Add support for shrinking lid space in LogDataStore.

5 files changed, 76 insertions, 16 deletions

diff --git a/searchlib/src/tests/docstore/logdatastore/logdatastore_test.cpp b/searchlib/src/tests/docstore/logdatastore/logdatastore_test.cpp
index 54218d5c73f..761d32224f5 100644
--- a/searchlib/src/tests/docstore/logdatastore/logdatastore_test.cpp
+++ b/searchlib/src/tests/docstore/logdatastore/logdatastore_test.cpp
@@ -765,7 +765,7 @@ struct Fixture {
         return serialNum++;
     }
 
-    Fixture(const vespalib::string &dirName,
+    Fixture(const vespalib::string &dirName = "tmp",
             bool dirCleanup = true,
             size_t maxFileSize = 4096 * 2)
         : executor(1, 0x10000),
@@ -803,6 +803,10 @@ struct Fixture {
             }
         }
     }
+    void compactLidSpace(uint32_t wantedDocIdLimit) {
+        store.compactLidSpace(wantedDocIdLimit);
+        assertDocIdLimit(wantedDocIdLimit);
+    }
     void assertDocIdLimit(uint32_t expDocIdLimit) {
         EXPECT_EQUAL(expDocIdLimit, store.getDocIdLimit());
     }
@@ -881,8 +885,7 @@ TEST("require that lid space can be compacted and entries from old files skipped
         TEST_DO(f.assertContent({10,100,101,102,20,200,201,202,30}, 203));
 
         f.assertDocIdLimit(203);
-        f.store.compactLidSpace(100);
-        f.assertDocIdLimit(100);
+        f.compactLidSpace(100);
         TEST_DO(f.assertContent({10,20,30}, 203));
 
         f.writeUntilNewChunk(31);
@@ -898,7 +901,7 @@ TEST("require that lid space can be compacted and entries from old files skipped
     }
 }
 
-TEST_F("require that getLid() is protected by docIdLimit", Fixture("tmp"))
+TEST_F("require that getLid() is protected by docIdLimit", Fixture)
 {
     f.write(1);
     vespalib::GenerationHandler::Guard guard = f.store.getLidReadGuard();
@@ -906,6 +909,45 @@ TEST_F("require that getLid() is protected by docIdLimit", Fixture("tmp"))
     EXPECT_FALSE(f.store.getLid(guard, 2).valid());
 }
 
+TEST_F("require that lid space can be compacted and shrunk", Fixture)
+{
+    f.write(1).write(2);
+    EXPECT_FALSE(f.store.canShrinkLidSpace());
+
+    f.compactLidSpace(2);
+    MemoryUsage before = f.store.getMemoryUsage();
+    EXPECT_TRUE(f.store.canShrinkLidSpace());
+    f.store.shrinkLidSpace();
+
+    MemoryUsage after = f.store.getMemoryUsage();
+    EXPECT_LESS(after.usedBytes(), before.usedBytes());
+}
+
+TEST_F("require that lid space can be increased after being compacted and then shrunk", Fixture)
+{
+    f.write(1).write(3);
+    TEST_DO(f.compactLidSpace(2));
+    f.write(2);
+    TEST_DO(f.assertDocIdLimit(3));
+    f.store.shrinkLidSpace();
+    TEST_DO(f.assertDocIdLimit(3));
+    TEST_DO(f.assertContent({1,2}, 3));
+}
+
+TEST_F("require that lid space can be shrunk only after read guards are deleted", Fixture)
+{
+    f.write(1).write(2);
+    EXPECT_FALSE(f.store.canShrinkLidSpace());
+    {
+        vespalib::GenerationHandler::Guard guard = f.store.getLidReadGuard();
+        f.compactLidSpace(2);
+        f.write(1); // trigger remove of old generations
+        EXPECT_FALSE(f.store.canShrinkLidSpace());
+    }
+    f.write(1); // trigger remove of old generations
+    EXPECT_TRUE(f.store.canShrinkLidSpace());
+}
+
 TEST_MAIN() {
     DummyFileHeaderContext::setCreator("logdatastore_test");
     TEST_RUN_ALL();
diff --git a/searchlib/src/vespa/searchlib/common/rcuvector.h b/searchlib/src/vespa/searchlib/common/rcuvector.h
index bde7fb2accb..aeaec9b1022 100644
--- a/searchlib/src/vespa/searchlib/common/rcuvector.h
+++ b/searchlib/src/vespa/searchlib/common/rcuvector.h
@@ -58,7 +58,7 @@ private:
     }
     void expand(size_t newCapacity);
     void expandAndInsert(const T & v);
-    virtual void onExpand();
+    virtual void onReallocation();
 
 public:
     using ValueType = T;
@@ -133,7 +133,7 @@ private:
     generation_t       _generation;
     GenerationHolder   _genHolderStore;
 
