[AURORA-1878] Increased executor logs can lead to task's running out of disk space

• Type: Task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.17.0
• Component/s: Executor
• Assignee: Joshua Cohen
• Reporter: Joshua Cohen
• Created: 11/Jan/17 17:20
• Updated: 01/Feb/17 07:54
• Resolved: 11/Jan/17 22:20

Description

After the health check for updates patch, this log statement is being emitted once every 500ms: https://github.com/apache/aurora/commit/2992c8b4#diff-6d60c873330419a828fb992f46d53372R121

This is due to this code:

        if status_result is not None:
          log.info('%s reported %s' % (status_checker.__class__.__name__, status_result))

Previously, status_result would be None unless the status checker had a terminal event. Now, status_result will always be set, but we only consider the status_result to be terminal if the status is not TASK_STARTING or TASK_RUNNING. So, for the healthy case, we log that the task is TASK_RUNNING every 500ms.