Merge pull request #20740 from vespa-engine/geirst/prepare-sampling-of-current-transient-disk-usage

Prepare for tracking the current transient disk usage across all docu…

15 files changed, 97 insertions, 81 deletions

diff --git a/searchcore/src/tests/proton/server/disk_mem_usage_filter/disk_mem_usage_filter_test.cpp b/searchcore/src/tests/proton/server/disk_mem_usage_filter/disk_mem_usage_filter_test.cpp
index 072b76c6443..ce85517ee09 100644
--- a/searchcore/src/tests/proton/server/disk_mem_usage_filter/disk_mem_usage_filter_test.cpp
+++ b/searchcore/src/tests/proton/server/disk_mem_usage_filter/disk_mem_usage_filter_test.cpp
@@ -121,9 +121,16 @@ TEST_F(DiskMemUsageFilterTest, both_disk_limit_and_memory_limit_can_be_reached)
               "capacity: 100, used: 90, diskUsed: 0.9, diskLimit: 0.8}}");
 }
 
+TEST_F(DiskMemUsageFilterTest, transient_disk_usage_is_tracked_in_usage_state_and_metrics)
+{
+    _filter.set_transient_resource_usage({40, 0});
+    EXPECT_EQ(0.4, _filter.usageState().transient_disk_usage());
+    EXPECT_EQ(0.4, _filter.get_metrics().get_transient_disk_usage());
+}
+
 TEST_F(DiskMemUsageFilterTest, transient_memory_usage_is_tracked_in_usage_state_and_metrics)
 {
-    _filter.set_transient_resource_usage(200, 0);
+    _filter.set_transient_resource_usage({0, 200});
     EXPECT_EQ(0.2, _filter.usageState().transient_memory_usage());
     EXPECT_EQ(0.2, _filter.get_metrics().get_transient_memory_usage());
 }
diff --git a/searchcore/src/tests/proton/server/disk_mem_usage_sampler/disk_mem_usage_sampler_test.cpp b/searchcore/src/tests/proton/server/disk_mem_usage_sampler/disk_mem_usage_sampler_test.cpp
index 7fa51bb8b95..0c80553e1e7 100644
--- a/searchcore/src/tests/proton/server/disk_mem_usage_sampler/disk_mem_usage_sampler_test.cpp
+++ b/searchcore/src/tests/proton/server/disk_mem_usage_sampler/disk_mem_usage_sampler_test.cpp
@@ -34,8 +34,7 @@ public:
         : _memory_usage(memory_usage),
           _disk_usage(disk_usage)
     {}
-    size_t get_transient_memory_usage() const override { return _memory_usage; }
-    size_t get_transient_disk_usage() const override { return _disk_usage; }
+    TransientResourceUsage get_transient_resource_usage() const override { return {_disk_usage, _memory_usage}; }
 };
 
 struct DiskMemUsageSamplerTest : public ::testing::Test {
@@ -46,7 +45,7 @@ struct DiskMemUsageSamplerTest : public ::testing::Test {
                                             50ms, make_hw_info()))
     {
         sampler.add_transient_usage_provider(std::make_shared<MyProvider>(50, 200));
-        sampler.add_transient_usage_provider(std::make_shared<MyProvider>(100, 199));
+        sampler.add_transient_usage_provider(std::make_shared<MyProvider>(100, 150));
     }
     const DiskMemUsageFilter& filter() const { return sampler.writeFilter(); }
 };
@@ -56,7 +55,7 @@ TEST_F(DiskMemUsageSamplerTest, resource_usage_is_sampled)
     // Poll for up to 20 seconds to get a sample.
     size_t i = 0;
     for (; i < static_cast<size_t>(20s / 50ms); ++i) {
-        if (filter().get_transient_memory_usage() > 0) {
+        if (filter().get_transient_resource_usage().memory() > 0) {
             break;
         }
         std::this_thread::sleep_for(50ms);
@@ -70,10 +69,10 @@ TEST_F(DiskMemUsageSamplerTest, resource_usage_is_sampled)
     EXPECT_EQ(filter().getMemoryStats().getAnonymousRss(), 0);
 #endif
     EXPECT_GT(filter().getDiskUsedSize(), 0);
-    EXPECT_EQ(150, filter().get_transient_memory_usage());
+    EXPECT_EQ(150, filter().get_transient_resource_usage().memory());
     EXPECT_EQ(150.0 / memory_size_bytes, filter().usageState().transient_memory_usage());
-    EXPECT_EQ(200, filter().get_transient_disk_usage());
-    EXPECT_EQ(200.0 / disk_size_bytes, filter().get_relative_transient_disk_usage());
+    EXPECT_EQ(350, filter().get_transient_resource_usage().disk());
+    EXPECT_EQ(350.0 / disk_size_bytes, filter().usageState().transient_disk_usage());
 }
 
