Uploaded image for project: 'IMPALA'
  1. IMPALA
  2. IMPALA-4541

test_codegen_mem_limit is flaky

    XMLWordPrintableJSON

Details

    • Bug
    • Status: Resolved
    • Blocker
    • Resolution: Fixed
    • Impala 2.8.0
    • Impala 2.8.0
    • Backend

    Description

      The test seems to exceed memory limit at another location so it doesn't consistently fail due to exceeding memory limit during codegen.
      The exhaustive RHEL7 build hit the following failure.

      Error Message
      
      query_test/test_query_mem_limit.py:123: in test_codegen_mem_limit     self.run_test_case('QueryTest/codegen-mem-limit', vector) common/impala_test_suite.py:324: in run_test_case     self.__verify_exceptions(test_section['CATCH'], str(e), use_db) common/impala_test_suite.py:215: in __verify_exceptions     (expected_str, actual_str) E   AssertionError: Unexpected exception string. Expected: Codegen failed to reserve E   Not found in actual: ImpalaBeeswaxException: Query aborted:Memory limit exceededFailed to allocate tuple bufferMemory Limit Exceeded by fragment: a44539d48545d15b:2  HDFS_SCAN_NODE (id=0) could not allocate 90.00 KB without exceeding limit.Query(a44539d48545d15b:ec0ea61000000000): Limit=100.00 KB Total=64.42 KB Peak=64.42 KB  Fragment a44539d48545d15b:2: Total=64.42 KB Peak=64.42 KB    HDFS_SCAN_NODE (id=0): Total=55.00 KB Peak=55.00 KB      Exprs: Total=8.00 KB Peak=8.00 KB    DataStreamSender (dst_id=1): Total=1.42 KB Peak=1.42 KB  Block Manager: Total=0 Peak=0
      Stacktrace
      
      query_test/test_query_mem_limit.py:123: in test_codegen_mem_limit
          self.run_test_case('QueryTest/codegen-mem-limit', vector)
      common/impala_test_suite.py:324: in run_test_case
          self.__verify_exceptions(test_section['CATCH'], str(e), use_db)
      common/impala_test_suite.py:215: in __verify_exceptions
          (expected_str, actual_str)
      E   AssertionError: Unexpected exception string. Expected: Codegen failed to reserve
      E   Not found in actual: ImpalaBeeswaxException: Query aborted:Memory limit exceededFailed to allocate tuple bufferMemory Limit Exceeded by fragment: a44539d48545d15b:2  HDFS_SCAN_NODE (id=0) could not allocate 90.00 KB without exceeding limit.Query(a44539d48545d15b:ec0ea61000000000): Limit=100.00 KB Total=64.42 KB Peak=64.42 KB  Fragment a44539d48545d15b:2: Total=64.42 KB Peak=64.42 KB    HDFS_SCAN_NODE (id=0): Total=55.00 KB Peak=55.00 KB      Exprs: Total=8.00 KB Peak=8.00 KB    DataStreamSender (dst_id=1): Total=1.42 KB Peak=1.42 KB  Block Manager: Total=0 Peak=0
      Standard Error
      
      -- connecting to: localhost:21000
      -- executing against localhost:21000
      use functional_parquet;
      
      SET disable_codegen=True;
      SET abort_on_error=1;
      SET exec_single_node_rows_threshold=0;
      SET batch_size=0;
      SET num_nodes=0;
      -- executing against localhost:21000
      set mem_limit=100k;
      
      -- executing against localhost:21000
      
      select *
      from alltypes
      where substr(string_col, 1) = "";
      
      -- executing against localhost:21000
      SET MEM_LIMIT=0;;
      
      

      Attachments

        Activity

          People

            tarmstrong Tim Armstrong
            kwho Michael Ho
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved: