fix(PrimProc): MCOL-5852 use only long string storage

for the group_concat data to reduce memory usage

dbcon/joblist/groupconcat.cpp

@@ -332,6 +332,7 @@ void GroupConcatAgUM::initialize()
   {
     fRowGroup = fGroupConcat->fRowGroup;
     fRowGroup.setUseStringTable(true);
+    fRowGroup.setUseOnlyLongString(true);
     fRowRGData.reinit(fRowGroup, 1);
     fRowGroup.setData(&fRowRGData);
     fRowGroup.resetRowGroup(0);
@@ -998,6 +999,7 @@ void GroupConcatNoOrder::initialize(const rowgroup::SP_GroupConcat& gcc)
   GroupConcator::initialize(gcc);
 
   fRowGroup = gcc->fRowGroup;
+  fRowGroup.setUseOnlyLongString(true);
   fRowsPerRG = 128;
   fErrorCode = ERR_AGGREGATION_TOO_BIG;
   fRm = gcc->fRm;
@@ -1010,7 +1012,7 @@ void GroupConcatNoOrder::initialize(const rowgroup::SP_GroupConcat& gcc)
 
   uint64_t newSize = fRowsPerRG * fRowGroup.getRowSize();
 
-  if (!fRm->getMemory(newSize, fSessionMemLimit))
+  if (fRm && !fRm->getMemory(newSize, fSessionMemLimit))
   {
     cerr << IDBErrorInfo::instance()->errorMsg(fErrorCode) << " @" << __FILE__ << ":" << __LINE__;
     throw IDBExcept(fErrorCode);
@@ -1020,6 +1022,7 @@ void GroupConcatNoOrder::initialize(const rowgroup::SP_GroupConcat& gcc)
 
   fData.reinit(fRowGroup, fRowsPerRG);
   fRowGroup.setData(&fData);
+  fRowGroup.setUseOnlyLongString(true);
   fRowGroup.resetRowGroup(0);
   fRowGroup.initRow(&fRow);
   fRowGroup.getRow(0, &fRow);

utils/rowgroup/rowgroup.cpp

@@ -79,7 +79,7 @@ uint64_t StringStore::storeString(const uint8_t* data, uint32_t len)
   if (mem.size() > 0)
     lastMC = (MemChunk*)mem.back().get();
 
-  if ((len + 4) >= CHUNK_SIZE)
+  if ((len + 4) >= CHUNK_SIZE || fUseOnlyLongStrings)
   {
     auto allocSize = len + sizeof(MemChunk) + 4;
     if (alloc)
@@ -316,8 +316,10 @@ RGData::RGData(const RowGroup& rg, uint32_t rowCount)
   RGDataSizeType s = rg.getDataSize(rowCount);
   rowData.reset(new uint8_t[s]);
 
-  if (rg.usesStringTable() && rowCount > 0)
+  if (rg.usesStringTable() && rowCount > 0) {
     strings.reset(new StringStore());
+    strings->useOnlyLongStrings(rg.usesOnlyLongString());
+  }
 
   userDataStore.reset();
   columnCount = rg.getColumnCount();
@@ -329,7 +331,10 @@ RGData::RGData(const RowGroup& rg)
   rowData.reset(new uint8_t[rg.getMaxDataSize()]);
 
   if (rg.usesStringTable())
+  {
     strings.reset(new StringStore());
+    strings->useOnlyLongStrings(rg.usesOnlyLongString());
+  }
 
   userDataStore.reset();
   columnCount = rg.getColumnCount();
@@ -345,6 +350,7 @@ RGData::RGData(const RowGroup& rg, allocators::CountingAllocator<RGDataBufType>&
   {
     allocators::CountingAllocator<StringStoreBufType> ssAlloc = _alloc;
     strings.reset(new StringStore(ssAlloc)); 
+    strings->useOnlyLongStrings(rg.usesOnlyLongString());
   }
 
   userDataStore.reset();
@@ -386,7 +392,7 @@ if (rg.usesStringTable())
 
