key) into 2 buckets POSTHOOK: type: CREATETABLE POSTHOOK: Output: database:default POSTHOOK: Output: default@hive_test_smb_bucket2 PREHOOK: query: -- empty partitions (HIVE-3205) explain extended SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL PREHOOK: type: QUERY POSTHOOK: query: -- empty partitions (HIVE-3205) explain extended SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL POSTHOOK: type: QUERY ABSTRACT SYNTAX TREE: TOK_QUERY TOK_FROM TOK_JOIN TOK_TABREF TOK_TABNAME hive_test_smb_bucket1 a TOK_TABREF TOK_TABNAME hive_test_smb_bucket2 b = . TOK_TABLE_OR_COL a key . TOK_TABLE_OR_COL b key TOK_INSERT TOK_DESTINATION TOK_DIR TOK_TMP_FILE TOK_SELECT TOK_HINTLIST TOK_HINT TOK_MAPJOIN TOK_HINTARGLIST b TOK_SELEXPR . TOK_TABLE_OR_COL b key k1 TOK_SELEXPR . TOK_TABLE_OR_COL b value TOK_SELEXPR . TOK_TABLE_OR_COL b ds TOK_SELEXPR . TOK_TABLE_OR_COL a key k2 TOK_WHERE and and = . TOK_TABLE_OR_COL a ds '2010-10-15' = . TOK_TABLE_OR_COL b ds '2010-10-15' TOK_FUNCTION TOK_ISNOTNULL . TOK_TABLE_OR_COL b key STAGE DEPENDENCIES: Stage-1 is a root stage Stage-0 depends on stages: Stage-1 STAGE PLANS: Stage: Stage-1 Map Reduce Map Operator Tree: TableScan alias: a Statistics: Num rows: 0 Data size: 0 Basic stats: NONE Column stats: NONE GatherStats: false Filter Operator isSamplingPred: false predicate: (key is not null and (ds = '2010-10-15')) (type: boolean) Statistics: Num rows: 0 Data size: 0 Basic stats: NONE Column stats: NONE Sorted Merge Bucket Map Join Operator condition map: Inner Join 0 to 1 keys: 0 key (type: int) 1 key (type: int) outputColumnNames: _col0, _col6, _col7 Position of Big Table: 0 Select Operator expressions: _col6 (type: int), _col7 (type: string), '2010-10-15' (type: string), _col0 (type: int) outputColumnNames: _col0, _col1, _col2, _col3 File Output Operator compressed: false GlobalTableId: 0 #### A masked pattern was here #### NumFilesPerFileSink: 1 #### A masked pattern was here #### table: input format: org.apache.hadoop.mapred.TextInputFormat output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat properties: columns _col0,_col1,_col2,_col3 columns.types int:string:string:int escape.delim \ hive.serialization.extend.additional.nesting.levels true serialization.format 1 serialization.lib org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe TotalFiles: 1 GatherStats: false MultiFileSpray: false Stage: Stage-0 Fetch Operator limit: -1 Processor Tree: ListSink PREHOOK: query: SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL PREHOOK: type: QUERY PREHOOK: Input: default@hive_test_smb_bucket1 PREHOOK: Input: default@hive_test_smb_bucket2 #### A masked pattern was here #### POSTHOOK: query: SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL POSTHOOK: type: QUERY POSTHOOK: Input: default@hive_test_smb_bucket1 POSTHOOK: Input: default@hive_test_smb_bucket2 #### A masked pattern was here #### PREHOOK: query: explain extended SELECT /* + MAPJOIN(a) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL PREHOOK: type: QUERY POSTHOOK: query: explain extended SELECT /* + MAPJOIN(a) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL POSTHOOK: type: QUERY ABSTRACT SYNTAX TREE: TOK_QUERY TOK_FROM TOK_JOIN TOK_TABREF TOK_TABNAME hive_test_smb_bucket1 a TOK_TABREF TOK_TABNAME hive_test_smb_bucket2 b = . TOK_TABLE_OR_COL a key . TOK_TABLE_OR_COL b key TOK_INSERT TOK_DESTINATION TOK_DIR TOK_TMP_FILE TOK_SELECT TOK_HINTLIST TOK_HINT TOK_MAPJOIN TOK_HINTARGLIST a TOK_SELEXPR . TOK_TABLE_OR_COL b key k1 TOK_SELEXPR . TOK_TABLE_OR_COL b value TOK_SELEXPR . TOK_TABLE_OR_COL b ds TOK_SELEXPR . TOK_TABLE_OR_COL a key k2 TOK_WHERE and and = . TOK_TABLE_OR_COL a ds '2010-10-15' = . TOK_TABLE_OR_COL b ds '2010-10-15' TOK_FUNCTION TOK_ISNOTNULL . TOK_TABLE_OR_COL b key STAGE DEPENDENCIES: Stage-1 is a root stage Stage-0 depends on stages: Stage-1 STAGE PLANS: Stage: Stage-1 Map Reduce Map Operator Tree: TableScan alias: b Statistics: Num rows: 0 Data size: 0 Basic stats: NONE Column stats: NONE GatherStats: false Filter Operator isSamplingPred: false predicate: (key is not null and (ds = '2010-10-15')) (type: boolean) Statistics: Num rows: 0 Data size: 0 Basic stats: NONE Column stats: NONE Sorted Merge Bucket Map Join Operator condition map: Inner Join 0 to 1 keys: 0 key (type: int) 1 key (type: int) outputColumnNames: _col0, _col6, _col7 Position of Big Table: 1 Select Operator expressions: _col6 (type: int), _col7 (type: string), '2010-10-15' (type: string), _col0 (type: int) outputColumnNames: _col0, _col1, _col2, _col3 File Output Operator compressed: false GlobalTableId: 0 #### A masked pattern was here #### NumFilesPerFileSink: 1 #### A masked pattern was here #### table: input format: org.apache.hadoop.mapred.TextInputFormat output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat properties: columns _col0,_col1,_col2,_col3 columns.types int:string:string:int escape.delim \ hive.serialization.extend.additional.nesting.levels true serialization.format 1 serialization.lib org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe TotalFiles: 1 GatherStats: false MultiFileSpray: false Stage: Stage-0 Fetch Operator limit: -1 Processor Tree: ListSink PREHOOK: query: SELECT /* + MAPJOIN(a) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL PREHOOK: type: QUERY PREHOOK: Input: default@hive_test_smb_bucket1 PREHOOK: Input: default@hive_test_smb_bucket2 #### A masked pattern was here #### POSTHOOK: query: SELECT /* + MAPJOIN(a) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL POSTHOOK: type: QUERY POSTHOOK: Input: default@hive_test_smb_bucket1 POSTHOOK: Input: default@hive_test_smb_bucket2 #### A masked pattern was here #### PREHOOK: query: insert overwrite table hive_test_smb_bucket1 partition (ds='2010-10-15') select key, value from src PREHOOK: type: QUERY PREHOOK: Input: default@src PREHOOK: Output: default@hive_test_smb_bucket1@ds=2010-10-15 POSTHOOK: query: insert overwrite table hive_test_smb_bucket1 partition (ds='2010-10-15') select key, value from src POSTHOOK: type: QUERY POSTHOOK: Input: default@src POSTHOOK: Output: default@hive_test_smb_bucket1@ds=2010-10-15 POSTHOOK: Lineage: hive_test_smb_bucket1 PARTITION(ds=2010-10-15).key EXPRESSION [(src)src.FieldSchema(name:key, type:string, comment:default), ] POSTHOOK: Lineage: hive_test_smb_bucket1 PARTITION(ds=2010-10-15).value SIMPLE [(src)src.FieldSchema(name:value, type:string, comment:default), ] PREHOOK: query: insert overwrite table hive_test_smb_bucket2 partition (ds='2010-10-15') select key, value from src PREHOOK: type: QUERY PREHOOK: Input: default@src PREHOOK: Output: default@hive_test_smb_bucket2@ds=2010-10-15 POSTHOOK: query: insert overwrite table hive_test_smb_bucket2 partition (ds='2010-10-15') select key, value from src POSTHOOK: type: QUERY POSTHOOK: Input: default@src POSTHOOK: Output: default@hive_test_smb_bucket2@ds=2010-10-15 POSTHOOK: Lineage: hive_test_smb_bucket2 PARTITION(ds=2010-10-15).key EXPRESSION [(src)src.FieldSchema(name:key, type:string, comment:default), ] POSTHOOK: Lineage: hive_test_smb_bucket2 PARTITION(ds=2010-10-15).value SIMPLE [(src)src.FieldSchema(name:value, type:string, comment:default), ] PREHOOK: query: explain create table smb_mapjoin9_results as SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL PREHOOK: type: CREATETABLE_AS_SELECT POSTHOOK: query: explain create table smb_mapjoin9_results as SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL POSTHOOK: type: CREATETABLE_AS_SELECT STAGE DEPENDENCIES: Stage-1 is a root stage Stage-7 depends on stages: Stage-1 , consists of Stage-4, Stage-3, Stage-5 Stage-4 Stage-0 depends on stages: Stage-4, Stage-3, Stage-6 Stage-9 depends on stages: Stage-0 Stage-2 depends on stages: Stage-9 Stage-3 Stage-5 Stage-6 depends on stages: Stage-5 STAGE PLANS: Stage: Stage-1 Map Reduce Map Operator Tree: TableScan alias: a Statistics: Num rows: 500 Data size: 5312 Basic stats: COMPLETE Column