[MESOS-1572] Consider automatic restarting of slaves/masters with soft/hard cpu lockups.

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: agent, master
• Labels: reliability repair
• Assignee: Unassigned
• Reporter: Benjamin Mahler
• Created: 09/Jul/14 00:58
• Updated: 09/Jul/14 01:25

Description

On Linux systems, when a soft or hard CPU lockup occurs on slaves, we have observed strange and undesirable things occur possibly due to kernel bugs.

With root access and sysctl, it should be possible to configure an automatic reboot of the machine when a soft/hard lockup occurs:
https://www.kernel.org/doc/Documentation/lockup-watchdogs.txt