 GTEST_MAIN_RUN_ALL_TESTS()
diff --git a/searchcore/src/vespa/searchcore/proton/common/i_transient_resource_usage_provider.h b/searchcore/src/vespa/searchcore/proton/common/i_transient_resource_usage_provider.h
index 5aeb84feb2c..833b6519046 100644
--- a/searchcore/src/vespa/searchcore/proton/common/i_transient_resource_usage_provider.h
+++ b/searchcore/src/vespa/searchcore/proton/common/i_transient_resource_usage_provider.h
@@ -7,6 +7,32 @@
 namespace proton {
 
 /**
+ * Class containing transient disk and memory usage (in bytes).
+ */
+class TransientResourceUsage {
+private:
+    size_t _disk;
+    size_t _memory;
+
+public:
+    TransientResourceUsage() noexcept
+        : _disk(0),
+          _memory(0)
+    {}
+    TransientResourceUsage(size_t disk_in,
+                           size_t memory_in) noexcept
+        : _disk(disk_in),
+          _memory(memory_in)
+    {}
+    size_t disk() const noexcept { return _disk; }
+    size_t memory() const noexcept { return _memory; }
+    void merge(const TransientResourceUsage& rhs) {
+        _disk += rhs.disk();
+        _memory += rhs.memory();
+    }
+};
+
+/**
  * Interface class providing a snapshot of transient resource usage.
  *
  * E.g. the memory used by the memory index and extra disk needed for running disk index fusion.
@@ -15,8 +41,7 @@ namespace proton {
 class ITransientResourceUsageProvider {
 public:
     virtual ~ITransientResourceUsageProvider() = default;
-    virtual size_t get_transient_memory_usage() const = 0;
-    virtual size_t get_transient_disk_usage() const = 0;
+    virtual TransientResourceUsage get_transient_resource_usage() const = 0;
 };
 
 }
diff --git a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.cpp b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.cpp
index 27eb636b475..8928045b814 100644
--- a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.cpp
+++ b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.cpp
@@ -131,6 +131,7 @@ DiskMemUsageFilter::recalcState(const Guard &guard)
     }
     DiskMemUsageState dmstate(ResourceUsageState(_config._diskLimit, diskUsed),
                               ResourceUsageState(_config._memoryLimit, memoryUsed),
+                              get_relative_transient_disk_usage(guard),
                               get_relative_transient_memory_usage(guard));
     notifyDiskMemUsage(guard, dmstate);
 }
@@ -153,9 +154,14 @@ DiskMemUsageFilter::getDiskUsedRatio(const Guard&) const
 double
 DiskMemUsageFilter::get_relative_transient_memory_usage(const Guard&) const
 {
-    return  static_cast<double>(_transient_memory_usage) / _hwInfo.memory().sizeBytes();
+    return  static_cast<double>(_transient_usage.memory()) / _hwInfo.memory().sizeBytes();
 }
 
+double
+DiskMemUsageFilter::get_relative_transient_disk_usage(const Guard&) const
+{
+    return  static_cast<double>(_transient_usage.disk()) / _hwInfo.disk().sizeBytes();
+}
 