 void RGData::reinit(const RowGroup& rg)
 {
-  reinit(rg, 8192);
+  reinit(rg, rgCommonSize);
 }
 
 void RGData::serialize(ByteStream& bs, RGDataSizeType amount) const
@@ -1128,6 +1134,7 @@ RowGroup::RowGroup(const RowGroup& r)
  , rgData(r.rgData)
  , strings(r.strings)
  , useStringTable(r.useStringTable)
+ , useOnlyLongStrings(r.useOnlyLongStrings)
  , hasCollation(r.hasCollation)
  , hasLongStringField(r.hasLongStringField)
  , sTableThreshold(r.sTableThreshold)
@@ -1160,6 +1167,7 @@ RowGroup& RowGroup::operator=(const RowGroup& r)
   rgData = r.rgData;
   strings = r.strings;
   useStringTable = r.useStringTable;
+  useOnlyLongStrings = r.useOnlyLongStrings;
   hasCollation = r.hasCollation;
   hasLongStringField = r.hasLongStringField;
   sTableThreshold = r.sTableThreshold;
@@ -1212,6 +1220,7 @@ void RowGroup::serialize(ByteStream& bs) const
   bs << (uint8_t)hasLongStringField;
   bs << sTableThreshold;
   bs.append((uint8_t*)&forceInline[0], sizeof(bool) * columnCount);
+  bs << (uint8_t)useOnlyLongStrings;
 }
 
 void RowGroup::deserialize(ByteStream& bs)
@@ -1238,6 +1247,8 @@ void RowGroup::deserialize(ByteStream& bs)
   forceInline.reset(new bool[columnCount]);
   memcpy(&forceInline[0], bs.buf(), sizeof(bool) * columnCount);
   bs.advance(sizeof(bool) * columnCount);
+  bs >> tmp8;
+  useOnlyLongStrings = (bool)tmp8;
   // offsets = (useStringTable ? &stOffsets[0] : &oldOffsets[0]);
   offsets = 0;
 

utils/rowgroup/rowgroup.h

@@ -63,7 +63,7 @@
 
 namespace rowgroup
 {
-const int16_t rgCommonSize = 8192;
+constexpr int16_t rgCommonSize = 8192;
 using RGDataSizeType = uint64_t;
 
 /*
@@ -172,6 +172,8 @@ class StringStore
   {
     return fUseStoreStringMutex;
   }
+  void useOnlyLongStrings(bool b) { fUseOnlyLongStrings = b; }
+  bool useOnlyLongStrings() const { return fUseOnlyLongStrings; }
 
   // This is an overlay b/c the underlying data needs to be any size,
   // and alloc'd in one chunk.  data can't be a separate dynamic chunk.
@@ -193,6 +195,7 @@ class StringStore
   std::vector<boost::shared_ptr<uint8_t[]>> longStrings;
   bool empty = true;
   bool fUseStoreStringMutex = false;  //@bug6065, make StringStore::storeString() thread safe
+  bool fUseOnlyLongStrings = false;
   boost::mutex fMutex;
   std::optional<allocators::CountingAllocator<StringStoreBufType>> alloc {};
 };
@@ -1556,6 +1559,8 @@ class RowGroup : public messageqcpp::Serializeable
 
   inline bool usesStringTable() const;
   inline void setUseStringTable(bool);
+  void setUseOnlyLongString(bool b) { useOnlyLongStrings = b; }
+  bool usesOnlyLongString() const { return useOnlyLongStrings ; }
 
   bool hasLongString() const
   {
@@ -1628,6 +1633,8 @@ class RowGroup : public messageqcpp::Serializeable
   RGData* rgData = nullptr;
   StringStore* strings = nullptr;  // note, strings and data belong to rgData
   bool useStringTable = true;
+  bool useOnlyLongStrings = false;
+  bool useAggregateDataStore = true;
   bool hasCollation = false;
   bool hasLongStringField = false;
   uint32_t sTableThreshold = 20;