diff --git itests/qtest/pom.xml itests/qtest/pom.xml index e194441..c3cbb89 100644 --- itests/qtest/pom.xml +++ itests/qtest/pom.xml @@ -36,7 +36,7 @@ false false - list_bucket_dml_10.q,input16_cc.q,scriptfile1.q,scriptfile1_win.q,bucket4.q,bucketmapjoin6.q,disable_merge_for_bucketing.q,reduce_deduplicate.q,smb_mapjoin_8.q,join1.q,groupby2.q,bucketizedhiveinputformat.q,bucketmapjoin7.q,optrstat_groupby.q,bucket_num_reducers.q,bucket5.q,load_fs2.q,bucket_num_reducers2.q,infer_bucket_sort_merge.q,infer_bucket_sort_reducers_power_two.q,infer_bucket_sort_dyn_part.q,infer_bucket_sort_bucketed_table.q,infer_bucket_sort_map_operators.q,infer_bucket_sort_num_buckets.q,leftsemijoin_mr.q,schemeAuthority.q,schemeAuthority2.q,truncate_column_buckets.q,remote_script.q,,load_hdfs_file_with_space_in_the_name.q,parallel_orderby.q,import_exported_table.q,stats_counter.q + stats_counter_partitioned.q,list_bucket_dml_10.q,input16_cc.q,scriptfile1.q,scriptfile1_win.q,bucket4.q,bucketmapjoin6.q,disable_merge_for_bucketing.q,reduce_deduplicate.q,smb_mapjoin_8.q,join1.q,groupby2.q,bucketizedhiveinputformat.q,bucketmapjoin7.q,optrstat_groupby.q,bucket_num_reducers.q,bucket5.q,load_fs2.q,bucket_num_reducers2.q,infer_bucket_sort_merge.q,infer_bucket_sort_reducers_power_two.q,infer_bucket_sort_dyn_part.q,infer_bucket_sort_bucketed_table.q,infer_bucket_sort_map_operators.q,infer_bucket_sort_num_buckets.q,leftsemijoin_mr.q,schemeAuthority.q,schemeAuthority2.q,truncate_column_buckets.q,remote_script.q,,load_hdfs_file_with_space_in_the_name.q,parallel_orderby.q,import_exported_table.q,stats_counter.q cluster_tasklog_retrieval.q,minimr_broken_pipe.q,mapreduce_stack_trace.q,mapreduce_stack_trace_turnoff.q,mapreduce_stack_trace_hadoop20.q,mapreduce_stack_trace_turnoff_hadoop20.q add_part_exist.q,alter1.q,alter2.q,alter4.q,alter5.q,alter_rename_partition.q,alter_rename_partition_authorization.q,archive.q,archive_corrupt.q,archive_multi.q,archive_mr_1806.q,archive_multi_mr_1806.q,authorization_1.q,authorization_2.q,authorization_4.q,authorization_5.q,authorization_6.q,authorization_7.q,ba_table1.q,ba_table2.q,ba_table3.q,ba_table_udfs.q,binary_table_bincolserde.q,binary_table_colserde.q,cluster.q,columnarserde_create_shortcut.q,combine2.q,constant_prop.q,create_nested_type.q,create_or_replace_view.q,create_struct_table.q,create_union_table.q,database.q,database_location.q,database_properties.q,ddltime.q,describe_database_json.q,drop_database_removes_partition_dirs.q,escape1.q,escape2.q,exim_00_nonpart_empty.q,exim_01_nonpart.q,exim_02_00_part_empty.q,exim_02_part.q,exim_03_nonpart_over_compat.q,exim_04_all_part.q,exim_04_evolved_parts.q,exim_05_some_part.q,exim_06_one_part.q,exim_07_all_part_over_nonoverlap.q,exim_08_nonpart_rename.q,exim_09_part_spec_nonoverlap.q,exim_10_external_managed.q,exim_11_managed_external.q,exim_12_external_location.q,exim_13_managed_location.q,exim_14_managed_location_over_existing.q,exim_15_external_part.q,exim_16_part_external.q,exim_17_part_managed.q,exim_18_part_external.q,exim_19_00_part_external_location.q,exim_19_part_external_location.q,exim_20_part_managed_location.q,exim_21_export_authsuccess.q,exim_22_import_exist_authsuccess.q,exim_23_import_part_authsuccess.q,exim_24_import_nonexist_authsuccess.q,global_limit.q,groupby_complex_types.q,groupby_complex_types_multi_single_reducer.q,index_auth.q,index_auto.q,index_auto_empty.q,index_bitmap.q,index_bitmap1.q,index_bitmap2.q,index_bitmap3.q,index_bitmap_auto.q,index_bitmap_rc.q,index_compact.q,index_compact_1.q,index_compact_2.q,index_compact_3.q,index_stale_partitioned.q,init_file.q,input16.q,input16_cc.q,input46.q,input_columnarserde.q,input_dynamicserde.q,input_lazyserde.q,input_testxpath3.q,input_testxpath4.q,insert2_overwrite_partitions.q,insertexternal1.q,join_thrift.q,lateral_view.q,load_binary_data.q,load_exist_part_authsuccess.q,load_nonpart_authsuccess.q,load_part_authsuccess.q,loadpart_err.q,lock1.q,lock2.q,lock3.q,lock4.q,merge_dynamic_partition.q,multi_insert.q,multi_insert_move_tasks_share_dependencies.q,null_column.q,ppd_clusterby.q,query_with_semi.q,rename_column.q,sample6.q,sample_islocalmode_hook.q,set_processor_namespaces.q,show_tables.q,source.q,split_sample.q,str_to_map.q,transform1.q,udaf_collect_set.q,udaf_context_ngrams.q,udaf_histogram_numeric.q,udaf_ngrams.q,udaf_percentile_approx.q,udf_array.q,udf_bitmap_and.q,udf_bitmap_or.q,udf_explode.q,udf_format_number.q,udf_map.q,udf_map_keys.q,udf_map_values.q,udf_max.q,udf_min.q,udf_named_struct.q,udf_percentile.q,udf_printf.q,udf_sentences.q,udf_sort_array.q,udf_split.q,udf_struct.q,udf_substr.q,udf_translate.q,udf_union.q,udf_xpath.q,udtf_stack.q,view.q,virtual_column.q diff --git ql/src/java/org/apache/hadoop/hive/ql/exec/FileSinkOperator.java ql/src/java/org/apache/hadoop/hive/ql/exec/FileSinkOperator.java index e7453c7..a3a737a 100644 --- ql/src/java/org/apache/hadoop/hive/ql/exec/FileSinkOperator.java +++ ql/src/java/org/apache/hadoop/hive/ql/exec/FileSinkOperator.java @@ -52,6 +52,7 @@ import org.apache.hadoop.hive.ql.plan.PlanUtils; import org.apache.hadoop.hive.ql.plan.SkewedColumnPositionPair; import org.apache.hadoop.hive.ql.plan.api.OperatorType; +import org.apache.hadoop.hive.ql.stats.CounterStatsPublisher; import org.apache.hadoop.hive.ql.stats.StatsPublisher; import org.apache.hadoop.hive.serde2.SerDeException; import org.apache.hadoop.hive.serde2.SerDeStats; @@ -982,7 +983,12 @@ private void publishStats() throws HiveException { } else { // for partitioned table, the key is // common key prefix + static partition spec + DynamicPartSpec + taskID - key = createKeyForStatsPublisher(taskID, spSpec, fspKey); + if (statsPublisher instanceof CounterStatsPublisher) { + key = conf.getStatsAggPrefix() + spSpec; + } else { + key = createKeyForStatsPublisher(taskID, spSpec, fspKey); + } + } Map statsToPublish = new HashMap(); for (String statType : fspValue.stat.getStoredStats()) { diff --git ql/src/java/org/apache/hadoop/hive/ql/exec/StatsTask.java ql/src/java/org/apache/hadoop/hive/ql/exec/StatsTask.java index 142af10..fbda77b 100644 --- ql/src/java/org/apache/hadoop/hive/ql/exec/StatsTask.java +++ ql/src/java/org/apache/hadoop/hive/ql/exec/StatsTask.java @@ -29,8 +29,10 @@ import org.apache.commons.logging.Log; import org.apache.commons.logging.LogFactory; import org.apache.hadoop.fs.FileStatus; +import org.apache.hadoop.fs.Path; import org.apache.hadoop.hive.common.StatsSetupConst; import org.apache.hadoop.hive.conf.HiveConf; +import org.apache.hadoop.hive.conf.HiveConf.ConfVars; import org.apache.hadoop.hive.metastore.Warehouse; import