 DiskMemUsageFilter::DiskMemUsageFilter(const HwInfo &hwInfo)
     : _lock(),
@@ -163,7 +169,7 @@ DiskMemUsageFilter::DiskMemUsageFilter(const HwInfo &hwInfo)
       _acceptWrite(true),
       _memoryStats(),
       _diskUsedSizeBytes(),
-      _transient_memory_usage(0u),
+      _transient_usage(),
       _config(),
       _state(),
       _dmstate(),
@@ -190,11 +196,10 @@ DiskMemUsageFilter::setDiskUsedSize(uint64_t diskUsedSizeBytes)
 }
 
 void
-DiskMemUsageFilter::set_transient_resource_usage(size_t transient_memory_usage, size_t transient_disk_usage)
+DiskMemUsageFilter::set_transient_resource_usage(const TransientResourceUsage& transient_usage)
 {
     Guard guard(_lock);
-    _transient_memory_usage = transient_memory_usage;
-    _transient_disk_usage = transient_disk_usage;
+    _transient_usage = transient_usage;
     recalcState(guard);
 }
 
@@ -220,25 +225,11 @@ DiskMemUsageFilter::getDiskUsedSize() const
     return _diskUsedSizeBytes;
 }
 
-size_t
-DiskMemUsageFilter::get_transient_memory_usage() const
-{
-    Guard guard(_lock);
-    return _transient_memory_usage;
-}
-
-size_t
-DiskMemUsageFilter::get_transient_disk_usage() const
-{
-    Guard guard(_lock);
-    return _transient_disk_usage;
-}
-
-double
-DiskMemUsageFilter::get_relative_transient_disk_usage() const
+TransientResourceUsage
+DiskMemUsageFilter::get_transient_resource_usage() const
 {
     Guard guard(_lock);
-    return  static_cast<double>(_transient_disk_usage) / _hwInfo.disk().sizeBytes();
+    return _transient_usage;
 }
 
 DiskMemUsageFilter::Config
diff --git a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.h b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.h
index fb18a12084f..ac328c3ddea 100644
--- a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.h
+++ b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_filter.h
@@ -6,6 +6,7 @@
 #include "disk_mem_usage_state.h"
 #include "disk_mem_usage_metrics.h"
 #include <vespa/searchcore/proton/common/hw_info.h>
+#include <vespa/searchcore/proton/common/i_transient_resource_usage_provider.h>
 #include <vespa/searchcore/proton/persistenceengine/i_resource_write_filter.h>
 #include <vespa/vespalib/util/process_memory_stats.h>
 #include <atomic>
@@ -47,8 +48,7 @@ private:
     // Following member variables are protected by _lock
     vespalib::ProcessMemoryStats _memoryStats;
     uint64_t                     _diskUsedSizeBytes;
-    size_t                       _transient_memory_usage;
-    size_t                       _transient_disk_usage;
+    TransientResourceUsage       _transient_usage;
     Config                       _config;
     State                        _state;
     DiskMemUsageState            _dmstate;
@@ -59,6 +59,7 @@ private:
     double getMemoryUsedRatio(const Guard &guard) const;
     double getDiskUsedRatio(const Guard &guard) const;
     double get_relative_transient_memory_usage(const Guard& guard) const;
+    double get_relative_transient_disk_usage(const Guard& guard) const;
     void notifyDiskMemUsage(const Guard &guard, DiskMemUsageState state);
 
