feat(replica): Support FlushDB command for replication #580

src/server/journal/executor.cc

@@ -12,7 +12,7 @@ namespace dfly {
 JournalExecutor::JournalExecutor(Service* service) : service_{service} {
 }
 
-void JournalExecutor::Execute(journal::ParsedEntry&& entry) {
+void JournalExecutor::Execute(journal::ParsedEntry& entry) {
   if (entry.payload) {
     io::NullSink null_sink;
     ConnectionContext conn_context{&null_sink, nullptr};

src/server/journal/executor.h

@@ -14,7 +14,7 @@ class Service;
 class JournalExecutor {
  public:
   JournalExecutor(Service* service);
-  void Execute(journal::ParsedEntry&& entry);
+  void Execute(journal::ParsedEntry& entry);
 
  private:
   Service* service_;

src/server/journal/serializer.cc

@@ -70,6 +70,7 @@ error_code JournalWriter::Write(const journal::Entry& entry) {
       return Write(entry.dbid);
     case journal::Op::COMMAND:
       RETURN_ON_ERR(Write(entry.txid));
+      RETURN_ON_ERR(Write(entry.shard_cnt));
       return std::visit([this](const auto& payload) { return Write(payload); }, entry.payload);
     default:
       break;
@@ -100,6 +101,10 @@ io::Result<uint16_t> JournalReader::ReadU16(io::Source* source) {
   return ReadPackedUIntTyped<uint16_t>(source);
 }
 
+io::Result<uint32_t> JournalReader::ReadU32(io::Source* source) {
+  return ReadPackedUIntTyped<uint32_t>(source);
+}
+
 io::Result<uint64_t> JournalReader::ReadU64(io::Source* source) {
   return ReadPackedUIntTyped<uint64_t>(source);
 }
@@ -153,6 +158,7 @@ io::Result<journal::ParsedEntry> JournalReader::ReadEntry(io::Source* source) {
   switch (entry.opcode) {
     case journal::Op::COMMAND:
       SET_OR_UNEXPECT(ReadU64(source), entry.txid);
+      SET_OR_UNEXPECT(ReadU32(source), entry.shard_cnt);
       entry.payload = CmdArgVec{};
       if (auto ec = Read(source, &*entry.payload); ec)
         return make_unexpected(ec);

src/server/journal/serializer.h

@@ -55,6 +55,7 @@ struct JournalReader {
   // TODO: Templated endian encoding to not repeat...?
   io::Result<uint8_t> ReadU8(io::Source* source);
   io::Result<uint16_t> ReadU16(io::Source* source);
+  io::Result<uint32_t> ReadU32(io::Source* source);
   io::Result<uint64_t> ReadU64(io::Source* source);
 
   // Read string into internal buffer and return size.

src/server/journal/types.h

@@ -22,6 +22,7 @@ struct EntryBase {
   TxId txid;
   Op opcode;
   DbIndex dbid;
+  uint32_t shard_cnt;
 };
 
 // This struct represents a single journal entry.
@@ -34,11 +35,11 @@ struct Entry : public EntryBase {
                    std::pair<std::string_view, ArgSlice>  // Command and its shard parts.
                    >;
 
-  Entry(TxId txid, DbIndex dbid, Payload pl)
-      : EntryBase{txid, journal::Op::COMMAND, dbid}, payload{pl} {
+  Entry(TxId txid, DbIndex dbid, Payload pl, uint32_t shard_cnt)
+      : EntryBase{txid, journal::Op::COMMAND, dbid, shard_cnt}, payload{pl} {
   }
 
-  Entry(journal::Op opcode, DbIndex dbid) : EntryBase{0, opcode, dbid}, payload{} {
+  Entry(journal::Op opcode, DbIndex dbid) : EntryBase{0, opcode, dbid, 0}, payload{} {
   }
 
   Payload payload;
@@ -50,11 +51,11 @@ struct ParsedEntry : public EntryBase {
 
   ParsedEntry() = default;
 
-  ParsedEntry(journal::Op opcode, DbIndex dbid) : EntryBase{0, opcode, dbid}, payload{} {
+  ParsedEntry(journal::Op opcode, DbIndex dbid) : EntryBase{0, opcode, dbid, 0}, payload{} {
   }
 
-  ParsedEntry(TxId txid, DbIndex dbid, Payload pl)
-      : EntryBase{txid, journal::Op::COMMAND, dbid}, payload{pl} {
+  ParsedEntry(TxId txid, DbIndex dbid, Payload pl, uint32_t shard_cnt)
+      : EntryBase{txid, journal::Op::COMMAND, dbid, shard_cnt}, payload{pl} {
   }
 
