Hello,

I discovered this bug yesterday. The simple fix, if your planning to stay 
on CoreOS alpha with Docker v1.12, is to remove the word 'daemon' from 
'/etc/systemd/system/docker.service.d/40-flannel.conf'.

Basically, if you setup kubernetes from scripts in the 
github.com/coreos/coreos-kubernetes repo on CoreOS,\ between the ~13 day 
window of PR #551 and #642(the fix) then rolling to CoreOS v1151.0.0+ will 
break your docker.service until you modify the drop-in.

The issue is that PR #551 added a drop-in to docker.service that overrides 
the `ExecStart` line to drop some parameters (for a different network 
configuration style). The problem is that the command line parameters to 
start dockerd changed between Docker v1.11 and v1.12 and v1.12 baulks at 
the use of the daemon argument. If the drop-in isn't present, CoreOS would 
just update the ExecStart line across versions correctly via the included 
unit file in /usr/lib.... Its really my fault for introducing a dependance 
on the ExecStart line remaining correct when I just needed to override two 
environment variables.

The more correct fix, that works for both Docker v1.11 and v1.12, is 
here: https://github.com/coreos/coreos-kubernetes/pull/642

Hope this helps, let me know if you have further questions.

Regards,

Patrick

On Friday, August 26, 2016 at 6:56:05 AM UTC-7, [email protected] wrote:
>
> Hi All,
>
> I'm going through the Vagrant instructions for running a single-machine 
> and I'm hitting issues. It appears that several services are not starting 
> correctly [1] and I waited awhile for all the downloads to happen like the 
> instructions say [2] and then I looked at docker service status [3] and 
> logs and the service was not able to start. Has anyone else see anything 
> like this?
>
> Thanks,
> Jason
>
> [1]
> $ vagrant 
> ssh                                                                           
>                                                                         
> master 1876aac ✗
> CoreOS alpha (1151.0.0)
> Last login: Fri Aug 26 13:48:26 2016 from 10.0.2.2
> Update Strategy: No Reboots
> Failed Units: 3
>   docker.service
>   update-engine.service
>   docker.socket
>
> [2] 
> https://coreos.com/kubernetes/docs/latest/kubernetes-on-vagrant-single.html
>
> [3] 
> core@localhost ~ $ systemctl status docker.service
> ● docker.service - Docker Application Container Engine
>    Loaded: loaded (/usr/lib/systemd/system/docker.service; disabled; 
> vendor preset: disabled)
>   Drop-In: /etc/systemd/system/docker.service.d
>            └─40-flannel.conf
>    Active: failed (Result: start-limit-hit) since Fri 2016-08-26 13:37:06 
> UTC; 12min ago
>      Docs: http://docs.docker.com
>   Process: 5031 ExecStart=/usr/lib/coreos/dockerd daemon --host=fd:// 
> $DOCKER_OPTS $DOCKER_CGROUPS $DOCKER_OPT_MTU (code=exited, status=1/FAILURE)
>  Main PID: 5031 (code=exited, status=1/FAILURE)
>
> Aug 26 13:37:06 localhost systemd[1]: Starting Docker Application 
> Container Engine...
> Aug 26 13:37:06 localhost dockerd[5031]: dockerd: "dockerd" requires 0 
> arguments.
> Aug 26 13:37:06 localhost dockerd[5031]: Usage:        dockerd [OPTIONS]
> Aug 26 13:37:06 localhost systemd[1]: docker.service: Main process exited, 
> code=exited, status=1/FAILURE
> Aug 26 13:37:06 localhost systemd[1]: Failed to start Docker Application 
> Container Engine.
> Aug 26 13:37:06 localhost systemd[1]: docker.service: Unit entered failed 
> state.
> Aug 26 13:37:06 localhost systemd[1]: docker.service: Failed with result 
> 'exit-code'.
> Aug 26 13:37:06 localhost systemd[1]: docker.service: Start request 
> repeated too quickly.
> Aug 26 13:37:06 localhost systemd[1]: Failed to start Docker Application 
> Container Engine.
> Aug 26 13:37:06 localhost systemd[1]: docker.service: Failed with result 
> 'start-limit-hit'.
>

Reply via email to