*MongoDB 3.2 Only*: Aggregation with self join can cause deadlock on cursor timeout

XMLWordPrintableJSON

    • Type: Task
    • Resolution: Won't Fix
    • Priority: Major - P3
    • None
    • Affects Version/s: 3.2.20
    • Component/s: None
    • None
    • Query
    • None
    • 3
    • None
    • None
    • None
    • None
    • None
    • None

      The stack trace produced by a self-deadlock:

      __lll_lock_wait
      _L_lock_854
      pthread_mutex_lock
      mongo::CursorManager::eraseCursor
      mongo::GlobalCursorIdCache::eraseCursor
      mongo::CursorManager::eraseCursorGlobalIfAuthorized
      mongo::receivedKillCursors
      mongo::assembleResponse
      mongo::DBDirectClient::say
      mongo::DBClientBase::killCursor
      mongo::DBClientCursor::kill
      mongo::DBClientCursor::~DBClientCursor
      mongo::DBClientCursor::~DBClientCursor
      mongo::DocumentSourceLookUp::~DocumentSourceLookUp
      mongo::DocumentSourceLookUp::~DocumentSourceLookUp
      std::deque<...>::_M_destroy_data_aux
      std::deque<...>::~deque
      mongo::Pipeline::~Pipeline
      mongo::PipelineProxyStage::~PipelineProxyStage
      mongo::PipelineProxyStage::~PipelineProxyStage
      mongo::PlanExecutor::~PlanExecutor
      mongo::ClientCursor::~ClientCursor
      mongo::CursorManager::timeoutCursors
      mongo::GlobalCursorIdCache::timeoutCursors
      mongo::ClientCursorMonitor::run
      mongo::BackgroundJob::jobBody
      execute_native_thread_routine
      start_thread
      clone
      

            Assignee:
            Backlog - Query Team (Inactive)
            Reporter:
            James Wahlin
            Votes:
            0 Vote for this issue
            Watchers:
            8 Start watching this issue

              Created:
              Updated:
              Resolved: