[MESOS-1649] Network isolator should tolerate slave crashes while doing isolate/cleanup.

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.20.0
• Epic Link: Container Network Monitoring
• Sprint: Q3 Sprint 2
• Story Points: 3
• Assignee: Jie Yu
• Reporter: Jie Yu
• Created: 29/Jul/14 05:43
• Updated: 06/Aug/14 16:36
• Resolved: 01/Aug/14 17:54

Description

A slave may crash while we are installing/removing filters. The slave recovery for the network isolator should tolerate those partially installed filters. Also, we want to avoid leaking a filter on host eth0 and host lo.

The current code cannot tolerate that, thus may cause the following error:

Failed to perform recovery: Collect failed: Failed to recover container d409a100-2afb-497c-864f-fe3002cf65d9 with pid 50405: No ephemeral ports found
To remedy this do as follows:
Step 1: rm -f /var/lib/mesos/meta/slaves/latest
       This ensures slave doesn't recover old live executors.
Step 2: Restart the slave.