[IMPALA-4463] Hang in test_mini_stress.py with MT_DOP=3

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Duplicate
• Affects Version/s: Impala 2.8.0
• Component/s: Backend
• Labels: multithreading
• Target Version: Impala 2.9.0
• Assignee: Alexander Behm
• Reporter: Alexander Behm
• Created: 10/Nov/16 06:23
• Updated: 16/Feb/17 01:59
• Resolved: 16/Feb/17 01:59

Description

A Jenkins run hung while running test_mini_stress.py::test_mini_stress.

Observations at the time of the hang:

• There are a total of 64 FE connections in use.
• There are 15 active queries that returned results. Their latest timeline event is "UnregisterQuery" from which there is no more progress.

Attached:

• Backtraces of all threads of the impalad processes at the time of the hang
• Raw JSON of the impalad metrics at the time of the hang

I have not yet tried to reproduce locally. This is my first run of this test in this configuration, so I assume it should be easy to reproduce.

bin/start-impala-cluster.py --impalad_args="--default_query_options=mt_dop=3"
tests/run-tests.py stress/test_mini_stress.py