 public:
@@ -66,13 +67,11 @@ public:
     ~DiskMemUsageFilter() override;
     void setMemoryStats(vespalib::ProcessMemoryStats memoryStats_in);
     void setDiskUsedSize(uint64_t diskUsedSizeBytes);
-    void set_transient_resource_usage(size_t transient_memory_usage, size_t transient_disk_usage);
+    void set_transient_resource_usage(const TransientResourceUsage& transient_usage);
     void setConfig(Config config);
     vespalib::ProcessMemoryStats getMemoryStats() const;
     uint64_t getDiskUsedSize() const;
-    size_t get_transient_memory_usage() const;
-    size_t get_transient_disk_usage() const;
-    double get_relative_transient_disk_usage() const;
+    TransientResourceUsage get_transient_resource_usage() const;
     Config getConfig() const;
     const HwInfo &getHwInfo() const { return _hwInfo; }
     DiskMemUsageState usageState() const;
diff --git a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.cpp b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.cpp
index 562e20a864c..230593c2c1d 100644
--- a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.cpp
+++ b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.cpp
@@ -14,6 +14,7 @@ DiskMemUsageMetrics::DiskMemUsageMetrics() noexcept
 DiskMemUsageMetrics::DiskMemUsageMetrics(const DiskMemUsageState &usage_state) noexcept
     : _disk_usage(usage_state.diskState().usage()),
       _disk_utilization(usage_state.diskState().utilization()),
+      _transient_disk_usage(usage_state.transient_disk_usage()),
       _memory_usage(usage_state.memoryState().usage()),
       _memory_utilization(usage_state.memoryState().utilization()),
       _transient_memory_usage(usage_state.transient_memory_usage())
@@ -25,6 +26,7 @@ DiskMemUsageMetrics::merge(const DiskMemUsageState &usage_state) noexcept
 {
     _disk_usage = std::max(_disk_usage, usage_state.diskState().usage());
     _disk_utilization = std::max(_disk_utilization, usage_state.diskState().utilization());
+    _transient_disk_usage = std::max(_transient_disk_usage, usage_state.transient_disk_usage());
     _memory_usage = std::max(_memory_usage, usage_state.memoryState().usage());
     _memory_utilization = std::max(_memory_utilization, usage_state.memoryState().utilization());
     _transient_memory_usage = std::max(_transient_memory_usage, usage_state.transient_memory_usage());
diff --git a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.h b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.h
index cb084b90a11..cb97eb4c891 100644
--- a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.h
+++ b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_metrics.h
@@ -14,6 +14,7 @@ class DiskMemUsageMetrics
 {
     double _disk_usage;
     double _disk_utilization;
+    double _transient_disk_usage;
     double _memory_usage;
     double _memory_utilization;
     double _transient_memory_usage;
@@ -24,6 +25,7 @@ public:
     void merge(const DiskMemUsageState &usage_state) noexcept;
     double get_disk_usage() const noexcept { return _disk_usage; }
     double get_disk_utilization() const noexcept { return _disk_utilization; }
+    double get_transient_disk_usage() const noexcept { return _transient_disk_usage; }
     double get_memory_usage() const noexcept { return _memory_usage; }
     double get_memory_utilization() const noexcept { return _memory_utilization; }
     double get_transient_memory_usage() const noexcept { return _transient_memory_usage; }
diff --git a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_sampler.cpp b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_sampler.cpp
index cfed4c0522a..2dc749ce26b 100644
--- a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_sampler.cpp
+++ b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_sampler.cpp
@@ -125,16 +125,14 @@ DiskMemUsageSampler::sampleMemoryUsage()
 void
 DiskMemUsageSampler::sample_transient_resource_usage()
 {
-    size_t transient_memory_usage_sum = 0;
-    size_t max_transient_disk_usage = 0;
+    TransientResourceUsage transient_usage;
     {
         std::lock_guard<std::mutex> guard(_lock);
         for (auto provider : _transient_usage_providers) {
-            transient_memory_usage_sum += provider->get_transient_memory_usage();
-            max_transient_disk_usage = std::max(max_transient_disk_usage, provider->get_transient_disk_usage());
+            transient_usage.merge(provider->get_transient_resource_usage());
         }
     }
-    _filter.set_transient_resource_usage(transient_memory_usage_sum, max_transient_disk_usage);
+    _filter.set_transient_resource_usage(transient_usage);
 }
 
 void
diff --git a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_state.h b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_state.h
index 769086574f0..b205b441bcf 100644
--- a/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_state.h
+++ b/searchcore/src/vespa/searchcore/proton/server/disk_mem_usage_state.h
@@ -8,27 +8,31 @@ namespace proton {
 
