Add CompactionStrategy methods to determine if compaction should be performed.

7 files changed, 10 insertions, 41 deletions

diff --git a/searchlib/src/tests/attribute/compaction/attribute_compaction_test.cpp b/searchlib/src/tests/attribute/compaction/attribute_compaction_test.cpp
index c3bb9481f0f..a029373821a 100644
--- a/searchlib/src/tests/attribute/compaction/attribute_compaction_test.cpp
+++ b/searchlib/src/tests/attribute/compaction/attribute_compaction_test.cpp
@@ -237,8 +237,7 @@ TEST_F("Compaction limits address space usage (dead) when free lists are NOT use
 {
     populate_and_hammer(f, true);
     AddressSpace afterSpace = f.getMultiValueAddressSpaceUsage("after");
-    // DEAD_ARRAYS_SLACK in multi value mapping is is 64k
-    EXPECT_GREATER(65536u, afterSpace.dead());
+    EXPECT_GREATER(search::CompactionStrategy::DEAD_ADDRESS_SPACE_SLACK, afterSpace.dead());
 }
 
 TEST_F("Compaction is not executed when free lists are used",
@@ -267,8 +266,7 @@ TEST_F("Compaction is peformed when compaction strategy is changed to enable com
         f._v->commit(); // new commit might trigger further compaction
         after2 = f.getMultiValueAddressSpaceUsage("after2");
     }
-    // DEAD_ARRAYS_SLACK in multi value mapping is is 64k
-    EXPECT_GREATER(65536u, after2.dead());
+    EXPECT_GREATER(search::CompactionStrategy::DEAD_ADDRESS_SPACE_SLACK, after2.dead());
 }
 
 TEST_MAIN() { TEST_RUN_ALL(); }
diff --git a/searchlib/src/tests/attribute/enum_attribute_compaction/enum_attribute_compaction_test.cpp b/searchlib/src/tests/attribute/enum_attribute_compaction/enum_attribute_compaction_test.cpp
index bd1b63c01d8..2faaa371fca 100644
--- a/searchlib/src/tests/attribute/enum_attribute_compaction/enum_attribute_compaction_test.cpp
+++ b/searchlib/src/tests/attribute/enum_attribute_compaction/enum_attribute_compaction_test.cpp
@@ -174,9 +174,8 @@ template <typename VectorType>
 void
 CompactionTest<VectorType>::test_enum_store_compaction()
 {
-    constexpr size_t DEAD_BYTES_SLACK = 0x10000u;
     constexpr uint32_t canary_stride = 256;
-    uint32_t dead_limit = DEAD_BYTES_SLACK / 8;
+    uint32_t dead_limit = search::CompactionStrategy::DEAD_BYTES_SLACK / 8;
     uint32_t doc_count = dead_limit * 3;
     if (_v->hasMultiValue() || std::is_same_v<VectorType,StringAttribute>) {
         doc_count /= 2;
diff --git a/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp b/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp
index a8a34b0ba8a..f45c3e4a1a0 100644
--- a/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp
+++ b/searchlib/src/tests/attribute/reference_attribute/reference_attribute_test.cpp
@@ -45,8 +45,6 @@ vespalib::string doc1("id:test:music::1");
 vespalib::string doc2("id:test:music::2");
 vespalib::string doc3("id:test:music::3");
 
-constexpr size_t DEAD_BYTES_SLACK = 0x10000u;
-
 }
 
 struct MyGidToLidMapperFactory : public search::attribute::test::MockGidToLidMapperFactory {
@@ -178,7 +176,7 @@ struct ReferenceAttributeTest : public ::testing::Test {
         search::attribute::Status newStatus = oldStatus;
         uint64_t iter = 0;
         AttributeGuard guard(_attr);
-        uint64_t dropCount = DEAD_BYTES_SLACK / sizeof(Reference);
+        uint64_t dropCount = search::CompactionStrategy::DEAD_BYTES_SLACK / sizeof(Reference);
         for (; iter < iterLimit; ++iter) {
             clear(2);
             set(2, toGid(doc2));
diff --git a/searchlib/src/vespa/searchlib/attribute/enumstore.hpp b/searchlib/src/vespa/searchlib/attribute/enumstore.hpp
index de68e7f044a..99d9a7fdfee 100644
--- a/searchlib/src/vespa/searchlib/attribute/enumstore.hpp
+++ b/searchlib/src/vespa/searchlib/attribute/enumstore.hpp
@@ -219,13 +219,6 @@ EnumStoreT<EntryT>::update_stat()
     return retval;
 }
 
