Report rewritten blocks telemetry

Author: jason-rl

src/overlaybd/lsmt/file.cpp

@@ -990,8 +990,12 @@ class LSMTFile : public LSMTReadOnlyFile {
         DataStat data_stat;
         data_stat.total_data_size = (buf.st_size - HeaderTrailer::SPACE);
         data_stat.valid_data_size = index()->block_count() * ALIGNMENT;
-        LOG_DEBUG("data_size: ` ( valid: ` )", data_stat.total_data_size,
-                  data_stat.valid_data_size);
+        // Get rewrite stats from the index
+        auto rw_stats = index()->rewrite_stats();
+        data_stat.total_blocks_written = rw_stats.total_blocks_written;
+        data_stat.rewritten_blocks = rw_stats.rewritten_blocks;
+        LOG_DEBUG("data_size: ` ( valid: ` ), rewrites: `/`", data_stat.total_data_size,
+                  data_stat.valid_data_size, data_stat.rewritten_blocks, data_stat.total_blocks_written);
         return data_stat;
     }
 

src/overlaybd/lsmt/file.h

@@ -98,6 +98,9 @@ class IFileRW : public IFileRO {
     struct DataStat {
         uint64_t total_data_size = -1; // size of total data
         uint64_t valid_data_size = -1; // size of valid data (excluding garbage)
+        // Block rewrite telemetry (in 512B block units)
+        uint64_t total_blocks_written = 0;  // Total blocks written
+        uint64_t rewritten_blocks = 0;      // Blocks that overwrote previous data
     };
     virtual DataStat data_stat() const = 0;
 

src/overlaybd/lsmt/index.cpp

@@ -256,6 +256,12 @@ class Index0 : public IComboIndex {
         }
     } alloc_blk;
 
+    // Block rewrite tracking for telemetry
+    struct rewrite_stats {
+        uint64_t total_blocks_written = 0;   // Total blocks written (in 512B units)
+        uint64_t rewritten_blocks = 0;       // Blocks that overwrote previous data
+    } m_rewrite_stats;
+
     // Index0(const set<SegmentMapping> &mapping) : mapping(mapping){};
 
     Index0(const SegmentMapping *pmappings = nullptr, size_t n = 0) {
@@ -314,6 +320,31 @@ class Index0 : public IComboIndex {
     virtual void insert(SegmentMapping m) override {
         if (m.length == 0)
             return;
+
+        // Track total blocks written for rewrite telemetry
+        m_rewrite_stats.total_blocks_written += m.length;
+
+        // Count blocks that will be overwritten (rewritten)
+        uint64_t rewritten = 0;
+        auto check_it = mapping.lower_bound(m);
+        // Check previous mapping for overlap
+        if (check_it != mapping.begin()) {
+            auto prev_it = std::prev(check_it);
+            if (prev_it->end() > m.offset) {
+                // Previous mapping overlaps with new write
+                rewritten += std::min(prev_it->end(), m.end()) - m.offset;
+            }
+        }
+        // Check current and subsequent mappings for overlap
+        for (auto it = check_it; it != mapping.end() && it->offset < m.end(); ++it) {
+            uint64_t overlap_start = std::max(it->offset, m.offset);
+            uint64_t overlap_end = std::min(it->end(), m.end());
+            if (overlap_end > overlap_start) {
+                rewritten += overlap_end - overlap_start;
+            }
+        }
+        m_rewrite_stats.rewritten_blocks += rewritten;
+
         alloc_blk += m;
         auto it = mapping.lower_bound(m);
         if (it == mapping.end()) {
@@ -369,6 +400,13 @@ class Index0 : public IComboIndex {
         return alloc_blk.m_alloc;
     }
 
+    virtual RewriteStats rewrite_stats() const override {
+        return RewriteStats{
+            m_rewrite_stats.total_blocks_written,
+            m_rewrite_stats.rewritten_blocks
+        };
+    }
+
     // returns the first and last mapping in the index
     // the there's no one, return an invalid mapping: [INVALID_OFFSET, 0) ==> 0
     virtual SegmentMapping front() const override {

src/overlaybd/lsmt/index.h

@@ -132,6 +132,12 @@ class IMemoryIndex {
     virtual IMemoryIndex *make_read_only_index() const = 0;
 };
 
+// Block rewrite statistics for telemetry
+struct RewriteStats {
+    uint64_t total_blocks_written = 0;   // Total blocks written (in 512B units)
+    uint64_t rewritten_blocks = 0;       // Blocks that overwrote previous data
+};
+
 // the level 0 memory index, which supports write
 class IMemoryIndex0 : public IMemoryIndex {
 public:
@@ -142,6 +148,9 @@ class IMemoryIndex0 : public IMemoryIndex {
     // memory allocation is aligned to the `alignment`
     virtual SegmentMapping *dump(size_t alignment = 0) const = 0;
     // virtual IMemoryIndex *make_read_only_index() const = 0;
+
+    // Get block rewrite statistics for telemetry
+    virtual RewriteStats rewrite_stats() const = 0;
 };
 
 class IComboIndex : public IMemoryIndex0 {

src/tools/overlaybd-commit.cpp

@@ -188,6 +188,17 @@ int main(int argc, char **argv) {
         out = fout;
     }
 
+    // Get data stats before commit for telemetry (includes rewrite frequency)
+    auto stats = fin->data_stat();
+    LOG_INFO("data_stat: total_data_size=`, valid_data_size=`, total_blocks_written=`, rewritten_blocks=`",
+             stats.total_data_size, stats.valid_data_size,
+             stats.total_blocks_written, stats.rewritten_blocks);
+
+    // Output rewrite stats as JSON to stdout for the caller to parse
+    // Format: {"total_blocks_written": N, "rewritten_blocks": M}
+    printf("{\"total_blocks_written\": %" PRIu64 ", \"rewritten_blocks\": %" PRIu64 "}\n",
+           stats.total_blocks_written, stats.rewritten_blocks);
+
     CommitArgs args(out);
     if (!uuid.empty()) {
         memset(args.uuid.data, 0, UUID::String::LEN);