[MESOS-7546] WAIT_NESTED_CONTAINER sometimes returns 404

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 1.2.0, 1.3.0
• Fix Version/s: 1.2.2, 1.3.1
• Component/s: containerization
• Labels: containerizer mesosphere
• Sprint: Mesosphere Sprint 57
• Story Points: 3
• Assignee: Gastón Kleiman
• Reporter: Gastón Kleiman
• Shepherd: Vinod Kone
• Created: 22/May/17 23:26
• Updated: 30/May/17 21:06
• Resolved: 30/May/17 21:06

Description

WAIT_NESTED_CONTAINER sometimes returns 404s even though the nested container has already exited and the parent task/executor is still running.

This happens when an agent uses more than one containerizer (e.g.,
docker,mesos, WAIT_NESTED_CONTAINER and the exit status of the nested container has already been checkpointed.

The root cause of this is a bug in the ComposingContainerizer in the following lines: https://github.com/apache/mesos/blob/1c7ffbeb505b3f5ab759202195f0b946a20cb803/src/slave/containerizer/composing.cpp#L620-L628