[MESOS-4417] Prevent allocator from crashing on successful recovery.

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 0.27.0
• Component/s: allocation
• Labels: mesosphere
• Target Version/s: 0.27.0
• Epic Link: Quota
• Sprint: Mesosphere Sprint 27
• Story Points: 3
• Assignee: Alex R
• Reporter: Alex R
• Shepherd: Joris Van Remoortere
• Created: 17/Jan/16 08:43
• Updated: 26/Nov/18 12:21
• Resolved: 26/Nov/18 12:21

Description

There might be a bug that may crash the master as pointed out by bmahler in https://reviews.apache.org/r/42222/:

It looks like if we trip the resume call in addSlave, this delayed resume will crash the master 
due to the CHECK(paused) that currently resides in resume.