[MESOS-6118] Agent would crash with docker container tasks due to host mount table read.

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Pending Closed
• Affects Version/s: 1.0.1
• Fix Version/s: 1.0.2, 1.1.0
• Component/s: agent
• Labels: linux slave

Environment

Build: 2016-08-26 23:06:27 by centos
Version: 1.0.1
Git tag: 1.0.1
Git SHA: 3611eb0b7eea8d144e9b2e840e0ba16f2f659ee3
systemd version `219` detected
Inializing systemd state
Created systemd slice: `/run/systemd/system/mesos_executors.slice`
Started systemd slice `mesos_executors.slice`
Using isolation: posix/cpu,posix/mem,filesystem/posix,network/cni
Using /sys/fs/cgroup/freezer as the freezer hierarchy for the Linux launcher

Linux ip-10-254-192-40 3.10.0-327.28.3.el7.x86_64 #1 SMP Thu Aug 18 19:05:49 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux

• Target Version/s: 1.0.2, 1.1.0
• Assignee: Kevin Klues
• Reporter: Jamie Briant
• Shepherd: Jie Yu
• Created: 02/Sep/16 00:08
• Updated: 21/May/20 19:21
• Resolved: 17/Oct/16 22:06

Description

I have a framework which schedules thousands of short running (a few seconds to a few minutes) of tasks, over a period of several minutes. In 1.0.1, the slave process will crash every few minutes (with systemd restarting it).

Crash is:

Sep 01 20:52:23 ip-10-254-192-99 mesos-slave: F0901 20:52:23.905678 1232 fs.cpp:140] Check failed: !visitedParents.contains(parentId)
Sep 01 20:52:23 ip-10-254-192-99 mesos-slave: *** Check failure stack trace: ***

Version 1.0.0 works without this issue.

Issue Links

• Blocked: MESOS-10131 Agent frequently dies with error "Cycle found in mount table hierarchy"
• is related to: MESOS-5969 Linux 'MountInfoTable' entries not sorted as expected