-    void onExpand() override;
+    void onReallocation() override;
 
 public:
     RcuVector();
diff --git a/searchlib/src/vespa/searchlib/common/rcuvector.hpp b/searchlib/src/vespa/searchlib/common/rcuvector.hpp
index c68016dea81..9424331d014 100644
--- a/searchlib/src/vespa/searchlib/common/rcuvector.hpp
+++ b/searchlib/src/vespa/searchlib/common/rcuvector.hpp
@@ -60,7 +60,7 @@ RcuVectorBase<T>::expand(size_t newCapacity) {
     size_t holdSize = tmpData->capacity() * sizeof(T);
     vespalib::GenerationHeldBase::UP hold(new RcuVectorHeld<Array>(holdSize, std::move(tmpData)));
     _genHolder.hold(std::move(hold));
-    onExpand();
+    onReallocation();
 }
 
 template <typename T>
@@ -95,6 +95,7 @@ RcuVectorBase<T>::shrink(size_t newSize)
         size_t holdSize = tmpData->capacity() * sizeof(T);
         vespalib::GenerationHeldBase::UP hold(new RcuVectorHeld<Array>(holdSize, std::move(tmpData)));
         _genHolder.hold(std::move(hold));
+        onReallocation();
     }
 }
 
@@ -144,11 +145,11 @@ RcuVectorBase<T>::getMemoryUsage() const
 
 template <typename T>
 void
-RcuVectorBase<T>::onExpand() { }
+RcuVectorBase<T>::onReallocation() { }
 
 template <typename T>
 void
-RcuVector<T>::onExpand() {
+RcuVector<T>::onReallocation() {
     _genHolderStore.transferHoldLists(_generation);
 }
 
diff --git a/searchlib/src/vespa/searchlib/docstore/logdatastore.cpp b/searchlib/src/vespa/searchlib/docstore/logdatastore.cpp
index 245f0ea912b..159c9882d31 100644
--- a/searchlib/src/vespa/searchlib/docstore/logdatastore.cpp
+++ b/searchlib/src/vespa/searchlib/docstore/logdatastore.cpp
@@ -54,7 +54,8 @@ LogDataStore::LogDataStore(vespalib::ThreadExecutor &executor,
       _initFlushSyncToken(0),
       _tlSyncer(tlSyncer),
       _bucketizer(bucketizer),
-      _currentlyCompacting()
+      _currentlyCompacting(),
+      _compactLidSpaceGeneration()
 {
     // Reserve space for 1TB summary in order to avoid locking.
     _fileChunks.reserve(LidInfo::getFileIdLimit());
@@ -886,15 +887,21 @@ LogDataStore::setLid(const LockGuard &guard, uint32_t lid, const LidInfo &meta)
         }
     } else {
         _lidInfo.ensure_size(lid+1, LidInfo());
-        _lidInfo.setGeneration(_genHandler.getNextGeneration());
-        _genHandler.incGeneration();
-        _genHandler.updateFirstUsedGeneration();
-        _lidInfo.removeOldGenerations(_genHandler.getFirstUsedGeneration());
-        updateDocIdLimit(_lidInfo.size());
+        incGeneration();
     }
+    updateDocIdLimit(lid + 1);
     _lidInfo[lid] = meta;
 }
 
+void
+LogDataStore::incGeneration()
+{
+    _lidInfo.setGeneration(_genHandler.getNextGeneration());
+    _genHandler.incGeneration();
+    _genHandler.updateFirstUsedGeneration();
+    _lidInfo.removeOldGenerations(_genHandler.getFirstUsedGeneration());
+}
+
 size_t
 LogDataStore::computeNumberOfSignificantBucketIdBits(const IBucketizer & bucketizer, FileId fileId) const
 {
@@ -1119,21 +1126,29 @@ void
 LogDataStore::compactLidSpace(uint32_t wantedDocLidLimit)
 {
     LockGuard guard(_updateLock);
+    assert(wantedDocLidLimit <= getDocIdLimit());
     for (size_t i = wantedDocLidLimit; i < _lidInfo.size(); ++i) {
         _lidInfo[i] = LidInfo();
     }
     setDocIdLimit(wantedDocLidLimit);
+    _compactLidSpaceGeneration = _genHandler.getCurrentGeneration();
+    incGeneration();
 }
 
 bool
 LogDataStore::canShrinkLidSpace() const
 {
-    return false;
+    return getDocIdLimit() < _lidInfo.size() &&
+           _compactLidSpaceGeneration < _genHandler.getFirstUsedGeneration();
 }
 
 void
 LogDataStore::shrinkLidSpace()
 {
+    assert(canShrinkLidSpace());
+    LockGuard guard(_updateLock);
+    _lidInfo.shrink(getDocIdLimit());
+    incGeneration();
 }
 
 } // namespace search
diff --git a/searchlib/src/vespa/searchlib/docstore/logdatastore.h b/searchlib/src/vespa/searchlib/docstore/logdatastore.h
index 037cac58fae..c426d9bb46c 100644
--- a/searchlib/src/vespa/searchlib/docstore/logdatastore.h
+++ b/searchlib/src/vespa/searchlib/docstore/logdatastore.h
@@ -304,6 +304,7 @@ private:
     }
     bool shouldCompactToActiveFile(size_t compactedSize) const;
     std::pair<bool, FileId> findNextToCompact();
+    void incGeneration();
 
     typedef std::vector<FileId> FileIdxVector;
     Config                                   _config;
@@ -322,6 +323,7 @@ private:
     transactionlog::SyncProxy               &_tlSyncer;
     IBucketizer::SP                          _bucketizer;
     NameIdSet                                _currentlyCompacting;
+    uint64_t                                 _compactLidSpaceGeneration;
 };
 
 } // namespace search