org.apache.hadoop.hive.ql.DriverContext; import org.apache.hadoop.hive.ql.ErrorMsg; @@ -226,7 +228,11 @@ private int aggregateStats() { updateStats(StatsSetupConst.statsRequireCompute, tblStats, statsAggregator, parameters, aggKey, atomic); statsAggregator.cleanUp(aggKey); + Map allStats = tblStats.stats; + } + + // The collectable stats for the aggregator needs to be cleared. // For eg. if a file is being loaded, the old number of rows are not valid else if (work.isClearAggregatorStats()) { @@ -236,7 +242,8 @@ else if (work.isClearAggregatorStats()) { } } } - + Map allStats = tblStats.stats; + // write table stats to metastore parameters = tTable.getParameters(); for (String statType : StatsSetupConst.statsRequireCompute) { @@ -278,16 +285,28 @@ else if (work.isClearAggregatorStats()) { // Statistics newPartStats = new Statistics(); - // In that case of a partition, the key for stats temporary store is - // "rootDir/[dynamic_partition_specs/]%" - String partitionID = Utilities.getHashedStatsPrefix( - work.getAggKey() + Warehouse.makePartPath(partn.getSpec()), maxPrefixLength); + String partitionID; + + if(HiveConf.getVar(conf, ConfVars.HIVESTATSDBCLASS).equals("counter")) { + partitionID = table.getTableName() +Path.SEPARATOR + Warehouse.makePartPath(partn.getSpec()); + // there is no need to aggregate stats in this case, but this should also work. + // also check non-partitioned code path. + } else { + // In that case of a partition, the key for stats temporary store is + // "rootDir/[dynamic_partition_specs/]%" + partitionID = Utilities.getHashedStatsPrefix( + work.getAggKey() + Warehouse.makePartPath(partn.getSpec()), maxPrefixLength); + } + LOG.info("Stats aggregator : " + partitionID); if (statsAggregator != null) { updateStats(StatsSetupConst.statsRequireCompute, newPartStats, statsAggregator, parameters, partitionID, atomic); + for (String stat : newPartStats.stats.keySet()) { + LOG.debug("Stat : " + stat + "\t" + newPartStats.stats.get(stat)); + } statsAggregator.cleanUp(partitionID); } else { for (String statType : StatsSetupConst.statsRequireCompute) { @@ -303,7 +322,9 @@ else if (work.isClearAggregatorStats()) { } } } - + for (String stat : newPartStats.stats.keySet()) { + LOG.debug("Stat : " + stat + "\t" + newPartStats.stats.get(stat)); + } /** * calculate fast statistics */ diff --git ql/src/java/org/apache/hadoop/hive/ql/exec/TableScanOperator.java ql/src/java/org/apache/hadoop/hive/ql/exec/TableScanOperator.java index 8d895f4..bad1383 100644 --- ql/src/java/org/apache/hadoop/hive/ql/exec/TableScanOperator.java +++ ql/src/java/org/apache/hadoop/hive/ql/exec/TableScanOperator.java @@ -28,6 +28,8 @@ import org.apache.hadoop.fs.Path; import org.apache.hadoop.hive.common.FileUtils; import org.apache.hadoop.hive.common.StatsSetupConst; +import org.apache.hadoop.hive.conf.HiveConf; +import org.apache.hadoop.hive.conf.HiveConf.ConfVars; import org.apache.hadoop.hive.ql.ErrorMsg; import org.apache.hadoop.hive.ql.metadata.HiveException; import org.apache.hadoop.hive.ql.metadata.VirtualColumn; @@ -296,11 +298,20 @@ private void publishStats() throws HiveException { conf.getStatsAggPrefix(), conf.getMaxStatsKeyPrefixLength()); key = keyPrefix + taskID; } else { - // In case of a partition, the key for temp storage