Hello Cory-

There are a number of errors about the mounts which should be non-fatal:

Aug 09 15:40:08 ip-172-31-16-118.ec2.internal dockerd[1107]:
time="2016-08-09T15:40:08.520711349Z" level=error msg="Clean up Error!
Cannot destroy container
26537dcd69eeb25dcd3a4867f81db909e5ca992efa1e6861d9e804770ec11239:
rmdriverfs: Driver overlay failed to remove root filesystem
26537dcd69eeb25dcd3a4867f81db909e5ca992efa1e6861d9e804770ec11239: mount
still active"
Aug 09 15:40:08 ip-172-31-16-118.ec2.internal dockerd[1107]:
time="2016-08-09T15:40:08.520749851Z" level=error msg="Handler for POST
/v1.22/containers/create returned error: Error relabeling upper directory:
operation not supported"

Could you grab the output of `mount` from these machines?

And it doesn't look like the daemon dies for several minutes after:

Aug 09 15:43:41 ip-172-31-16-118.ec2.internal systemd[1]: Stopping Docker
Application Container Engine...

I really don't know what is getting messed up to cause this. What are the
arguments to docker run?

It looks sort of a like a repeat of
https://github.com/coreos/bugs/issues/1301 which was about SELinux and
OverlayFS so I am going to cc in Nick, Matthew, and Crawford.

Cheers,

Brandon

On Tue, Aug 9, 2016 at 10:41 AM Cory Keane <[email protected]> wrote:

> Hey Brandon,
>
> I decided to just email the logs directly to you.
>
> Thanks!
>
> On Tue, Aug 9, 2016 at 12:37 PM, Brandon Philips <
> [email protected]> wrote:
>
>> journalctl --no-pager -u docker.service > foo.log
>>
>> I think it turns on no-pager by default if it knows stdout isn't a tty
>>
>> On Tue, Aug 9, 2016 at 10:35 AM Cory Keane <[email protected]> wrote:
>>
>>> No worries Brandon. I will upload them. But what's the easiest way to
>>> export them? The command you told me to run makes it so I have to press the
>>> space bar to cycle through all the lines of the file.
>>>
>>> Thanks so much.
>>>
>>> On Tue, Aug 9, 2016 at 12:34 PM, Brandon Philips <
>>> [email protected]> wrote:
>>>
>>>> Hey Cory-
>>>>
>>>> It is really hard to know what to look for in the logs. If you don't
>>>> want to share the logs publicly you can email them to me personally and I
>>>> will take a look.
>>>>
>>>> Cheers,
>>>>
>>>> Brandon
>>>>
>>>> On Tue, Aug 9, 2016 at 8:42 AM Cory Keane <[email protected]> wrote:
>>>>
>>>>> Brandon, I apologize. It appears that this worked for one of the
>>>>> machines experiencing this issue, but not the other machine. Since I 
>>>>> posted
>>>>> this, another one of our fleet machines auto updated to what appears to be
>>>>> this faulty version.
>>>>>
>>>>> So please let me know what I am looking for specifically when running
>>>>> the `journalctl -u docker` command.
>>>>>
>>>>>
>>>
>>>
>>> --
>>> *Cory Keane*
>>> Co-Founder & CTO
>>> ThreadMeUp.com
>>>
>>
>
>
> --
> *Cory Keane*
> Co-Founder & CTO
> ThreadMeUp.com
>

Reply via email to