[SLIDER-309] add functional tests of sliding-window failure handling

• Type: Sub-task
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: Slider 0.40
• Component/s: appmaster, test
• Assignee: Unassigned
• Reporter: Steve Loughran
• Created: 12/Aug/14 13:29
• Updated: 03/Nov/14 19:00

Description

There's some testing of the sliding window logic at the mock level, and of configurable thresholds in the HBase provider.

But not in the agent. We need

1. functional test to verify that a failure rate less than (threshold/window-size) is acceptable.
2. test to verify window is reset on AM failure (manually verified)
3. test to verify that a cluster flex resets the values and allows the threshold to be changed

Issue Links

• relates to: SLIDER-77 use a window for tracking container failures