   Payload payload;

src/server/journal_test.cc

@@ -95,13 +95,14 @@ TEST(Journal, WriteRead) {
   auto slice = [v = &slices](auto... ss) { return StoreSlice(v, ss...); };
   auto list = [v = &lists](auto... ss) { return StoreList(v, ss...); };
 
-  std::vector<journal::Entry> test_entries = {{0, 0, make_pair("MSET", slice("A", "1", "B", "2"))},
-                                              {1, 0, make_pair("MSET", slice("C", "3"))},
-                                              {2, 0, list("DEL", "A", "B")},
-                                              {3, 1, list("LPUSH", "l", "v1", "v2")},
-                                              {4, 0, make_pair("MSET", slice("D", "4"))},
-                                              {5, 1, list("DEL", "l1")},
-                                              {6, 2, list("SET", "E", "2")}};
+  std::vector<journal::Entry> test_entries = {
+      {0, 0, make_pair("MSET", slice("A", "1", "B", "2")), 2},
+      {0, 0, make_pair("MSET", slice("C", "3")), 2},
+      {1, 0, list("DEL", "A", "B"), 2},
+      {2, 1, list("LPUSH", "l", "v1", "v2"), 1},
+      {3, 0, make_pair("MSET", slice("D", "4")), 1},
+      {4, 1, list("DEL", "l1"), 1},
+      {5, 2, list("SET", "E", "2"), 1}};
 
   // Write all entries to string file.
   io::StringSink ss;

src/server/rdb_load.cc

@@ -1983,7 +1983,7 @@ error_code RdbLoaderBase::HandleJournalBlob(Service* service, DbIndex dbid) {
   while (done < num_entries) {
     journal::ParsedEntry entry{};
     SET_OR_RETURN(journal_reader_.ReadEntry(&bs), entry);
-    ex.Execute(std::move(entry));
+    ex.Execute(entry);
     done++;
   }
 

src/server/replica.cc

@@ -97,9 +97,11 @@ Replica::Replica(string host, uint16_t port, Service* se) : service_(*se) {
   master_context_.port = port;
 }
 
-Replica::Replica(const MasterContext& context, uint32_t dfly_flow_id, Service* service)
+Replica::Replica(const MasterContext& context, uint32_t dfly_flow_id, Service* service,
+                 std::shared_ptr<Replica::MultiShardExecution> shared_exe_data)
     : service_(*service), master_context_(context) {
   master_context_.dfly_flow_id = dfly_flow_id;
+  multi_shard_exe_ = shared_exe_data;
 }
 
 Replica::~Replica() {
@@ -427,13 +429,13 @@ error_code Replica::InitiatePSync() {
 // Initialize and start sub-replica for each flow.
 error_code Replica::InitiateDflySync() {
   DCHECK_GT(num_df_flows_, 0u);
-
+  multi_shard_exe_.reset(new MultiShardExecution());
   shard_flows_.resize(num_df_flows_);
   for (unsigned i = 0; i < num_df_flows_; ++i) {
-    shard_flows_[i].reset(new Replica(master_context_, i, &service_));
+    shard_flows_[i].reset(new Replica(master_context_, i, &service_, multi_shard_exe_));
   }
 
-  // Blocked on untill all flows got full sync cut.
+  // Blocked on until all flows got full sync cut.
   fibers_ext::BlockingCounter sync_block{num_df_flows_};
 
   auto err_handler = [this, sync_block](const auto& ge) mutable {
@@ -705,14 +707,63 @@ void Replica::StableSyncDflyFb(Context* cntx) {
       cntx->Error(res.error(), "Journal format error");
       return;
     }
-
-    executor.Execute(std::move(res.value()));
-
+    ExecuteEntry(&executor, res.value());
     last_io_time_ = sock_->proactor()->GetMonotonicTimeNs();
   }
   return;
 }
 
