[IMPALA-11379] perf-AB-test Jenkins job fails with "Working copy is dirty"

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: Impala 4.2.0
• Fix Version/s: Impala 4.2.0
• Component/s: Infrastructure
• Target Version: Impala 4.2.0
• Epic Color: ghx-label-2
• Assignee: Joe McDonnell
• Reporter: Joe McDonnell
• Created: 21/Jun/22 23:20
• Updated: 02/Jul/22 03:27
• Resolved: 23/Jun/22 00:23

Description

When using the perf-AB-test job on upstream Jenkins, it fails with this message:

14:50:56 + [[ ! -v HASH_B ]]
14:50:56 + [[ ! -v IMPALAD_ARGS ]]
14:50:56 + IMPALAD_ARGS=
14:50:56 + ./bin/single_node_perf_run.py --iterations 10 --scale 42 --table_formats parquet/none --workload tpch --num_impalads 3 --query_names '.*' --load --start_minicluster --impalad_args '' 66cde5ba8f707e61b1b46dabe114819de8b4f75d 06b1db4675b509b9461f7489d81a2c74930aa413
14:50:56  M shell/ext-py/kerberos-1.3.1/kerberos.egg-info/PKG-INFO
14:50:58 Traceback (most recent call last):
14:50:58   File "./bin/single_node_perf_run.py", line 359, in <module>
14:50:58     main()
14:50:58   File "./bin/single_node_perf_run.py", line 340, in main
14:50:58     raise Exception("Working copy is dirty. Consider 'git stash' and try again.")
14:50:58 Exception: Working copy is dirty. Consider 'git stash' and try again.

This is comparing the latest two master commits, and it seems to reproduce in general for everything. We should figure out how to fix this and add some check in GVO to prevent it in future.