[YETUS-570] Report and optionally kill stale JVMs between unit test modules

• Type: New Feature
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.6.0
• Fix Version/s: 0.7.0
• Component/s: Precommit

Release Note

<!-- markdown -->

This change enables two new features:

* A thread monitor to help determine how `--proclimit` should be configured.

* A process reaper to automatically eliminate processes that match certain names after some operations to limit the impact of "stuck" unit tests.

• Assignee: Allen Wittenauer
• Reporter: Allen Wittenauer
• Created: 27/Oct/17 14:24
• Updated: 03/Nov/17 03:15
• Resolved: 03/Nov/17 03:15

Description

YETUS-561 does a great job of preventing fork bombs and stale processes from destroying machines. However, Yetus should do a better job of:

a) Reporting when that happens
b) Preventing that from happening

Issue Links

• relates to: HDFS-12711 deadly hdfs test
• requires: YETUS-561 Ability to limit user process counts and Docker container's RAM usage