Hi, I just ran into this issue when I was trying to stop and recreate my Network Agents today so they using agent-instancev:0.8.3, is there any suggestions apart from a kernel update to fix the issue?
I managed to get 1/3 hosts updraded, but the other 2 hosts are proving to be troublesome to update as the Docker Daemon keeps getting stuck.
What version of Rancher are you running? What version of Docker and what OS are you using?
This is actually a Docker issue and not Rancher specific. We are just good at triggering the Docker issue. 
We are just good at triggering the Docker issue.
No way they can ignore it as a fluke 
We’re on v1.1.3 and the Ubuntu 14.04 (kernel: 3.13.0-46-generic) Docker 1.9.1/1.12.1
We have 3 hosts, with network agents running v0.4.1.
I have one host that is on Docker 1.12.1 and it restarted fine, but we currently can’t invest the time to wait for 1.9.1 → 1.12.1 to rebuild all the image layers 
Trying to restart the Network Agents breaks the hosts on 1.9.1 so we had to hold back on updating.
EDIT:
Also for some reason purging the Network Agent on the host with docker 1.12.1 didn’t seem to recreate it, we had to relaunch the agent on that host. I’m guessing its probably something to do with the agent-instance version being too old.
