From 18005fdeb70066452636e6da3e906e65c9fec4a3 Mon Sep 17 00:00:00 2001
From: Lizhe Ji <lizheji@meta.com>
Date: Mon, 15 Jun 2026 11:03:50 -0700
Subject: [PATCH 1/2] Add metadata_dim_ and total_rows_written_ members to
 EmbeddingRocksDB (#5883)

Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/2802


Declare two new EmbeddingRocksDB member fields used by the metadata column
family feature later in the stack:
- metadata_dim_: width of the metaheader stored in the separate metadata CF
- total_rows_written_: cumulative count of rows written

Reviewed By: EddyLXJ

Differential Revision: D108342270
---
 .../ssd_split_embeddings_cache/ssd_table_batched_embeddings.h   | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h b/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h
index 6539992ab6..8ea3f82a25 100644
--- a/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h
+++ b/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h
@@ -1396,10 +1396,12 @@ class EmbeddingRocksDB : public kv_db::EmbeddingKVDB {
   std::atomic<int64_t> fwd_l1_eviction_dur_{0};
   std::atomic<int64_t> bwd_l1_cnflct_miss_write_back_dur_{0};
   std::atomic<int64_t> flush_write_dur_{0};
+  std::atomic<int64_t> total_rows_written_{0}; // cumulative actual rows written
 
   std::unordered_map<const SnapshotHandle*, std::unique_ptr<SnapshotHandle>>
       snapshots_;
   int64_t max_D_;
+  int64_t metadata_dim_;
   int64_t elem_size_;
   std::vector<int64_t> sub_table_dims_;
   std::vector<int64_t> sub_table_hash_cumsum_;

From 581a46b91de1649e41bb88aaea1acd1e50e83c7a Mon Sep 17 00:00:00 2001
From: Lizhe Ji <lizheji@meta.com>
Date: Mon, 15 Jun 2026 11:03:50 -0700
Subject: [PATCH 2/2] Add metadata column family fields to EmbeddingRocksDB
 (#5890)

Summary:

Declare the metadata column family member fields (metadata_cf_handles_,
cf_options_) and add the required includes. These are used by the metadata
methods and existing-method wiring that follow in the stack.

Differential Revision: D108357928
---
 .../ssd_split_embeddings_cache/ssd_table_batched_embeddings.h | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h b/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h
index 8ea3f82a25..99af7d7d9e 100644
--- a/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h
+++ b/fbgemm_gpu/src/ssd_split_embeddings_cache/ssd_table_batched_embeddings.h
@@ -9,6 +9,8 @@
 #pragma once
 
 #include <algorithm>
+#include <cstring>
+#include <ctime>
 #include <iostream>
 #include <memory>
 
@@ -1376,11 +1378,13 @@ class EmbeddingRocksDB : public kv_db::EmbeddingKVDB {
   friend class CheckpointHandle;
 
   std::vector<std::unique_ptr<rocksdb::DB>> dbs_;
+  std::vector<rocksdb::ColumnFamilyHandle*> metadata_cf_handles_;
   std::vector<std::unique_ptr<Initializer>> initializers_;
   std::unique_ptr<folly::CPUThreadPoolExecutor> executor_;
   rocksdb::ReadOptions ro_{};
   rocksdb::WriteOptions wo_{};
   std::shared_ptr<rocksdb::RateLimiter> rate_limiter_;
+  rocksdb::ColumnFamilyOptions cf_options_;
   std::vector<int64_t> shard_flush_compaction_deadlines_;
   bool done_staggered_flushes_;
   int64_t memtable_flush_offset_;