Update listener (#4925)

* remove Listener::apply * remove Listener::apply * Revert "remove Listener::apply" This reverts commit 7898a20.
vesoft-inc · Nov 23, 2022 · b251e2f · b251e2f
1 parent 2c447a9
commit b251e2f
Show file tree

Hide file tree

Showing 5 changed files with 241 additions and 148 deletions.
diff --git a/src/kvstore/Listener.cpp b/src/kvstore/Listener.cpp
@@ -10,7 +10,6 @@
 #include "kvstore/LogEncoder.h"
 
 DEFINE_int32(listener_commit_interval_secs, 1, "Listener commit interval");
-DEFINE_int32(listener_commit_batch_size, 1000, "Max batch size when listener commit");
 DEFINE_uint32(ft_request_retry_times, 3, "Retry times if fulltext request failed");
 DEFINE_int32(ft_bulk_batch_size, 100, "Max batch size when bulk insert");
 DEFINE_int32(listener_pursue_leader_threshold, 1000, "Catch up with the leader's threshold");
@@ -161,123 +160,7 @@ void Listener::doApply() {
   });
 }
 
-void Listener::processLogs() {
-  std::unique_ptr<LogIterator> iter;
-  {
-    std::lock_guard<std::mutex> guard(raftLock_);
-    if (lastApplyLogId_ >= committedLogId_) {
-      return;
-    }
-    iter = wal_->iterator(lastApplyLogId_ + 1, committedLogId_);
-  }
-
-  LogID lastApplyId = -1;
-  // the kv pair which can sync to remote safely
-  std::vector<KV> data;
-  while (iter->valid()) {
-    lastApplyId = iter->logId();
-
-    auto log = iter->logMsg();
-    if (log.empty()) {
-      // skip the heartbeat
-      ++(*iter);
-      continue;
-    }
-
-    DCHECK_GE(log.size(), sizeof(int64_t) + 1 + sizeof(uint32_t));
-    switch (log[sizeof(int64_t)]) {
-      case OP_PUT: {
-        auto pieces = decodeMultiValues(log);
-        DCHECK_EQ(2, pieces.size());
-        data.emplace_back(pieces[0], pieces[1]);
-        break;
-      }
-      case OP_MULTI_PUT: {
-        auto kvs = decodeMultiValues(log);
-        DCHECK_EQ(0, kvs.size() % 2);
-        for (size_t i = 0; i < kvs.size(); i += 2) {
-          data.emplace_back(kvs[i], kvs[i + 1]);
-        }
-        break;
-      }
-      case OP_REMOVE:
-      case OP_REMOVE_RANGE:
-      case OP_MULTI_REMOVE: {
-        break;
-      }
-      case OP_BATCH_WRITE: {
-        auto batch = decodeBatchValue(log);
-        for (auto& op : batch) {
-          // OP_BATCH_REMOVE and OP_BATCH_REMOVE_RANGE is igored
-          if (op.first == BatchLogType::OP_BATCH_PUT) {
-            data.emplace_back(op.second.first, op.second.second);
-          }
-        }
-        break;
-      }
-      case OP_TRANS_LEADER:
-      case OP_ADD_LEARNER:
-      case OP_ADD_PEER:
-      case OP_REMOVE_PEER: {
-        break;
-      }
-      default: {
-        VLOG(2) << idStr_ << "Unknown operation: " << static_cast<int32_t>(log[0]);
-      }
-    }
 
-    if (static_cast<int32_t>(data.size()) > FLAGS_listener_commit_batch_size) {
-      break;
-    }
-    ++(*iter);
-  }
-
-  // apply to state machine
-  if (lastApplyId != -1 && apply(data)) {
-    std::lock_guard<std::mutex> guard(raftLock_);
-    lastApplyLogId_ = lastApplyId;
-    persist(committedLogId_, term_, lastApplyLogId_);
-    VLOG(2) << idStr_ << "Listener succeeded apply log to " << lastApplyLogId_;
-    lastApplyTime_ = time::WallClock::fastNowInMilliSec();
-  }
-}
-
-std::tuple<nebula::cpp2::ErrorCode, int64_t, int64_t> Listener::commitSnapshot(
-    const std::vector<std::string>& rows,
-    LogID committedLogId,
-    TermID committedLogTerm,
-    bool finished) {
-  VLOG(2) << idStr_ << "Listener is committing snapshot.";
-  int64_t count = 0;
-  int64_t size = 0;
-  std::vector<KV> data;
-  data.reserve(rows.size());
-  for (const auto& row : rows) {
-    count++;
-    size += row.size();
-    auto kv = decodeKV(row);
-    data.emplace_back(kv.first, kv.second);
-  }
-  if (!apply(data)) {
-    LOG(INFO) << idStr_ << "Failed to apply data while committing snapshot.";
-    return {
-        nebula::cpp2::ErrorCode::E_RAFT_PERSIST_SNAPSHOT_FAILED, kNoSnapshotCount, kNoSnapshotSize};
-  }
-  if (finished) {
-    CHECK(!raftLock_.try_lock());
-    leaderCommitId_ = committedLogId;
-    lastApplyLogId_ = committedLogId;
-    persist(committedLogId, committedLogTerm, lastApplyLogId_);
-    lastApplyTime_ = time::WallClock::fastNowInMilliSec();
-    LOG(INFO) << folly::sformat(
-        "Commit snapshot to : committedLogId={},"
-        "committedLogTerm={}, lastApplyLogId_={}",
-        committedLogId,
-        committedLogTerm,
-        lastApplyLogId_);
-  }
-  return {nebula::cpp2::ErrorCode::SUCCEEDED, count, size};
-}
 
