[HAMA-909] Improve Mesos Scheduler's Fault Tolerance

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.7.0
• Assignee: Jeffrey Ryan Fenchel
• Reporter: Jeffrey Ryan Fenchel
• Created: 03/Jun/14 05:19
• Updated: 05/Jun/14 01:03
• Resolved: 04/Jun/14 03:46

Description

There are several updates to address issues with the recent Mesos integration:
1) Use thread-safe collections in the Mesos scheduler where necessary.
2) Add handling for recovery tasks.
3) Improve cleanup of Mesos tasks for job failures and user initiated job terminations.