 /**
  * Class used to describe state of disk and memory usage relative to configured limits.
- * In addition relative transient memory usage is tracked.
+ * In addition relative transient disk and memory usage are tracked.
  */
 class DiskMemUsageState
 {
     ResourceUsageState _diskState;
     ResourceUsageState _memoryState;
+    double _transient_disk_usage;
     double _transient_memory_usage;
 
 public:
     DiskMemUsageState() = default;
     DiskMemUsageState(const ResourceUsageState &diskState_,
                       const ResourceUsageState &memoryState_,
+                      double transient_disk_usage_ = 0,
                       double transient_memory_usage_ = 0)
         : _diskState(diskState_),
           _memoryState(memoryState_),
+          _transient_disk_usage(transient_disk_usage_),
           _transient_memory_usage(transient_memory_usage_)
     {
     }
     bool operator==(const DiskMemUsageState &rhs) const {
         return ((_diskState == rhs._diskState) &&
                 (_memoryState == rhs._memoryState) &&
+                (_transient_disk_usage == rhs._transient_disk_usage) &&
                 (_transient_memory_usage == rhs._transient_memory_usage));
     }
     bool operator!=(const DiskMemUsageState &rhs) const {
@@ -36,6 +40,7 @@ public:
     }
     const ResourceUsageState &diskState() const { return _diskState; }
     const ResourceUsageState &memoryState() const { return _memoryState; }
+    double transient_disk_usage() const { return _transient_disk_usage; }
     double transient_memory_usage() const { return _transient_memory_usage; }
     bool aboveDiskLimit(double resourceLimitFactor) const { return diskState().aboveLimit(resourceLimitFactor); }
     bool aboveMemoryLimit(double resourceLimitFactor) const { return memoryState().aboveLimit(resourceLimitFactor); }
diff --git a/searchcore/src/vespa/searchcore/proton/server/documentdb.cpp b/searchcore/src/vespa/searchcore/proton/server/documentdb.cpp
index b9903d626aa..f052d663ba6 100644
--- a/searchcore/src/vespa/searchcore/proton/server/documentdb.cpp
+++ b/searchcore/src/vespa/searchcore/proton/server/documentdb.cpp
@@ -97,14 +97,11 @@ public:
     DocumentDBResourceUsageProvider(const DocumentDB& doc_db) noexcept
         : _doc_db(doc_db)
     {}
-    size_t get_transient_memory_usage() const override {
-        return _doc_db.getReadySubDB()->getSearchableStats().memoryUsage().allocatedBytes();
-    }
-    size_t get_transient_disk_usage() const override {
-        // We estimate the transient disk usage for the next disk index fusion
-        // as the size of the largest disk index.
-        // TODO: Change this to actually measure the size of the fusion disk index(es).
-        return _doc_db.getReadySubDB()->getSearchableStats().max_component_size_on_disk();
+    TransientResourceUsage get_transient_resource_usage() const override {
+        // Transient disk usage is measured as the total disk usage of all current fusion indexes.
+        // Transient memory usage is measured as the total memory usage of all memory indexes.
+        auto stats = _doc_db.getReadySubDB()->getSearchableStats();
+        return {stats.fusion_size_on_disk(), stats.memoryUsage().allocatedBytes()};
     }
 };
 
diff --git a/searchcore/src/vespa/searchcore/proton/server/proton.cpp b/searchcore/src/vespa/searchcore/proton/server/proton.cpp
index 7c52d3d96ff..b128fe16e5e 100644
--- a/searchcore/src/vespa/searchcore/proton/server/proton.cpp
+++ b/searchcore/src/vespa/searchcore/proton/server/proton.cpp
@@ -758,7 +758,7 @@ Proton::updateMetrics(const metrics::MetricLockGuard &)
         metrics.resourceUsage.memory.set(dm_metrics.get_memory_usage());
         metrics.resourceUsage.memoryUtilization.set(dm_metrics.get_memory_utilization());
         metrics.resourceUsage.transient_memory.set(dm_metrics.get_transient_memory_usage());
-        metrics.resourceUsage.transient_disk.set(usageFilter.get_relative_transient_disk_usage());
+        metrics.resourceUsage.transient_disk.set(dm_metrics.get_transient_disk_usage());
         metrics.resourceUsage.memoryMappings.set(usageFilter.getMemoryStats().getMappingsCount());
         metrics.resourceUsage.openFileDescriptors.set(FastOS_File::count_open_files());
         metrics.resourceUsage.feedingBlocked.set((usageFilter.acceptWriteOperation() ? 0.0 : 1.0));
diff --git a/searchcore/src/vespa/searchcore/proton/server/resource_usage_explorer.cpp b/searchcore/src/vespa/searchcore/proton/server/resource_usage_explorer.cpp
index 4b5e97a827b..3750c5afb5f 100644
--- a/searchcore/src/vespa/searchcore/proton/server/resource_usage_explorer.cpp
+++ b/searchcore/src/vespa/searchcore/proton/server/resource_usage_explorer.cpp
@@ -44,16 +44,16 @@ ResourceUsageExplorer::get_state(const vespalib::slime::Inserter &inserter, bool
         disk.setDouble("usage", usageState.diskState().usage());
         disk.setDouble("limit", usageState.diskState().limit());
         disk.setDouble("utilization", usageState.diskState().utilization());
+        disk.setDouble("transient", usageState.transient_disk_usage());
         convertDiskStatsToSlime(_usage_filter.getHwInfo(), _usage_filter.getDiskUsedSize(), disk.setObject("stats"));
 
