fix disk manager thread conflict

conf/nebula-storaged.conf.default

@@ -54,6 +54,9 @@
 # One path per Rocksdb instance.
 --data_path=data/storage
 
+# Minimum reserved bytes of each data path
+--minimum_reserved_bytes=268435456
+
 # The default reserved bytes for one batch operation
 --rocksdb_batch_size=4096
 # The default block cache size used in BlockBasedTable.

conf/nebula-storaged.conf.production

@@ -54,6 +54,9 @@
 # One path per Rocksdb instance.
 --data_path=data/storage
 
+# Minimum reserved bytes of each data path
+--minimum_reserved_bytes=268435456
+
 # The default reserved bytes for one batch operation
 --rocksdb_batch_size=4096
 # The default block cache size used in BlockBasedTable. (MB)

src/kvstore/DiskManager.cpp

@@ -48,8 +48,8 @@ StatusOr<std::vector<std::string>> DiskManager::path(GraphSpaceID spaceId) {
         return Status::Error("Space not found");
     }
     std::vector<std::string> paths;
-    for (const auto& [path, _] : spaceIt->second) {
-        paths.emplace_back(path);
+    for (const auto& partEntry : spaceIt->second) {
+        paths.emplace_back(partEntry.first);
     }
     return paths;
 }
@@ -70,6 +70,7 @@ StatusOr<std::string> DiskManager::path(GraphSpaceID spaceId, PartitionID partId
 void DiskManager::addPartToPath(GraphSpaceID spaceId,
                                 PartitionID partId,
                                 const std::string& path) {
+    std::lock_guard<std::mutex> lg(lock_);
     try {
         auto canonical = boost::filesystem::canonical(path);
         auto dataPath = canonical.parent_path().parent_path();
@@ -85,6 +86,7 @@ void DiskManager::addPartToPath(GraphSpaceID spaceId,
 void DiskManager::removePartFromPath(GraphSpaceID spaceId,
                                      PartitionID partId,
                                      const std::string& path) {
+    std::lock_guard<std::mutex> lg(lock_);
     try {
         auto canonical = boost::filesystem::canonical(path);
         auto dataPath = canonical.parent_path().parent_path();

src/kvstore/DiskManager.h

@@ -65,6 +65,9 @@ class DiskManager {
 
     // the index in dataPaths_ for a given space + part
     std::unordered_map<GraphSpaceID, std::unordered_map<PartitionID, size_t>> partIndex_;
+
+    // lock used to protect partPath_ and partIndex_
+    std::mutex lock_;
 };
 
 }  // namespace kvstore

src/kvstore/NebulaStore.cpp

@@ -338,7 +338,7 @@ std::shared_ptr<Part> NebulaStore::newPart(GraphSpaceID spaceId,
                                            KVEngine* engine,
                                            bool asLearner,
                                            const std::vector<HostAddr>& defaultPeers) {
-    auto walPath = folly::stringPrintf("%s/wal/%d", engine->getDataRoot(), partId);
+    auto walPath = folly::stringPrintf("%s/wal/%d", engine->getWalRoot(), partId);
     auto part = std::make_shared<Part>(spaceId,
                                        partId,
                                        raftAddr_,

src/kvstore/raftex/RaftPart.cpp

@@ -768,7 +768,7 @@ void RaftPart::appendLogsInternal(AppendLogsIterator iter, TermID termId) {
     } while (false);
 
     if (!checkAppendLogResult(res)) {
-        LOG(ERROR) << idStr_ << "Failed append logs";
+        LOG_EVERY_N(WARNING, 100) << idStr_ << "Failed to write wal";
         return;
     }
     // Step 2: Replicate to followers
@@ -805,7 +805,7 @@ void RaftPart::replicateLogs(folly::EventBase* eb,
     } while (false);
 
     if (!checkAppendLogResult(res)) {
-        LOG(ERROR) << idStr_ << "Replicate logs failed";
+        LOG(WARNING) << idStr_ << "replicateLogs failed because of not leader or term changed";
         return;
     }
 
@@ -911,7 +911,9 @@ void RaftPart::processAppendLogResponses(
         } while (false);
 
         if (!checkAppendLogResult(res)) {
-            LOG(ERROR) << idStr_ << "processAppendLogResponses failed!";
+            LOG(WARNING)
+                << idStr_
+                << "processAppendLogResponses failed because of not leader or term changed";
             return;
         }