[MESOS-4449] SegFault on agent during executor startup

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 0.26.0
• Fix Version/s: 0.26.1, 0.27.0
• Labels: mesosphere

Environment

Some setup details:

• Master and agents running in separate docker containers on the same host.
• Containers based upon Ubuntu 14.04 using Mesosphere produced Mesos deb files. For more details see (https://github.com/ContainerSolutions/minimesos-docker)
• This only occurs with 0.26, not with 0.25.

• Target Version/s: 0.27.0
• Sprint: Mesosphere Sprint 27
• Story Points: 1
• Assignee: Anand Mazumdar
• Reporter: Philip Winder
• Shepherd: Joris Van Remoortere
• Created: 21/Jan/16 16:16
• Updated: 26/Nov/18 12:21
• Resolved: 26/Nov/18 12:21

Description

When repeatedly performing our system tests we have found that we get a segfault on one of the agents. It probably occurs about one time in ten. I have attached the full log from that agent. I've attached the log from the agent that failed and the master (although I think this is less helpful).

To reproduce

• I have no idea. It seems to occur at certain times. E.g. like if a packet is created right on a minute boundary or something. But I don't think it's something caused by our code because the timestamps are stamped by mesos. I was surprised not to find a bug already open.

Issue Links

• is related to: MESOS-4451 Enable `-Wnull-dereference` when building Mesos