         Cursor &memory = object.setObject("memory");
         memory.setDouble("usage", usageState.memoryState().usage());
         memory.setDouble("limit", usageState.memoryState().limit());
         memory.setDouble("utilization", usageState.memoryState().utilization());
+        memory.setDouble("transient", usageState.transient_memory_usage());
         memory.setLong("physicalMemory", _usage_filter.getHwInfo().memory().sizeBytes());
         convertMemoryStatsToSlime(_usage_filter.getMemoryStats(), memory.setObject("stats"));
-        size_t transient_memory = _usage_filter.get_transient_memory_usage();
-        memory.setLong("transient", transient_memory);
 
         Cursor &address_space = object.setObject("attribute_address_space");
         address_space.setDouble("usage", attr_usage.get_usage());
diff --git a/searchcorespi/src/vespa/searchcorespi/index/indexmaintainer.cpp b/searchcorespi/src/vespa/searchcorespi/index/indexmaintainer.cpp
index a2bd19c3d29..6489cb156ce 100644
--- a/searchcorespi/src/vespa/searchcorespi/index/indexmaintainer.cpp
+++ b/searchcorespi/src/vespa/searchcorespi/index/indexmaintainer.cpp
@@ -117,6 +117,7 @@ public:
     {
         return _index->createBlueprint(requestContext, fields, term);
     }
+    // TODO: Calculate the total disk size of current fusion indexes and set fusion_size_on_disk().
     search::SearchableStats getSearchableStats() const override { return _index->getSearchableStats(); }
     search::SerialNum getSerialNum() const override {
         return _index->getSerialNum();
diff --git a/searchlib/src/tests/util/searchable_stats/searchable_stats_test.cpp b/searchlib/src/tests/util/searchable_stats/searchable_stats_test.cpp
index b89324c71f5..ed857d5776b 100644
--- a/searchlib/src/tests/util/searchable_stats/searchable_stats_test.cpp
+++ b/searchlib/src/tests/util/searchable_stats/searchable_stats_test.cpp
@@ -7,43 +7,35 @@ LOG_SETUP("searchable_stats_test");
 
 using namespace search;
 
-TEST(SearchableStatsTest, merge_also_tracks_max_size_on_disk_for_component)
+TEST(SearchableStatsTest, stats_can_be_merged)
 {
     SearchableStats stats;
     EXPECT_EQ(0u, stats.memoryUsage().allocatedBytes());
     EXPECT_EQ(0u, stats.docsInMemory());
     EXPECT_EQ(0u, stats.sizeOnDisk());
-    EXPECT_EQ(0u, stats.max_component_size_on_disk());
+    EXPECT_EQ(0u, stats.fusion_size_on_disk());
     {
         SearchableStats rhs;
         EXPECT_EQ(&rhs.memoryUsage(vespalib::MemoryUsage(100,0,0,0)), &rhs);
         EXPECT_EQ(&rhs.docsInMemory(10), &rhs);
         EXPECT_EQ(&rhs.sizeOnDisk(1000), &rhs);
-        EXPECT_EQ(1000u, rhs.max_component_size_on_disk());
+        EXPECT_EQ(&rhs.fusion_size_on_disk(500), &rhs);
         EXPECT_EQ(&stats.merge(rhs), &stats);
     }
     EXPECT_EQ(100u, stats.memoryUsage().allocatedBytes());
     EXPECT_EQ(10u, stats.docsInMemory());
     EXPECT_EQ(1000u, stats.sizeOnDisk());
-    EXPECT_EQ(1000u, stats.max_component_size_on_disk());
+    EXPECT_EQ(500u, stats.fusion_size_on_disk());
 
