From be5711cf0debaa44debf6224e82295e065a07cb9 Mon Sep 17 00:00:00 2001
From: drrtuy <drrtuy@gmail.com>
Date: Tue, 4 Mar 2025 21:10:39 +0000
Subject: [PATCH] feat(): replace getMaxDataSize with getMaxDataSizeWithStrings
 to accurately account for mem

---
 dbcon/joblist/diskjoinstep.cpp | 16 +++++++++-------
 dbcon/joblist/tuple-bps.cpp    |  5 +++--
 utils/rowgroup/rowstorage.cpp  |  2 +-
 3 files changed, 13 insertions(+), 10 deletions(-)

diff --git a/dbcon/joblist/diskjoinstep.cpp b/dbcon/joblist/diskjoinstep.cpp
index 86e6fa886..3bcd6945f 100644
--- a/dbcon/joblist/diskjoinstep.cpp
+++ b/dbcon/joblist/diskjoinstep.cpp
@@ -519,16 +519,18 @@ void DiskJoinStep::joinFcn(const uint32_t threadID)
               // cout << "inserting a full RG" << endl;
               if (thjs)
               {
-                // FIXME: Possible false positive. Something wrong with this calculation, just put a warning
-                // until fixed.
-                if (!thjs->getMemory(l_outputRG.getMaxDataSize()))
+                if (!thjs->getMemory(l_outputRG.getMaxDataSizeWithStrings()))
                 {
                   // FIXME: This is also looks wrong.
                   // calculate guess of size required for error message
-                  uint64_t memReqd = (unmatched.size() * outputRG.getDataSize(1)) / 1048576;
-                  uint64_t memLimit = thjs->resourceManager->getConfiguredUMMemLimit() / 1048576;
-                  std::cerr << "DiskJoin::joinFcn() possible OOM for the join result, mem required: "
-                            << memReqd << " mem limit: " << memLimit << std::endl;
+                  uint64_t memReqd = (l_outputRG.getMaxDataSizeWithStrings()) / 1048576;
+                  Message::Args args;
+                  args.add(memReqd);
+                  args.add(thjs->resourceManager->getConfiguredUMMemLimit() / 1048576);
+                  std::cerr << logging::IDBErrorInfo::instance()->errorMsg(logging::ERR_JOIN_RESULT_TOO_BIG,
+                                                                           args)
+                            << " @" << __FILE__ << ":" << __LINE__;
+                  throw logging::IDBExcept(logging::ERR_JOIN_RESULT_TOO_BIG, args);
                 }
               }
 
diff --git a/dbcon/joblist/tuple-bps.cpp b/dbcon/joblist/tuple-bps.cpp
index 5696a1376..e12f250d3 100644
--- a/dbcon/joblist/tuple-bps.cpp
+++ b/dbcon/joblist/tuple-bps.cpp
@@ -275,9 +275,10 @@ uint64_t TupleBPS::JoinLocalData::generateJoinResultSet(const uint32_t depth,
         uint64_t baseRid = local_outputRG.getBaseRid();
         outputData.push_back(joinedData);
         // Don't let the join results buffer get out of control.
-        if (tbps->resourceManager()->getMemory(local_outputRG.getMaxDataSize(), false))
+        auto outputDataSize = local_outputRG.getMaxDataSizeWithStrings();
+        if (tbps->resourceManager()->getMemory(outputDataSize, false))
         {
-          memSizeForOutputRG += local_outputRG.getMaxDataSize();
+          memSizeForOutputRG += outputDataSize;
         }
         else
         {
diff --git a/utils/rowgroup/rowstorage.cpp b/utils/rowgroup/rowstorage.cpp
index ec5699ada..bec3faa65 100644
--- a/utils/rowgroup/rowstorage.cpp
+++ b/utils/rowgroup/rowstorage.cpp
@@ -1361,7 +1361,7 @@ class RowGroupStorage
   {
     messageqcpp::ByteStream bs;
     fRowGroupOut->setData(rgdata);
-    rgdata->serialize(bs, fRowGroupOut->getDataSize());
+    rgdata->serialize(bs, fRowGroupOut->getSizeWithStrings());
 
     int errNo;
     if ((errNo = fDumper->write(makeRGFilename(rgid), (char*)bs.buf(), bs.length())) != 0)