 void Listener::resetListener() {
   std::lock_guard<std::mutex> g(raftLock_);

diff --git a/src/kvstore/Listener.h b/src/kvstore/Listener.h
@@ -9,6 +9,7 @@
 #include "common/base/Base.h"
 #include "common/meta/SchemaManager.h"
 #include "kvstore/Common.h"
+#include "kvstore/LogEncoder.h"
 #include "kvstore/raftex/Host.h"
 #include "kvstore/raftex/RaftPart.h"
 #include "kvstore/wal/FileBasedWal.h"
@@ -180,14 +181,6 @@ class Listener : public raftex::RaftPart {
    */
   virtual LogID lastApplyLogId() = 0;
 
-  /**
-   * @brief Apply data into listener's state machine
-   *
-   * @param data Key/value to apply
-   * @return True if succeed. False if failed.
-   */
-  virtual bool apply(const std::vector<KV>& data) = 0;
-
   /**
    * @brief Persist commitLogId commitLogTerm and lastApplyLogId
    */
@@ -272,31 +265,13 @@ class Listener : public raftex::RaftPart {
                      ClusterID clusterId,
                      folly::StringPiece log) override;
 
-  /**
-   * @brief If the listener falls behind way to much than leader, the leader will send all its data
-   * in snapshot by batch, listener need to implement this method to apply the batch to state
-   * machine. The return value is a pair of <logs count, logs size> of this batch.
-   *
-   * @param data Data to apply
-   * @param committedLogId Commit log id of snapshot
-   * @param committedLogTerm Commit log term of snapshot
-   * @param finished Whether spapshot is finished
-   * @return std::tuple<nebula::cpp2::ErrorCode, int64_t, int64_t> Return {ok, count, size} if
-   * succeed, else return {errorcode, -1, -1}
-   */
-  std::tuple<nebula::cpp2::ErrorCode, int64_t, int64_t> commitSnapshot(
-      const std::vector<std::string>& data,
-      LogID committedLogId,
-      TermID committedLogTerm,
-      bool finished) override;
-
   /**
    * @brief Background job thread will trigger doApply to apply data into state machine periodically
    */
   void doApply();
 
   // Process logs and then call apply to execute
-  virtual void processLogs();
+  virtual void processLogs() = 0;
 
  protected:
   LogID leaderCommitId_ = 0;

diff --git a/src/kvstore/plugins/elasticsearch/ESListener.cpp b/src/kvstore/plugins/elasticsearch/ESListener.cpp
@@ -10,6 +10,7 @@
 
 DECLARE_uint32(ft_request_retry_times);
 DECLARE_int32(ft_bulk_batch_size);
+DEFINE_int32(listener_commit_batch_size, 1000, "Max batch size when listener commit");
 
 namespace nebula {
 namespace kvstore {
@@ -244,5 +245,122 @@ bool ESListener::writeDatum(const std::vector<nebula::plugin::DocItem>& items) c
   return true;
 }
 
