Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
Impala 2.3.0
-
None
Description
/data/2/jenkins/workspace/impala-CDH5-nightly-data-load/repos/Impala/be/src/exprs/expr-test.cc:397 Value of: *reinterpret_cast<bool*>(result) Actual: false Expected: expected_result Which is: true unix_timestamp() between 1446367138 and 1446367148
Here is the relevant expr-test code (not sure exactly which of the two statements failed):
// Test that the epoch is reasonable. Allow a few seconds to compensate for execution // time. int tolerance_in_seconds = 5; time_t unix_time = (posix_time::microsec_clock::local_time() - from_time_t(0)) .total_seconds(); stringstream expr_sql; expr_sql << "unix_timestamp() between " << unix_time - tolerance_in_seconds << " and " << unix_time + tolerance_in_seconds; TestValue(expr_sql.str(), TYPE_BOOLEAN, true); { ScopedLocalUnixTimestampConversionOverride use_local; unix_time = time(NULL); expr_sql.str(""); expr_sql << "unix_timestamp() between " << unix_time - tolerance_in_seconds << " and " << unix_time + tolerance_in_seconds; TestValue(expr_sql.str(), TYPE_BOOLEAN, true); }
My guess is that there was some hiccup in the VM causing it to hang for > 5 seconds, but I don't know if there's any way to verify this. I can't find any BE test logs (maybe I don't know where to look...). Marking as a flaky test for now.