[MESOS-3351] duplicated slave id in master after master failover

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.25.0
• Component/s: master
• Labels: race-condition uuid

Environment

Mac OS (Darwin da-macbookair.cn.ibm.com 14.5.0 Darwin Kernel Version 14.5.0: Wed Jul 29 02:26:53 PDT 2015; root:xnu-2782.40.9~1/RELEASE_X86_64 x86_64)

• Target Version/s: 0.25.0
• Flags: Patch
• Assignee: Klaus Ma
• Reporter: Klaus Ma
• Shepherd: Vinod Kone
• Created: 01/Sep/15 06:21
• Updated: 18/Sep/15 23:08
• Resolved: 18/Sep/15 23:08

Description

When a slave register to master, master will generate a slave ID for it by masterInfo.id + "-S" + nextSlaveId (in master.cpp) to avoid duplicate slaveId. But if master failover, nextSlaveId was reset to 0 which may trigger duplicated slaveId between old slave & new slave.

For now, it's only reproduced in Mac OS unstably, and can NOT reproduce in Ubuntu; not sure the other OS.

Issue Links

• blocks: MESOS-3070 Master CHECK failure if a framework uses duplicated task id.