-namespace {
-
-// minimum dead bytes in enum store before consider compaction
-constexpr size_t DEAD_BYTES_SLACK = 0x10000u;
-constexpr size_t DEAD_ADDRESS_SPACE_SLACK = 0x10000u;
-
-}
 template <typename EntryT>
 std::unique_ptr<IEnumStore::EnumIndexRemapper>
 EnumStoreT<EntryT>::consider_compact(const CompactionStrategy& compaction_strategy)
@@ -234,10 +227,8 @@ EnumStoreT<EntryT>::consider_compact(const CompactionStrategy& compaction_strate
     size_t dead_bytes = _cached_values_memory_usage.deadBytes();
     size_t used_address_space = _cached_values_address_space_usage.used();
     size_t dead_address_space = _cached_values_address_space_usage.dead();
-    bool compact_memory = ((dead_bytes >= DEAD_BYTES_SLACK) &&
-                           (used_bytes * compaction_strategy.getMaxDeadBytesRatio() < dead_bytes));
-    bool compact_address_space = ((dead_address_space >= DEAD_ADDRESS_SPACE_SLACK) &&
-                                  (used_address_space * compaction_strategy.getMaxDeadAddressSpaceRatio() < dead_address_space));
+    bool compact_memory = compaction_strategy.should_compact_memory(used_bytes, dead_bytes);
+    bool compact_address_space = compaction_strategy.should_compact_address_space(used_address_space, dead_address_space);
     if (compact_memory || compact_address_space) {
         return compact_worst(compact_memory, compact_address_space);
     }
diff --git a/searchlib/src/vespa/searchlib/attribute/multi_value_mapping_base.cpp b/searchlib/src/vespa/searchlib/attribute/multi_value_mapping_base.cpp
index b8378b059f6..c57c18e1b0e 100644
--- a/searchlib/src/vespa/searchlib/attribute/multi_value_mapping_base.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/multi_value_mapping_base.cpp
@@ -6,14 +6,6 @@
 
 namespace search::attribute {
 
-namespace {
-
-// minimum dead bytes in multi value mapping before consider compaction
-constexpr size_t DEAD_BYTES_SLACK = 0x10000u;
-constexpr size_t DEAD_ARRAYS_SLACK = 0x10000u;
-
-}
-
 MultiValueMappingBase::MultiValueMappingBase(const vespalib::GrowStrategy &gs,
                                              vespalib::GenerationHolder &genHolder)
     : _indices(gs, genHolder),
@@ -89,10 +81,8 @@ MultiValueMappingBase::considerCompact(const CompactionStrategy &compactionStrat
     size_t deadBytes = _cachedArrayStoreMemoryUsage.deadBytes();
     size_t usedArrays = _cachedArrayStoreAddressSpaceUsage.used();
     size_t deadArrays = _cachedArrayStoreAddressSpaceUsage.dead();
-    bool compactMemory = ((deadBytes >= DEAD_BYTES_SLACK) &&
-                          (usedBytes * compactionStrategy.getMaxDeadBytesRatio() < deadBytes));
-    bool compactAddressSpace = ((deadArrays >= DEAD_ARRAYS_SLACK) &&
-                                (usedArrays * compactionStrategy.getMaxDeadAddressSpaceRatio() < deadArrays));
+    bool compactMemory = compactionStrategy.should_compact_memory(usedBytes, deadBytes);
+    bool compactAddressSpace = compactionStrategy.should_compact_address_space(usedArrays, deadArrays);
     if (compactMemory || compactAddressSpace) {
         compactWorst(compactMemory, compactAddressSpace);
         return true;
diff --git a/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp b/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp
index 2af5bfdf225..f9faed34d90 100644
--- a/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp
+++ b/searchlib/src/vespa/searchlib/attribute/reference_attribute.cpp
@@ -27,9 +27,6 @@ using document::IdParseException;
 
 namespace {
 
-// minimum dead bytes in unique store before consider compaction
-constexpr size_t DEAD_BYTES_SLACK = 0x10000u;
-
 const vespalib::string uniqueValueCountTag = "uniqueValueCount";
 
 uint64_t
@@ -289,8 +286,7 @@ ReferenceAttribute::considerCompact(const CompactionStrategy &compactionStrategy
 {
     size_t usedBytes = _cached_unique_store_values_memory_usage.usedBytes();
     size_t deadBytes = _cached_unique_store_values_memory_usage.deadBytes();
-    bool compactMemory = ((deadBytes >= DEAD_BYTES_SLACK) &&
-                          (usedBytes * compactionStrategy.getMaxDeadBytesRatio() < deadBytes));
+    bool compactMemory = compactionStrategy.should_compact_memory(usedBytes, deadBytes);
     if (compactMemory) {
         compactWorst();
         return true;
diff --git a/searchlib/src/vespa/searchlib/tensor/tensor_attribute.cpp b/searchlib/src/vespa/searchlib/tensor/tensor_attribute.cpp
index 042ecba0901..e8a25e0e23e 100644
--- a/searchlib/src/vespa/searchlib/tensor/tensor_attribute.cpp
+++ b/searchlib/src/vespa/searchlib/tensor/tensor_attribute.cpp
@@ -27,9 +27,6 @@ namespace {
 
 constexpr uint32_t TENSOR_ATTRIBUTE_VERSION = 0;
 
-// minimum dead bytes in tensor attribute before consider compaction
-constexpr size_t DEAD_SLACK = 0x10000u;
-
 Value::UP
 createEmptyTensor(const ValueType &type)
 {
@@ -90,7 +87,7 @@ TensorAttribute::onCommit()
         Status &status = getStatus();
         size_t used = status.getUsed();
         size_t dead = status.getDead();
-        if ((dead >= DEAD_SLACK) && (dead * 5 > used)) {
+        if (getConfig().getCompactionStrategy().should_compact_memory(used, dead)) {
             compactWorst();
         }
     }