+void ESListener::processLogs() {
+  std::unique_ptr<LogIterator> iter;
+  {
+    std::lock_guard<std::mutex> guard(raftLock_);
+    if (lastApplyLogId_ >= committedLogId_) {
+      return;
+    }
+    iter = wal_->iterator(lastApplyLogId_ + 1, committedLogId_);
+  }
+
+  LogID lastApplyId = -1;
+  // the kv pair which can sync to remote safely
+  std::vector<KV> data;
+  while (iter->valid()) {
+    lastApplyId = iter->logId();
+
+    auto log = iter->logMsg();
+    if (log.empty()) {
+      // skip the heartbeat
+      ++(*iter);
+      continue;
+    }
+
+    DCHECK_GE(log.size(), sizeof(int64_t) + 1 + sizeof(uint32_t));
+    switch (log[sizeof(int64_t)]) {
+      case OP_PUT: {
+        auto pieces = decodeMultiValues(log);
+        DCHECK_EQ(2, pieces.size());
+        data.emplace_back(pieces[0], pieces[1]);
+        break;
+      }
+      case OP_MULTI_PUT: {
+        auto kvs = decodeMultiValues(log);
+        DCHECK_EQ(0, kvs.size() % 2);
+        for (size_t i = 0; i < kvs.size(); i += 2) {
+          data.emplace_back(kvs[i], kvs[i + 1]);
+        }
+        break;
+      }
+      case OP_REMOVE:
+      case OP_REMOVE_RANGE:
+      case OP_MULTI_REMOVE: {
+        break;
+      }
+      case OP_BATCH_WRITE: {
+        auto batch = decodeBatchValue(log);
+        for (auto& op : batch) {
+          // OP_BATCH_REMOVE and OP_BATCH_REMOVE_RANGE is igored
+          if (op.first == BatchLogType::OP_BATCH_PUT) {
+            data.emplace_back(op.second.first, op.second.second);
+          }
+        }
+        break;
+      }
+      case OP_TRANS_LEADER:
+      case OP_ADD_LEARNER:
+      case OP_ADD_PEER:
+      case OP_REMOVE_PEER: {
+        break;
+      }
+      default: {
+        VLOG(2) << idStr_ << "Unknown operation: " << static_cast<int32_t>(log[0]);
+      }
+    }
+
+    if (static_cast<int32_t>(data.size()) > FLAGS_listener_commit_batch_size) {
+      break;
+    }
+    ++(*iter);
+  }
+  // apply to state machine
+  if (lastApplyId != -1 && apply(data)) {
+    std::lock_guard<std::mutex> guard(raftLock_);
+    lastApplyLogId_ = lastApplyId;
+    persist(committedLogId_, term_, lastApplyLogId_);
+    VLOG(2) << idStr_ << "Listener succeeded apply log to " << lastApplyLogId_;
+    lastApplyTime_ = time::WallClock::fastNowInMilliSec();
+  }
+}
+
+std::tuple<nebula::cpp2::ErrorCode, int64_t, int64_t> ESListener::commitSnapshot(
+    const std::vector<std::string>& rows,
+    LogID committedLogId,
+    TermID committedLogTerm,
+    bool finished) {
+  VLOG(2) << idStr_ << "Listener is committing snapshot.";
+  int64_t count = 0;
+  int64_t size = 0;
+  std::vector<KV> data;
+  data.reserve(rows.size());
+  for (const auto& row : rows) {
+    count++;
+    size += row.size();
+    auto kv = decodeKV(row);
+    data.emplace_back(kv.first, kv.second);
+  }
+  if (!apply(data)) {
+    LOG(INFO) << idStr_ << "Failed to apply data while committing snapshot.";
+    return {
+        nebula::cpp2::ErrorCode::E_RAFT_PERSIST_SNAPSHOT_FAILED, kNoSnapshotCount, kNoSnapshotSize};
+  }
+  if (finished) {
+    CHECK(!raftLock_.try_lock());
+    leaderCommitId_ = committedLogId;
+    lastApplyLogId_ = committedLogId;
+    persist(committedLogId, committedLogTerm, lastApplyLogId_);
+    lastApplyTime_ = time::WallClock::fastNowInMilliSec();
+    LOG(INFO) << folly::sformat(
+        "Commit snapshot to : committedLogId={},"
+        "committedLogTerm={}, lastApplyLogId_={}",
+        committedLogId,
+        committedLogTerm,
+        lastApplyLogId_);
+  }
+  return {nebula::cpp2::ErrorCode::SUCCEEDED, count, size};
+}
+
 }  // namespace kvstore
 }  // namespace nebula
diff --git a/src/kvstore/plugins/elasticsearch/ESListener.h b/src/kvstore/plugins/elasticsearch/ESListener.h
@@ -71,7 +71,7 @@ class ESListener : public Listener {
    * @param data Key/value to apply
    * @return True if succeed. False if failed.
    */
-  bool apply(const std::vector<KV>& data) override;
+  bool apply(const std::vector<KV>& data);
 
   /**
    * @brief Persist commitLogId commitLogTerm and lastApplyLogId
@@ -92,6 +92,14 @@ class ESListener : public Listener {
    */
   LogID lastApplyLogId() override;
 
+  void processLogs() override;
+
+  std::tuple<nebula::cpp2::ErrorCode, int64_t, int64_t> commitSnapshot(
+      const std::vector<std::string>& data,
+      LogID committedLogId,
+      TermID committedLogTerm,
+      bool finished) override;
+
  private:
   /**
    * @brief Write last commit id, last commit term, last apply id to a file