From a75511ae081c020e2a1359387dc06601087a1667 Mon Sep 17 00:00:00 2001 From: airborne12 Date: Wed, 31 Jul 2024 19:39:17 +0800 Subject: [PATCH] [Feature](inverted index) add no need read data optimize config (#38584) pick from #36686 --- .../olap/rowset/segment_v2/segment_iterator.cpp | 6 ++++++ .../org/apache/doris/qe/SessionVariable.java | 16 ++++++++++++++++ gensrc/thrift/PaloInternalService.thrift | 2 ++ 3 files changed, 24 insertions(+) diff --git a/be/src/olap/rowset/segment_v2/segment_iterator.cpp b/be/src/olap/rowset/segment_v2/segment_iterator.cpp index 4c29c7d12b..57754705cf 100644 --- a/be/src/olap/rowset/segment_v2/segment_iterator.cpp +++ b/be/src/olap/rowset/segment_v2/segment_iterator.cpp @@ -1173,6 +1173,9 @@ Status SegmentIterator::_apply_inverted_index_on_block_column_predicate( } bool SegmentIterator::_need_read_data(ColumnId cid) { + if (_opts.runtime_state && !_opts.runtime_state->query_options().enable_no_need_read_data_opt) { + return true; + } // only support DUP_KEYS and UNIQUE_KEYS with MOW if (!((_opts.tablet_schema->keys_type() == KeysType::DUP_KEYS || (_opts.tablet_schema->keys_type() == KeysType::UNIQUE_KEYS && @@ -2782,6 +2785,9 @@ void SegmentIterator::_calculate_pred_in_remaining_conjunct_root( bool SegmentIterator::_no_need_read_key_data(ColumnId cid, vectorized::MutableColumnPtr& column, size_t nrows_read) { + if (_opts.runtime_state && !_opts.runtime_state->query_options().enable_no_need_read_data_opt) { + return false; + } if (_opts.tablet_schema->keys_type() != KeysType::DUP_KEYS) { return false; } diff --git a/fe/fe-core/src/main/java/org/apache/doris/qe/SessionVariable.java b/fe/fe-core/src/main/java/org/apache/doris/qe/SessionVariable.java index 606259c21c..2f8d8cdf6f 100644 --- a/fe/fe-core/src/main/java/org/apache/doris/qe/SessionVariable.java +++ b/fe/fe-core/src/main/java/org/apache/doris/qe/SessionVariable.java @@ -390,6 +390,7 @@ public class SessionVariable implements Serializable, Writable { public static final String ENABLE_INVERTED_INDEX_QUERY = "enable_inverted_index_query"; public static final String ENABLE_PUSHDOWN_COUNT_ON_INDEX = "enable_count_on_index_pushdown"; + public static final String ENABLE_NO_NEED_READ_DATA_OPT = "enable_no_need_read_data_opt"; public static final String GROUP_BY_AND_HAVING_USE_ALIAS_FIRST = "group_by_and_having_use_alias_first"; public static final String DROP_TABLE_IF_CTAS_FAILED = "drop_table_if_ctas_failed"; @@ -1426,6 +1427,11 @@ public class SessionVariable implements Serializable, Writable { "是否启用count_on_index pushdown。", "Set whether to pushdown count_on_index."}) public boolean enablePushDownCountOnIndex = true; + // Whether enable no need read data opt in segment_iterator. + @VariableMgr.VarAttr(name = ENABLE_NO_NEED_READ_DATA_OPT, needForward = true, description = { + "是否启用no_need_read_data opt。", "Set whether to enable no_need_read_data opt."}) + public boolean enableNoNeedReadDataOpt = true; + // Whether enable pushdown minmax to scan node of unique table. @VariableMgr.VarAttr(name = ENABLE_PUSHDOWN_MINMAX_ON_UNIQUE, needForward = true, description = { "是否启用pushdown minmax on unique table。", "Set whether to pushdown minmax on unique table."}) @@ -3239,6 +3245,14 @@ public class SessionVariable implements Serializable, Writable { this.enablePushDownCountOnIndex = enablePushDownCountOnIndex; } + public boolean isEnableNoNeedReadDataOpt() { + return enableNoNeedReadDataOpt; + } + + public void setEnableNoNeedReadDataOpt(boolean enableNoNeedReadDataOpt) { + this.enableNoNeedReadDataOpt = enableNoNeedReadDataOpt; + } + public int getMaxTableCountUseCascadesJoinReorder() { return this.maxTableCountUseCascadesJoinReorder; } @@ -3378,6 +3392,8 @@ public class SessionVariable implements Serializable, Writable { tResult.setEnableInvertedIndexQuery(enableInvertedIndexQuery); + tResult.setEnableNoNeedReadDataOpt(enableNoNeedReadDataOpt); + if (dryRunQuery) { tResult.setDryRunQuery(true); } diff --git a/gensrc/thrift/PaloInternalService.thrift b/gensrc/thrift/PaloInternalService.thrift index 995385ddc9..e2e25619ab 100644 --- a/gensrc/thrift/PaloInternalService.thrift +++ b/gensrc/thrift/PaloInternalService.thrift @@ -301,6 +301,8 @@ struct TQueryOptions { 112: optional i32 max_column_reader_num = 0 113: optional bool enable_force_spill = false; + + 116: optional bool enable_no_need_read_data_opt = true; 117: optional bool read_csv_empty_line_as_null = false;