     stats.merge(SearchableStats()
                         .memoryUsage(vespalib::MemoryUsage(150,0,0,0))
                         .docsInMemory(15)
-                        .sizeOnDisk(1500));
+                        .sizeOnDisk(1500)
+                        .fusion_size_on_disk(800));
     EXPECT_EQ(250u, stats.memoryUsage().allocatedBytes());
     EXPECT_EQ(25u, stats.docsInMemory());
     EXPECT_EQ(2500u, stats.sizeOnDisk());
-    EXPECT_EQ(1500u, stats.max_component_size_on_disk());
-
-    stats.merge(SearchableStats()
-                        .memoryUsage(vespalib::MemoryUsage(120,0,0,0))
-                        .docsInMemory(12)
-                        .sizeOnDisk(1200));
-    EXPECT_EQ(370u, stats.memoryUsage().allocatedBytes());
-    EXPECT_EQ(37u, stats.docsInMemory());
-    EXPECT_EQ(3700u, stats.sizeOnDisk());
-    EXPECT_EQ(1500u, stats.max_component_size_on_disk());
+    EXPECT_EQ(1300u, stats.fusion_size_on_disk());
 }
 
 GTEST_MAIN_RUN_ALL_TESTS()
diff --git a/searchlib/src/vespa/searchlib/util/searchable_stats.h b/searchlib/src/vespa/searchlib/util/searchable_stats.h
index 970076c12d0..e785a4c4483 100644
--- a/searchlib/src/vespa/searchlib/util/searchable_stats.h
+++ b/searchlib/src/vespa/searchlib/util/searchable_stats.h
@@ -15,11 +15,11 @@ class SearchableStats
 private:
     vespalib::MemoryUsage _memoryUsage;
     size_t _docsInMemory;
-    size_t _sizeOnDisk;
-    size_t _max_component_size_on_disk;
+    size_t _sizeOnDisk; // in bytes
+    size_t _fusion_size_on_disk; // in bytes
 
 public:
-    SearchableStats() : _memoryUsage(), _docsInMemory(0), _sizeOnDisk(0), _max_component_size_on_disk(0) {}
+    SearchableStats() : _memoryUsage(), _docsInMemory(0), _sizeOnDisk(0), _fusion_size_on_disk(0) {}
     SearchableStats &memoryUsage(const vespalib::MemoryUsage &usage) {
         _memoryUsage = usage;
         return *this;
@@ -32,22 +32,20 @@ public:
     size_t docsInMemory() const { return _docsInMemory; }
     SearchableStats &sizeOnDisk(size_t value) {
         _sizeOnDisk = value;
-        _max_component_size_on_disk = value;
         return *this;
     }
     size_t sizeOnDisk() const { return _sizeOnDisk; }
-
-    /**
-     * Returns the max disk size used by a single Searchable component,
-     * e.g. among the components that are merged into a SearchableStats instance via merge().
-     */
-    size_t max_component_size_on_disk() const { return _max_component_size_on_disk; }
+    SearchableStats& fusion_size_on_disk(size_t value) {
+        _fusion_size_on_disk = value;
+        return *this;
+    }
+    size_t fusion_size_on_disk() const { return _fusion_size_on_disk; }
 
     SearchableStats &merge(const SearchableStats &rhs) {
         _memoryUsage.merge(rhs._memoryUsage);
         _docsInMemory += rhs._docsInMemory;
         _sizeOnDisk += rhs._sizeOnDisk;
-        _max_component_size_on_disk = std::max(_max_component_size_on_disk, rhs._sizeOnDisk);
+        _fusion_size_on_disk += rhs._fusion_size_on_disk;
         return *this;
     }
 };