is - // "tableName + partitionSpecs + taskID" - String keyPrefix = Utilities.getHashedStatsPrefix( - conf.getStatsAggPrefix() + pspecs, conf.getMaxStatsKeyPrefixLength()); - key = keyPrefix + taskID; + + if (HiveConf.getVar(hconf, ConfVars.HIVESTATSDBCLASS).equals("counter")) { + // Can we do something within CounterStatsPublisher to avoid this special logic + // in TSOp. + // also, do we need same changes in FileSinkOp? + // Also, check non-partitioned tables case. + key = Utilities.appendPathSeparator(conf.getStatsAggPrefix() + pspecs); + } else { + // In case of a partition, the key for temp storage is + // "tableName + partitionSpecs + taskID" + String keyPrefix = Utilities.getHashedStatsPrefix( + conf.getStatsAggPrefix() + pspecs, conf.getMaxStatsKeyPrefixLength()); + key = keyPrefix + taskID; + } } for(String statType : stats.get(pspecs).getStoredStats()) { statsToPublish.put(statType, Long.toString(stats.get(pspecs).getStat(statType))); diff --git ql/src/java/org/apache/hadoop/hive/ql/exec/Utilities.java ql/src/java/org/apache/hadoop/hive/ql/exec/Utilities.java index 365c22a..eaa3f05 100644 --- ql/src/java/org/apache/hadoop/hive/ql/exec/Utilities.java +++ ql/src/java/org/apache/hadoop/hive/ql/exec/Utilities.java @@ -2291,7 +2291,7 @@ public static String getHashedStatsPrefix(String statsPrefix, int maxPrefixLengt return ret; } - private static String appendPathSeparator(String path) { + public static String appendPathSeparator(String path) { if (!path.endsWith(Path.SEPARATOR)) { path = path + Path.SEPARATOR; } diff --git ql/src/java/org/apache/hadoop/hive/ql/optimizer/GenMRTableScan1.java ql/src/java/org/apache/hadoop/hive/ql/optimizer/GenMRTableScan1.java index 0268f98..af729e6 100644 --- ql/src/java/org/apache/hadoop/hive/ql/optimizer/GenMRTableScan1.java +++ ql/src/java/org/apache/hadoop/hive/ql/optimizer/GenMRTableScan1.java @@ -35,6 +35,7 @@ import org.apache.hadoop.hive.ql.exec.TableScanOperator; import org.apache.hadoop.hive.ql.exec.Task; import org.apache.hadoop.hive.ql.exec.TaskFactory; +import org.apache.hadoop.hive.ql.exec.mr.MapRedTask; import org.apache.hadoop.hive.ql.io.rcfile.stats.PartialScanWork; import org.apache.hadoop.hive.ql.lib.Node; import org.apache.hadoop.hive.ql.lib.NodeProcessor; @@ -95,6 +96,7 @@ public Object process(Node nd, Stack stack, NodeProcessorCtx opProcCtx, StatsWork statsWork = new StatsWork(parseCtx.getQB().getParseInfo().getTableSpec()); statsWork.setAggKey(op.getConf().getStatsAggPrefix()); + statsWork.setSourceTask((MapRedTask)currTask); statsWork.setStatsReliable( parseCtx.getConf().getBoolVar(HiveConf.ConfVars.HIVE_STATS_RELIABLE)); Task statsTask = TaskFactory.get(statsWork, parseCtx.getConf()); diff --git ql/src/java/org/apache/hadoop/hive/ql/stats/CounterStatsAggregator.java ql/src/java/org/apache/hadoop/hive/ql/stats/CounterStatsAggregator.java index 2cc2519..17f5d7a 100644 --- ql/src/java/org/apache/hadoop/hive/ql/stats/CounterStatsAggregator.java +++ ql/src/java/org/apache/hadoop/hive/ql/stats/CounterStatsAggregator.java @@ -19,6 +19,7 @@ package org.apache.hadoop.hive.ql.stats; import java.io.IOException; +import java.util.Iterator; import org.apache.commons.logging.Log; import org.apache.commons.logging.LogFactory; @@ -26,6 +27,7 @@ import org.apache.hadoop.hive.ql.exec.mr.ExecDriver; import org.apache.hadoop.hive.ql.exec.mr.MapRedTask; import org.apache.hadoop.mapred.Counters; +import org.apache.hadoop.mapred.Counters.Counter; import org.apache.hadoop.mapred.JobClient; import org.apache.hadoop.mapred.JobConf; import org.apache.hadoop.mapred.RunningJob; @@ -46,6 +48,7 @@ public boolean connect(Configuration hconf, MapRedTask sourceTask) { counters = job.getCounters(); } } catch (Exception e) { + LOG.error(e); LOG.error("Failed to get Job instance for " + sourceTask.getJobID()); } return counters != null; @@ -57,13 +60,17 @@ private JobConf toJobConf(Configuration hconf) { @Override public String aggregateStats(String keyPrefix, String statType) { - long value = 0; - for (String groupName : counters.getGroupNames()) { - if (groupName.startsWith(keyPrefix)) { - value += counters.getGroup(groupName).getCounter(statType); - } - } - return String.valueOf(value); + LOG.debug("key : " + keyPrefix + "\t" + statType); + for (String cntrName : counters.getGroupNames()) { + LOG.info("Counter grp name: " + cntrName); + Iterator cntItr = counters.getGroup(cntrName).iterator(); + while(cntItr.hasNext()) { + Counter cntr = cntItr.next(); + LOG.info("Counter value: " + cntr.getName() + "\t" + cntr.getValue()); + } + + } + return String.valueOf(counters.getGroup(keyPrefix).getCounter(statType)); } @Override diff --git ql/src/test/queries/clientpositive/stats_counter_partitioned.q ql/src/test/queries/clientpositive/stats_counter_partitioned.q new file mode 100644 index 0000000..282012f --- /dev/null +++ ql/src/test/queries/clientpositive/stats_counter_partitioned.q @@ -0,0 +1,17 @@ +set hive.stats.dbclass=counter; + +set hive.stats.autogather=true; + + +create table dummy (key string, value string) partitioned by (ds string, hr string); +load data local inpath '../../data/files/kv1.txt' into table dummy partition (ds='2008',hr='12'); +load data local inpath '../../data/files/kv1.txt' into table dummy partition (ds='2008',hr='11'); +analyze table dummy partition (ds,hr) compute statistics; + describe formatted dummy partition (ds='2008', hr='11'); + describe formatted dummy partition (ds='2008', hr='12'); + + + + +drop table dummy; + diff --git ql/src/test/results/clientpositive/stats_counter_partitioned.q.out ql/src/test/results/clientpositive/stats_counter_partitioned.q.out new file mode 100644 index 0000000..4f3be33 --- /dev/null +++ ql/src/test/results/clientpositive/stats_counter_partitioned.q.out @@ -0,0 +1,123 @@ +PREHOOK: query: create table dummy (key string, value string) partitioned by (ds string, hr string) +PREHOOK: type: CREATETABLE +POSTHOOK: query: create table dummy (key string, value string) partitioned by (ds string, hr string) +POSTHOOK: type: CREATETABLE +POSTHOOK: Output: default@dummy +PREHOOK: query: load data local inpath '../../data/files/kv1.txt' into table dummy partition (ds='2008',hr='12') +PREHOOK: type: LOAD +PREHOOK: Output: default@dummy +POSTHOOK: query: load data local inpath '../../data/files/kv1.txt' into table dummy partition (ds='2008',hr='12') +POSTHOOK: type: LOAD +POSTHOOK: Output: default@dummy +POSTHOOK: Output: default@dummy@ds=2008/hr=12 +PREHOOK: query: load data local inpath '../../data/files/kv1.txt' into table dummy partition (ds='2008',hr='11') +PREHOOK: type: LOAD +PREHOOK: Output: default@dummy +POSTHOOK: query: load data local inpath '../../data/files/kv1.txt' into table dummy partition (ds='2008',hr='11') +POSTHOOK: type: LOAD +POSTHOOK: Output: default@dummy +POSTHOOK: Output: default@dummy@ds=2008/hr=11 +PREHOOK: query: analyze table dummy partition (ds,hr) compute statistics +PREHOOK: type: QUERY +PREHOOK: Input: default@dummy +PREHOOK: Input: default@dummy@ds=2008/hr=11 +PREHOOK: Input: default@dummy@ds=2008/hr=12 +PREHOOK: Output: default@dummy +PREHOOK: Output: default@dummy@ds=2008/hr=11 +PREHOOK: Output: default@dummy@ds=2008/hr=12 +POSTHOOK: query: analyze table dummy partition (ds,hr) compute statistics +POSTHOOK: type: QUERY +POSTHOOK: Input: default@dummy +POSTHOOK: Input: default@dummy@ds=2008/hr=11 +POSTHOOK: Input: default@dummy@ds=2008/hr=12 +POSTHOOK: Output: default@dummy +POSTHOOK: Output: default@dummy@ds=2008/hr=11 +POSTHOOK: Output: default@dummy@ds=2008/hr=12 +PREHOOK: query: describe formatted dummy partition (ds='2008', hr='11') +PREHOOK: type: DESCTABLE +POSTHOOK: query: describe formatted dummy partition (ds='2008', hr='11') +POSTHOOK: type: DESCTABLE +# col_name data_type comment + +key string None +value string None + +# Partition Information +# col_name data_type comment + +ds string None +hr string None + +# Detailed Partition Information +Partition Value: [2008, 11] +Database: default +Table: dummy +#### A masked pattern was here #### +Protect Mode: None +#### A masked pattern was here #### +Partition Parameters: + COLUMN_STATS_ACCURATE true + numFiles 1 + numRows 500 + rawDataSize 5312 + totalSize 5812 +#### A masked pattern was here #### + +# Storage Information +SerDe Library: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe +InputFormat: org.apache.hadoop.mapred.TextInputFormat +OutputFormat: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat +Compressed: No +Num Buckets: -1 +Bucket Columns: [] +Sort Columns: [] +Storage Desc Params: + serialization.format 1 +PREHOOK: query: describe formatted dummy partition (ds='2008', hr='12') +PREHOOK: type: DESCTABLE +POSTHOOK: query: describe formatted dummy partition (ds='2008', hr='12') +POSTHOOK: type: DESCTABLE +# col_name data_type comment + +key string None +value string None + +# Partition Information +# col_name data_type comment + +ds string None +hr string None + +# Detailed Partition Information +Partition Value: [2008, 12] +Database: default +Table: dummy +#### A masked pattern was here #### +Protect Mode: None +#### A masked pattern was here #### +Partition Parameters: + COLUMN_STATS_ACCURATE true + numFiles 1 + numRows 500 + rawDataSize 5312 + totalSize 5812 +#### A masked pattern was here #### + +# Storage Information +SerDe Library: org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe +InputFormat: org.apache.hadoop.mapred.TextInputFormat +OutputFormat: org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat +Compressed: No +Num Buckets: -1 +Bucket Columns: [] +Sort Columns: [] +Storage Desc Params: + serialization.format 1 +PREHOOK: query: drop table dummy +PREHOOK: type: DROPTABLE +PREHOOK: Input: default@dummy +PREHOOK: Output: default@dummy +POSTHOOK: query: drop table dummy +POSTHOOK: type: DROPTABLE +POSTHOOK: Input: default@dummy +POSTHOOK: Output: default@dummy