Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Cannot Reproduce
-
Impala 2.3.0
-
None
Description
Many test failures. Many of them look like the impalad died, but I can't find signs of this in the logs (probably missing something). The rest are due to the sorter not having enough memory, e.g.:
E ImpalaBeeswaxException: ImpalaBeeswaxException: E Query aborted: E Memory limit exceeded E Query did not have enough memory to get the minimum required buffers in the block manager. E E E E Memory Limit Exceeded E Query(d4094dc19c26e3c:b320dcd0712d6396) Limit: Consumption=8.04 MB E Fragment d4094dc19c26e3c:b320dcd0712d6397: Consumption=16.00 KB E EXCHANGE_NODE (id=14): Consumption=0 E DataStreamRecvr: Consumption=8.00 KB E Block Manager: Limit=80.64 GB Consumption=0 E Fragment d4094dc19c26e3c:b320dcd0712d6398: Consumption=8.03 MB E SORT_NODE (id=7): Consumption=0 E AGGREGATION_NODE (id=13): Consumption=8.01 MB E EXCHANGE_NODE (id=12): Consumption=0 E DataStreamSender: Consumption=8.00 KB E WARNING: The following tables are missing relevant table and/or column statistics. E functional_hbase.alltypesagg,functional_seq_snap.alltypes
There is one other failure:
select count(*) over () from functional.alltypestiny where month = 1 and bool_col = sleep(1000);
query_test/test_rows_availability.py:57: in test_rows_availability self.client.QUERY_STATES['FINISHED'], timeout=20) common/impala_service.py:162: in wait_for_query_state assert target_state == query_state, 'Did not reach query state in time' E AssertionError: Did not reach query state in time