Summary: (#1313)

If log recycling is enabled with the rocksdb (recycle_log_file_num=16)
 db->Writebatch is erroring out with keynotfound after ~5-6 hours of run
 (1M seq but can happen to any workload I guess).See my detailed bug
 report here (https://github.com/facebook/rocksdb/issues/1303).
 This commit is the fix for this, a check is been added not to delete
 the log file if it is already there in the recycle list.

Test Plan:
 Unit tested it and ran the similar profile. Not reproducing anymore.

diff --git a/db/db_impl.cc b/db/db_impl.cc
index 5b4ad092a8334188c6fea25b9929212f1e2afb1b..e2c26bf5ebc753abb59d5a6ea392012780fb12d5 100644 (file)
--- a/db/db_impl.cc
+++ b/db/db_impl.cc
@@ -812,38 +812,6 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
   job_context->prev_log_number = versions_->prev_log_number();
 
   versions_->AddLiveFiles(&job_context->sst_live);
-  if (doing_the_full_scan) {
-    for (size_t path_id = 0; path_id < db_options_.db_paths.size(); path_id++) {
-      // set of all files in the directory. We'll exclude files that are still
-      // alive in the subsequent processings.
-      std::vector<std::string> files;
-      env_->GetChildren(db_options_.db_paths[path_id].path,
-                        &files);  // Ignore errors
-      for (std::string file : files) {
-        // TODO(icanadi) clean up this mess to avoid having one-off "/" prefixes
-        job_context->full_scan_candidate_files.emplace_back(
-            "/" + file, static_cast<uint32_t>(path_id));
-      }
-    }
-
-    //Add log files in wal_dir
-    if (db_options_.wal_dir != dbname_) {
-      std::vector<std::string> log_files;
-      env_->GetChildren(db_options_.wal_dir, &log_files);  // Ignore errors
-      for (std::string log_file : log_files) {
-        job_context->full_scan_candidate_files.emplace_back(log_file, 0);
-      }
-    }
-    // Add info log files in db_log_dir
-    if (!db_options_.db_log_dir.empty() && db_options_.db_log_dir != dbname_) {
-      std::vector<std::string> info_log_files;
-      // Ignore errors
-      env_->GetChildren(db_options_.db_log_dir, &info_log_files);
-      for (std::string log_file : info_log_files) {
-        job_context->full_scan_candidate_files.emplace_back(log_file, 0);
-      }
-    }
-  }
 
   if (!alive_log_files_.empty()) {
     uint64_t min_log_number = job_context->log_number;
@@ -883,6 +851,59 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     assert(!logs_.empty());
   }
 
+  if (doing_the_full_scan) {
+    for (size_t path_id = 0; path_id < db_options_.db_paths.size(); path_id++) {
+      // set of all files in the directory. We'll exclude files that are still
+      // alive in the subsequent processings.
+      std::vector<std::string> files;
+      env_->GetChildren(db_options_.db_paths[path_id].path,
+                        &files);  // Ignore errors
+      for (std::string file : files) {
+        // TODO(icanadi) clean up this mess to avoid having one-off "/" prefixes
+        job_context->full_scan_candidate_files.emplace_back(
+            "/" + file, static_cast<uint32_t>(path_id));
+      }
+    }
+
+    //Add log files in wal_dir
+    if (db_options_.wal_dir != dbname_) {
+      std::vector<std::string> log_files;
+      env_->GetChildren(db_options_.wal_dir, &log_files);  // Ignore errors
+      InfoLogPrefix info_log_prefix(!db_options_.db_log_dir.empty(), dbname_);
+      for (std::string log_file : log_files) {
+        uint64_t number;
+        FileType type;
+        // Ignore file if we cannot recognize it.
+        if (!ParseFileName(log_file, &number, info_log_prefix.prefix, &type)) {
+          Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
+              "Unrecognized log file %s \n",log_file.c_str());
+          continue;
+        }
+        // If the log file is already in the log recycle list , don't put
+        // it in the candidate list.
+        if (std::find(log_recycle_files.begin(), log_recycle_files.end(),number) != 
+            log_recycle_files.end()) {
+
+          Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
+              "Log %" PRIu64 " Already added in the recycle list, skipping.\n", 
+              number);
+          continue;
+        }
+
+        job_context->full_scan_candidate_files.emplace_back(log_file, 0);
+      }
+    }
+    // Add info log files in db_log_dir
+    if (!db_options_.db_log_dir.empty() && db_options_.db_log_dir != dbname_) {
+      std::vector<std::string> info_log_files;
+      // Ignore errors
+      env_->GetChildren(db_options_.db_log_dir, &info_log_files);
+      for (std::string log_file : info_log_files) {
+        job_context->full_scan_candidate_files.emplace_back(log_file, 0);
+      }
+    }
+  }
+
   // We're just cleaning up for DB::Write().
   assert(job_context->logs_to_free.empty());
   job_context->logs_to_free = logs_to_free_;