Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Duplicate
-
Impala 2.2
-
None
-
None
Description
Some times repeated runs fail at test_spilling. It seems that Impalad crashes. The error from .FATAL:
F0616 14:40:45.102733 19161 buffered-block-mgr.cc:261] Check failed: client->num_tmp_reserved_buffers_ == 0 (5 vs. 0)
This is in:
bool BufferedBlockMgr::TryAcquireTmpReservation(Client* client, int num_buffers) {
TestSpilling.test_spilling[exec_option: {'disable_codegen': False, 'abort_on_error': 1, 'exec_single_node_rows_threshold': 0, 'batch_size': 0, 'num_nodes': 0} | table_format: parquet/none] self = <test_spillling.TestSpilling object at 0x23c0ad0> vector = <tests.common.test_vector.TestVector object at 0x2217ad0> @pytest.mark.execute_serially @CustomClusterTestSuite.with_args( impalad_args="--read_size=200000", catalogd_args="--load_catalog_in_background=false") def test_spilling(self, vector): new_vector = deepcopy(vector) # remove this. the test cases set this explicitly. del new_vector.get_value('exec_option')['num_nodes'] > self.run_test_case('QueryTest/spilling', new_vector) > raise ImpalaBeeswaxException(self.__build_error_message(u), u) E ImpalaBeeswaxException: ImpalaBeeswaxException: E INNER EXCEPTION: <class 'socket.error'> E MESSAGE: [Errno 104] Connection reset by peer
The last query executed:
-- executing against localhost:21000 set max_block_mgr_memory=100m; -- executing against localhost:21000 select count(l1.l_tax) from lineitem l1, lineitem l2, lineitem l3 where l1.l_tax < 0.01 and l2.l_tax < 0.04 and l1.l_orderkey = l2.l_orderkey and l1.l_orderkey = l3.l_orderkey and l1.l_comment = l3.l_comment and l1.l_shipdate = l3.l_shipdate;
Attachments
Issue Links
- is related to
-
IMPALA-2099 BufferedBlockMgr DCHECK client->num_tmp_reserved_buffers_ == 0 failure w/ RM enabled
- Resolved
- relates to
-
IMPALA-2245 Check failed: client->num_tmp_reserved_buffers_ == 0 (1 vs. 0)
- Resolved