+void Replica::ExecuteEntry(JournalExecutor* executor, journal::ParsedEntry& entry) {
+  if (entry.shard_cnt <= 1) {  // not multi shard cmd
+    executor->Execute(entry);
+    return;
+  }
+
+  // Multi shard command flow:
+  //  step 1: Fiber wait until all the fibers that should execute this tranaction got
+  //    to the journal entry of the transaction.
+  //  step 2: execute the command (All fibers)
+  //  step 3: Fiber wait until all fibers finished the execution
+  //  By step 1 we enforce that replica will execute multi shard commands that finished on master
+  //  By step 3 we ensures the correctness of flushall/flushdb commands
+
+  // TODO: this implemantaion does not support atomicity in replica
+  // Although multi shard transaction happen in step 2 very close to each other,
+  // user can query replica between executions.
+  // To support atomicity we should have one fiber in step 2 which will excute all the entries of
+  // the transaction together. In case of global comand such as flushdb the command can be executed
+  // by only one fiber.
+
+  // TODO: support error handler in this flow
+
+  // Only the first fiber to reach the transaction will create data for transaction in map
+  multi_shard_exe_->map_mu.lock();
+  auto [it, was_insert] = multi_shard_exe_->tx_sync_execution.emplace(entry.txid, entry.shard_cnt);
+
+  // Note: we must release the mutex befor calling wait on barrier
+  multi_shard_exe_->map_mu.unlock();
+
+  VLOG(2) << "txid: " << entry.txid << " unique_shard_cnt_: " << entry.shard_cnt
+          << " was_insert: " << was_insert;
+
+  // step 1
+  it->second.barrier.wait();
+  // step 2
+  executor->Execute(entry);
+  // step 3
+  it->second.barrier.wait();
+
+  // Note: erase from map can be done only after all fibers returned from wait.
+  // The last fiber which will decrease the counter to 0 will be the one to erase the data from map
+  auto val = it->second.counter.fetch_sub(1, std::memory_order_relaxed);
+  VLOG(2) << "txid: " << entry.txid << " unique_shard_cnt_: " << entry.shard_cnt
+          << " counter: " << val;
+  if (val == 1) {
+    std::lock_guard lg{multi_shard_exe_->map_mu};
+    multi_shard_exe_->tx_sync_execution.erase(entry.txid);
+  }
+}
+
 error_code Replica::ReadRespReply(base::IoBuf* io_buf, uint32_t* consumed) {
   DCHECK(parser_);
 

src/server/replica.h

@@ -3,13 +3,16 @@
 //
 #pragma once
 
+#include <boost/fiber/barrier.hpp>
 #include <boost/fiber/fiber.hpp>
+#include <boost/fiber/mutex.hpp>
 #include <variant>
 
 #include "base/io_buf.h"
 #include "facade/facade_types.h"
 #include "facade/redis_parser.h"
 #include "server/common.h"
+#include "server/journal/types.h"
 #include "util/fiber_socket_base.h"
 #include "util/fibers/fibers_ext.h"
 
@@ -21,6 +24,7 @@ namespace dfly {
 
 class Service;
 class ConnectionContext;
+class JournalExecutor;
 
 class Replica {
  private:
@@ -46,6 +50,19 @@ class Replica {
     R_SYNC_OK = 0x10,
   };
 
+  struct MultiShardExecution {
+    boost::fibers::mutex map_mu;
+
+    struct TxExecutionSync {
+      boost::fibers::barrier barrier;
+      std::atomic_uint32_t counter;
+      TxExecutionSync(uint32_t counter) : barrier(counter), counter(counter) {
+      }
+    };
+
+    std::unordered_map<TxId, TxExecutionSync> tx_sync_execution;
+  };
+
  public:
   Replica(std::string master_host, uint16_t port, Service* se);
   ~Replica();
@@ -81,7 +98,8 @@ class Replica {
 
  private: /* Main dlfly flow mode functions */
   // Initialize as single dfly flow.
-  Replica(const MasterContext& context, uint32_t dfly_flow_id, Service* service);
+  Replica(const MasterContext& context, uint32_t dfly_flow_id, Service* service,
+          std::shared_ptr<MultiShardExecution> shared_exe_data);
 
   // Start replica initialized as dfly flow.
   std::error_code StartFullSyncFlow(util::fibers_ext::BlockingCounter block, Context* cntx);
@@ -122,6 +140,8 @@ class Replica {
   // Send command, update last_io_time, return error.
   std::error_code SendCommand(std::string_view command, facade::ReqSerializer* serializer);
 
+  void ExecuteEntry(JournalExecutor* executor, journal::ParsedEntry& entry);
+
  public: /* Utility */
   struct Info {
     std::string host;
@@ -154,6 +174,8 @@ class Replica {
   MasterContext master_context_;
   std::unique_ptr<util::LinuxSocketBase> sock_;
 
+  std::shared_ptr<MultiShardExecution> multi_shard_exe_;
+
   // MainReplicationFb in standalone mode, FullSyncDflyFb in flow mode.
   ::boost::fibers::fiber sync_fb_;
   std::vector<std::unique_ptr<Replica>> shard_flows_;

src/server/transaction.cc

@@ -1221,7 +1221,7 @@ void Transaction::LogJournalOnShard(EngineShard* shard) {
     entry_payload =
         make_pair(facade::ToSV(cmd_with_full_args_.front()), ShardArgsInShard(shard->shard_id()));
   }
-  journal->RecordEntry(journal::Entry{txid_, db_index_, entry_payload});
+  journal->RecordEntry(journal::Entry{txid_, db_index_, entry_payload, unique_shard_cnt_});
 }
 
 void Transaction::BreakOnShutdown() {