[Bug 1184710] New: dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

Bug ID: 1184710
           Summary: dnsmasq needs to be restarted after reboot for dns to
                    work in a Docker container
           Product: Fedora
           Version: 21
         Component: docker-io
          Assignee: [e-mail]
          Reporter: [e-mail]
        QA Contact: [e-mail]
                CC: [e-mail], [e-mail],
                    [e-mail], [e-mail],
                    [e-mail], [e-mail],
                    [e-mail], [e-mail],
                    [e-mail], [e-mail], [e-mail],
                    [e-mail], [e-mail]
   External Bug ID: Red Hat Bugzilla 1128208
   External Bug ID: Red Hat Bugzilla 1128208

Description of problem:

Docker replies on dnsmasq to have 'listen-address' set to the docker bridge,
and have bind-interfaces option turned on. However in this configuration
dnsmasq has to start after docker in order for DNS to work in a docker
container.

Version-Release number of selected component (if applicable):

docker-io 1.4.0
dnsmasq 2.72

How reproducible:

Reboot the system with docker, and dnsmasq enabled.

Steps to Reproduce:
1. Ensure docker is using the default dns option of 172.17.42.1
2. Use the attached dnsmasq.conf
3. Reboot the system
4. Launch a docker container:

`docker run -i -t centos /usr/bin/ping www.redhat.com`

Actual results:

Cannot resolve hostname

Expected results:

Response from 'akamai' or similar

Additional info:

See related issue #1128208
 I tried added a systemd 'After' for dnsmasq on docker.service, but it doesn't
seem to be honoured by systemd.

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

Lokesh Mandvekar  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

--- Comment #1 from Lokesh Mandvekar  ---
(In reply to Jason Shepherd from comment #0)
> Version-Release number of selected component (if applicable):
>
> docker-io 1.4.0

Could you try this with the latest stable rpm? (1.4.1-8.fc21)
> dnsmasq 2.72
>
> How reproducible:
>
> Reboot the system with docker, and dnsmasq enabled.
>
> Steps to Reproduce:
> Ensure docker is using the default dns option of 172.17.42.1
> Use the attached dnsmasq.conf

Hmm, don't see anything attached. I'll try reproducing this once I have your
dnsmasq.conf
> Reboot the system
>
> Launch a docker container:
>
> `docker run -i -t centos /usr/bin/ping www.redhat.com`

This works fine on my system, but then again, I'm not using dnsmasq:

$ rpm -q docker-io
docker-io-1.4.1-8.fc21.x86_64

$ sudo docker run -it centos /usr/bin/ping www.redhat.com
PING e1890.b.akamaiedge.net (23.193.135.214) 56(84) bytes of data.
64 bytes from a23-193-135-214.deploy.static.akamaitechnologies.com
(23.193.135.214): icmp_seq=1 ttl=51 time=14.4 ms

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

--- Comment #2 from Jason Shepherd  ---
Created attachment 989895
  --> https://bugzilla.redhat.com/attachment.cgi?id=989895&action=edit
DNSMASQ configuration

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

--- Comment #3 from Jason Shepherd  ---
Hi Lokesh,

Thanks for looking into this. On further investigation I think the problem
could be related to use an LVM partition mounted at /var/lib/docker. However I
haven't been able to verify that, it's just been reported by others.

[jshepher@jshepher Documents]$ sudo lvdisplay /dev/vg_dhcp-1-52/lv_docker
  --- Logical volume ---
  LV Path                /dev/vg_dhcp-1-52/lv_docker
  LV Name                lv_docker
  VG Name                vg_dhcp-1-52
  LV UUID                4vFfQI-4Axe-cWza-tzAZ-QcuJ-e1cQ-UNVPV5
  LV Write Access        read/write
  LV Creation host, time jshepher.usersys.redhat.com, 2014-04-03 13:17:18 +1000
  LV Status              available
  # open                 1
  LV Size                125.00 GiB
  Current LE             4000
  Segments               2
  Allocation             inherit
  Read ahead sectors     auto
  - currently set to     256
  Block device           253:4

cat /etc/fstab:

/dev/mapper/vg_dhcp--1--52-lv_docker     /var/lib/docker   ext4 defaults 1 2

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

--- Comment #5 from Jason Shepherd  ---
I tried with the latest version, 1.4.1-8.fc21, but it didn't make any
difference.

I need to restart dnsmasq on the host for dns name resolution to work inside
the container.

sudo systemctl restart dnsmasq

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

Daniel Walsh  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

Fedora Admin XMLRPC Client  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|[e-mail]             |[e-mail]

--- Comment #7 from Fedora Admin XMLRPC Client  ---
This package has changed ownership in the Fedora Package Database.  Reassigning
to the new owner of this component.

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

Daniel Walsh  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|unspecified                 |medium
           Assignee|[e-mail]         |[e-mail]

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

Daniel Walsh  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Assignee|[e-mail]             |[e-mail]

--- Comment #8 from Daniel Walsh  ---
Sally could you confirm if this is still a problem.

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

--- Comment #9 from Sally  ---
$ rpm -q docker-io
docker-io-1.6.0-4.git350a636.fc21.x86_64

$ docker run -i -t centos /usr/bin/ping www.redhat.com
PING e1890.b.akamaiedge.net (23.196.44.61) 56(84) bytes of data.
64 bytes from a23-196-44-61.deploy.static.akamaitechnologies.com
(23.196.44.61): icmp_seq=1 ttl=48 time=33.6 ms
64 bytes from a23-196-44-61.deploy.static.akamaitechnologies.com
(23.196.44.61): icmp_seq=2 ttl=48 time=39.8 ms

No errors.  I followed steps outlined in Description.

[Bug 1184710] dnsmasq needs to be restarted after reboot for dns to work in a Docker container

[Red Hat Bugzilla]

https://bugzilla.redhat.com/show_bug.cgi?id=1184710

Jason Shepherd  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |CLOSED
         Resolution|---                         |CURRENTRELEASE
        Last Closed|                            |2015-07-08 19:07:11

--- Comment #10 from Jason Shepherd  ---
Also seems to be fixed for me.