fix: Resolve the inconsistency between member configuration and log index in snapshot meta

cmake/braft.cmake

@@ -20,7 +20,7 @@ ExternalProject_Add(
         DEPENDS brpc
         # The pr on braft is not merged, so I am using my own warehouse to run the test for the time being
         GIT_REPOSITORY "https://github.com/pikiwidb/braft.git"
-        GIT_TAG v1.1.2-alpha2
+        GIT_TAG v1.1.2.1
         GIT_SHALLOW true
         CMAKE_ARGS
         -DCMAKE_BUILD_TYPE=${LIB_BUILD_TYPE}

src/pikiwidb.cc

@@ -227,7 +227,7 @@ static int InitLimit() {
     limit.rlim_cur = maxfiles;
     limit.rlim_max = maxfiles;
     if (setrlimit(RLIMIT_NOFILE, &limit) != -1) {
-      WARN("your 'limit -n ' of {} is not enough for PikiwiDB to start. PikiwiDB have successfully reconfig it to ",
+      WARN("your 'limit -n ' of {} is not enough for PikiwiDB to start. PikiwiDB have successfully reconfig it to {}",
            old_limit, limit.rlim_cur);
     } else {
       ERROR(
@@ -282,7 +282,7 @@ int main(int ac, char* av[]) {
     daemonize();
   }
 
-  InitLimit();
+  // InitLimit(); // Code problem, program startup direct segment error
   pstd::InitRandom();
   SignalSetup();
   InitLogs();

src/praft/praft.cc

@@ -259,6 +259,16 @@ storage::LogIndex PRaft::GetLastLogIndex(bool is_flush) {
   return node_->get_last_log_index(is_flush);
 }
 
+void PRaft::GetConfigurationByIndex(const int64_t index, braft::ConfigurationEntry* conf,
+                                    braft::ConfigurationEntry* learner_conf) {
+  if (!node_) {
+    ERROR("Node is not initialized");
+    return;
+  }
+
+  node_->get_configuration(index, conf, learner_conf);
+}
+
 void PRaft::SendNodeRequest(PClient* client) {
   assert(client);
 
@@ -663,8 +673,9 @@ int PRaft::on_snapshot_load(braft::SnapshotReader* reader) {
     2. When a node is improperly shut down and restarted, the minimum flush-index should
        be obtained as the starting point for fault recovery.
     */
+    // replay from <replay_point + 1>
     uint64_t replay_point = PSTORE.GetBackend(db_id_)->GetStorage()->GetSmallestFlushedLogIndex();
-    node_->set_self_playback_point(replay_point);
+    node_->set_last_applied_index_and_term(replay_point);
     is_node_first_start_up_ = false;
     INFO("set replay_point: {}", replay_point);
 

src/praft/praft.h

@@ -14,6 +14,7 @@
 #include <tuple>
 #include <vector>
 
+#include "braft/configuration_manager.h"
 #include "braft/file_system_adaptor.h"
 #include "braft/raft.h"
 #include "brpc/server.h"
@@ -140,6 +141,8 @@ class PRaft : public braft::StateMachine {
   butil::Status GetListPeers(std::vector<braft::PeerId>* peers);
   storage::LogIndex GetTerm(uint64_t log_index);
   storage::LogIndex GetLastLogIndex(bool is_flush = false);
+  void GetConfigurationByIndex(const int64_t index, braft::ConfigurationEntry* conf,
+                               braft::ConfigurationEntry* learner_conf);
 
   bool IsInitialized() const { return node_ != nullptr && server_ != nullptr; }
 

src/praft/psnapshot.cc

@@ -10,6 +10,8 @@
 
 #include "psnapshot.h"
 
+#include "braft/configuration.h"
+#include "braft/configuration_manager.h"
 #include "braft/local_file_meta.pb.h"
 #include "braft/snapshot.h"
 #include "butil/files/file_path.h"
@@ -79,12 +81,27 @@ braft::FileAdaptor* PPosixFileSystemAdaptor::open(const std::string& path, int o
       PSTORE.HandleTaskSpecificDB(tasks);
       AddAllFiles(snapshot_path, &snapshot_meta_memtable, snapshot_path);
 
-      // update snapshot last log index and last_log_term
+      // update snapshot last log index, last_log_term, conf of last log index and learners
       auto& new_meta = const_cast<braft::SnapshotMeta&>(snapshot_meta_memtable.meta());
       auto last_log_index = PSTORE.GetBackend(db_id)->GetStorage()->GetSmallestFlushedLogIndex();
       new_meta.set_last_included_index(last_log_index);
       auto last_log_term = PRAFT.GetTerm(last_log_index);
       new_meta.set_last_included_term(last_log_term);
+      braft::ConfigurationEntry conf_entry;
+      braft::ConfigurationEntry learner_conf_entry;
+      PRAFT.GetConfigurationByIndex(last_log_index, &conf_entry, &learner_conf_entry);
+      new_meta.clear_peers();
+      for (auto iter = conf_entry.conf.begin(); iter != conf_entry.conf.end(); ++iter) {
+        *new_meta.add_peers() = iter->to_string();
+      }
+      new_meta.clear_old_peers();
+      for (auto iter = conf_entry.old_conf.begin(); iter != conf_entry.old_conf.end(); ++iter) {
+        *new_meta.add_old_peers() = iter->to_string();
+      }
+      new_meta.clear_learners();
+      for (auto iter = learner_conf_entry.conf.begin(); iter != learner_conf_entry.conf.end(); ++iter) {
+        *new_meta.add_learners() = iter->to_string();
+      }
       INFO("Succeed to fix db_{} snapshot meta: {}, {}", db_id, last_log_index, last_log_term);
 
       auto rc = snapshot_meta_memtable.save_to_file(fs, meta_path);