Add auto prefetching parameters to db_bench and db_stress (#10632)

Summary:
Same as title

Pull Request resolved: https://github.com/facebook/rocksdb/pull/10632

Test Plan: make crash_test -j32

Reviewed By: anand1976

Differential Revision: D39241479

Pulled By: akankshamahajan15

fbshipit-source-id: 5db5b0c007da786bacc1b30d8926d36d6d029b87

db_stress_tool/db_stress_common.h

@@ -314,6 +314,11 @@ DECLARE_int64(preclude_last_level_data_seconds);
 
 DECLARE_int32(verify_iterator_with_expected_state_one_in);
 
+DECLARE_uint64(readahead_size);
+DECLARE_uint64(initial_auto_readahead_size);
+DECLARE_uint64(max_auto_readahead_size);
+DECLARE_uint64(num_file_reads_for_auto_readahead);
+
 constexpr long KB = 1024;
 constexpr int kRandomValueMaxFactor = 3;
 constexpr int kValueMaxLen = 100;

db_stress_tool/db_stress_gflags.cc

@@ -1035,4 +1035,13 @@ DEFINE_int32(verify_iterator_with_expected_state_one_in, 0,
              "chance that the iterator is verified against the expected state "
              "file, instead of comparing keys between two iterators.");
 
+DEFINE_uint64(readahead_size, 0, "Iterator readahead size");
+DEFINE_uint64(initial_auto_readahead_size, 0,
+              "Initial auto readahead size for prefetching during Iteration");
+DEFINE_uint64(max_auto_readahead_size, 0,
+              "Max auto readahead size for prefetching during Iteration");
+DEFINE_uint64(
+    num_file_reads_for_auto_readahead, 0,
+    "Num of sequential reads to enable auto prefetching during Iteration");
+
 #endif  // GFLAGS

db_stress_tool/db_stress_test_base.cc

@@ -635,6 +635,7 @@ void StressTest::OperateDb(ThreadState* thread) {
       FLAGS_rate_limit_user_ops ? Env::IO_USER : Env::IO_TOTAL;
   read_opts.async_io = FLAGS_async_io;
   read_opts.adaptive_readahead = FLAGS_adaptive_readahead;
+  read_opts.readahead_size = FLAGS_readahead_size;
   WriteOptions write_opts;
   if (FLAGS_rate_limit_auto_wal_flush) {
     write_opts.rate_limiter_priority = Env::IO_USER;
@@ -2950,6 +2951,11 @@ void InitializeOptionsFromFlags(
   block_based_options.prepopulate_block_cache =
       static_cast<BlockBasedTableOptions::PrepopulateBlockCache>(
           FLAGS_prepopulate_block_cache);
+  block_based_options.initial_auto_readahead_size =
+      FLAGS_initial_auto_readahead_size;
+  block_based_options.max_auto_readahead_size = FLAGS_max_auto_readahead_size;
+  block_based_options.num_file_reads_for_auto_readahead =
+      FLAGS_num_file_reads_for_auto_readahead;
   options.table_factory.reset(NewBlockBasedTableFactory(block_based_options));
   options.db_write_buffer_size = FLAGS_db_write_buffer_size;
   options.write_buffer_size = FLAGS_write_buffer_size;

tools/db_bench_tool.cc

@@ -1236,6 +1236,14 @@ DEFINE_uint64(
     "BlockBasedTableOptions.initial_auto_readahead_size and doubles on every "
     "additional read upto max_auto_readahead_size");
 
+DEFINE_uint64(
+    num_file_reads_for_auto_readahead,
+    ROCKSDB_NAMESPACE::BlockBasedTableOptions()
+        .num_file_reads_for_auto_readahead,
+    "Rocksdb implicit readahead is enabled if reads are sequential and "
+    "num_file_reads_for_auto_readahead indicates after how many sequential "
+    "reads into that file internal auto prefetching should be start.");
+
 static enum ROCKSDB_NAMESPACE::CompressionType StringToCompressionType(
     const char* ctype) {
   assert(ctype);
@@ -4372,6 +4380,8 @@ class Benchmark {
           FLAGS_max_auto_readahead_size;
       block_based_options.initial_auto_readahead_size =
           FLAGS_initial_auto_readahead_size;
+      block_based_options.num_file_reads_for_auto_readahead =
+          FLAGS_num_file_reads_for_auto_readahead;
       BlockBasedTableOptions::PrepopulateBlockCache prepopulate_block_cache =
           block_based_options.prepopulate_block_cache;
       switch (FLAGS_prepopulate_block_cache) {

tools/db_crashtest.py

@@ -180,6 +180,10 @@ default_params = {
     "secondary_cache_uri":  lambda: random.choice(
         ["", "compressed_secondary_cache://capacity=8388608"]),
     "allow_data_in_errors": True,
+    "readahead_size": lambda: random.choice([0, 16384, 524288]),
+    "initial_auto_readahead_size": lambda: random.choice([0, 16384, 524288]),
+    "max_auto_readahead_size": lambda: random.choice([0, 16384, 524288]),
+    "num_file_reads_for_auto_readahead": lambda: random.choice([0, 1, 2]),
 }
 
 _TEST_DIR_ENV_VAR = 'TEST_TMPDIR'