stats: NONE Filter Operator predicate: key is not null (type: boolean) Statistics: Num rows: 250 Data size: 2656 Basic stats: COMPLETE Column stats: NONE Sorted Merge Bucket Map Join Operator condition map: Inner Join 0 to 1 keys: 0 key (type: int) 1 key (type: int) outputColumnNames: _col0, _col6, _col7 Select Operator expressions: _col6 (type: int), _col7 (type: string), '2010-10-15' (type: string), _col0 (type: int) outputColumnNames: _col0, _col1, _col2, _col3 File Output Operator compressed: false table: input format: org.apache.hadoop.mapred.TextInputFormat output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe name: default.smb_mapjoin9_results Stage: Stage-7 Conditional Operator Stage: Stage-4 Move Operator files: hdfs directory: true #### A masked pattern was here #### Stage: Stage-0 Move Operator files: hdfs directory: true #### A masked pattern was here #### Stage: Stage-9 Create Table Operator: Create Table columns: k1 int, value string, ds string, k2 int input format: org.apache.hadoop.mapred.TextInputFormat output format: org.apache.hadoop.hive.ql.io.IgnoreKeyTextOutputFormat serde name: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe name: default.smb_mapjoin9_results Stage: Stage-2 Stats-Aggr Operator Stage: Stage-3 Map Reduce Map Operator Tree: TableScan File Output Operator compressed: false table: input format: org.apache.hadoop.mapred.TextInputFormat output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe name: default.smb_mapjoin9_results Stage: Stage-5 Map Reduce Map Operator Tree: TableScan File Output Operator compressed: false table: input format: org.apache.hadoop.mapred.TextInputFormat output format: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat serde: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe name: default.smb_mapjoin9_results Stage: Stage-6 Move Operator files: hdfs directory: true #### A masked pattern was here #### PREHOOK: query: create table smb_mapjoin9_results as SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL PREHOOK: type: CREATETABLE_AS_SELECT PREHOOK: Input: default@hive_test_smb_bucket1 PREHOOK: Input: default@hive_test_smb_bucket1@ds=2010-10-15 PREHOOK: Input: default@hive_test_smb_bucket2 PREHOOK: Input: default@hive_test_smb_bucket2@ds=2010-10-15 PREHOOK: Output: database:default PREHOOK: Output: default@smb_mapjoin9_results POSTHOOK: query: create table smb_mapjoin9_results as SELECT /* + MAPJOIN(b) */ b.key as k1, b.value, b.ds, a.key as k2 FROM hive_test_smb_bucket1 a JOIN hive_test_smb_bucket2 b ON a.key = b.key WHERE a.ds = '2010-10-15' and b.ds='2010-10-15' and b.key IS NOT NULL POSTHOOK: type: CREATETABLE_AS_SELECT POSTHOOK: Input: default@hive_test_smb_bucket1 POSTHOOK: Input: default@hive_test_smb_bucket1@ds=2010-10-15 POSTHOOK: Input: default@hive_test_smb_bucket2 POSTHOOK: Input: default@hive_test_smb_bucket2@ds=2010-10-15 POSTHOOK: Output: database:default POSTHOOK: Output: default@smb_mapjoin9_results PREHOOK: query: drop table smb_mapjoin9_results PREHOOK: type: DROPTABLE PREHOOK: Input: default@smb_mapjoin9_results PREHOOK: Output: default@smb_mapjoin9_results POSTHOOK: query: drop table smb_mapjoin9_results POSTHOOK: type: DROPTABLE POSTHOOK: Input: default@smb_mapjoin9_results POSTHOOK: Output: default@smb_mapjoin9_results PREHOOK: query: drop table hive_test_smb_bucket1 PREHOOK: type: DROPTABLE PREHOOK: Input: default@hive_test_smb_bucket1 PREHOOK: Output: default@hive_test_smb_bucket1 POSTHOOK: query: drop table hive_test_smb_bucket1 POSTHOOK: type: DROPTABLE POSTHOOK: Input: default@hive_test_smb_bucket1 POSTHOOK: Output: default@hive_test_smb_bucket1 PREHOOK: query: drop table hive_test_smb_bucket2 PREHOOK: type: DROPTABLE PREHOOK: Input: default@hive_test_smb_bucket2 PREHOOK: Output: default@hive_test_smb_bucket2 POSTHOOK: query: drop table hive_test_smb_bucket2 POSTHOOK: type: DROPTABLE POSTHOOK: Input: default@hive_test_smb_bucket2 POSTHOOK: Output: default@hive_test_smb_bucket2