Bug 1118234

Summary:	Docker doesn't update DNS IP when it's changed on system
Product:	[Fedora] Fedora	Reporter:	Maxime Carron <maxime.carron>
Component:	docker-io	Assignee:	Lokesh Mandvekar <lsm5>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Fedora Extras Quality Assurance <extras-qa>
Severity:	medium	Docs Contact:
Priority:	unspecified
Version:	rawhide	CC:	admiller, dwalsh, golang-updates, hushan.jia, mattdm, mgoldman, s, vbatts
Target Milestone:	---
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2015-01-27 03:26:17 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Maxime Carron 2014-07-10 09:11:32 UTC

Description of problem:
When the DNS server is changed on the system, docker keep using the one used at its startup.

Version-Release number of selected component (if applicable):
EL 7.0
docker : 1.0.0-1

How reproducible:
Always.

Steps to Reproduce:
1. Start docker (systemctl start docker)
2. Do any operation which needs DNS resolution (docker pull fedora:latest)
=> System DNS is used
3. Change system DNS to 8.8.8.8 or any different DNS
4. Do some other operation which needs DNS resolution (docker pull fedora:rawhide)
5. You should see that first is still used

Actual results:
No DNS update

Expected results:
Docker should use the current system DNS

Additional info:
This is especially disturbing on a laptop, because a docker restart is needed everytime I change network (when first DNS is not reachable anymore, usually when DNS has a private IP)

Comment 1 Lokesh Mandvekar 2014-07-11 17:07:49 UTC

Hmm, don't have a fix yet, but it seems:

Case 1: change DNS from reachable to unreachable address, DNS settings take effect right away

Case 2: change DNS from unreachable to reachable address, the very first docker operation fails, but subsequent ones work without a restart.

Sample run follows (on a rawhide instance):

lsm5 @ nagato : /home/lsm5 $ docker pull fedora:latest
Pulling repository fedora
3f2fed40e4b0: Download complete 
511136ea3c5a: Download complete 
fd241224e9cf: Download complete 
lsm5 @ nagato : /home/lsm5 $ cat /etc/resolv.conf 
nameserver 8.8.8.8
nameserver 8.8.4.4
lsm5 @ nagato : /home/lsm5 $ sudo mv /etc/resolv.conf /etc/resolv.conf.safe
lsm5 @ nagato : /home/lsm5 $ sudo vim /etc/resolv.conf      
lsm5 @ nagato : /home/lsm5 $ cat /etc/resolv.conf
nameserver 1.1.1.1
lsm5 @ nagato : /home/lsm5 $ docker pull fedora:rawhide
Pulling repository fedora
2014/07/11 11:43:36 Get https://cdn-registry-1.docker.io/v1/repositories/library/fedora/tags: dial tcp: lookup cdn-registry-1.docker.io on 1.1.1.1:53: no answer from server
lsm5 @ nagato : /home/lsm5 $ sudo mv /etc/resolv.conf.safe /etc/resolv.conf
lsm5 @ nagato : /home/lsm5 $ cat /etc/resolv.conf 
nameserver 8.8.8.8
nameserver 8.8.4.4
lsm5 @ nagato : /home/lsm5 $ docker pull fedora:rawhide
Pulling repository fedora
2014/07/11 11:46:07 Get https://index.docker.io/v1/repositories/fedora/images: dial tcp: lookup index.docker.io on 1.1.1.1:53: no answer from server
lsm5 @ nagato : /home/lsm5 $ docker pull fedora:rawhide
Pulling repository fedora
64fd7993bcaf: Download complete 
64fd7993bcaf: Pulling image (rawhide) from fedora 
511136ea3c5a: Download complete 


---/var/log/messages---

Jul 11 11:40:14 nagato docker: [584bdf56] -job pull(fedora, latest) = OK (0)
Jul 11 11:42:45 nagato docker: 2014/07/11 11:42:45 POST /images/create?fromImage=fedora&tag=rawhide
Jul 11 11:42:45 nagato docker: [584bdf56] +job pull(fedora, rawhide)
Jul 11 11:43:36 nagato docker: [error] server.go:1185 Get https://cdn-registry-1.docker.io/v1/repositories/library/fedora/tags: dial tcp: lookup cdn-registry-1.docker.io on 1.1.1.1:53: no answer from server
Jul 11 11:43:36 nagato docker: Get https://cdn-registry-1.docker.io/v1/repositories/library/fedora/tags: dial tcp: lookup cdn-registry-1.docker.io on 1.1.1.1:53: no answer from server
Jul 11 11:43:36 nagato docker: [584bdf56] -job pull(fedora, rawhide) = ERR (1)
Jul 11 11:45:47 nagato docker: 2014/07/11 11:45:47 POST /images/create?fromImage=fedora&tag=rawhide
Jul 11 11:45:47 nagato docker: [584bdf56] +job pull(fedora, rawhide)
Jul 11 11:46:07 nagato docker: Get https://index.docker.io/v1/repositories/fedora/images: dial tcp: lookup index.docker.io on 1.1.1.1:53: no answer from server
Jul 11 11:46:07 nagato docker: [584bdf56] -job pull(fedora, rawhide) = ERR (1)
Jul 11 11:46:11 nagato docker: 2014/07/11 11:46:11 POST /images/create?fromImage=fedora&tag=rawhide
Jul 11 11:46:11 nagato docker: [584bdf56] +job pull(fedora, rawhide)
Jul 11 11:46:16 nagato kernel: [37692.370612] EXT4-fs (dm-2): mounted filesystem with ordered data mode. Opts: discard
Jul 11 11:46:16 nagato kernel: EXT4-fs (dm-2): mounted filesystem with ordered data mode. Opts: discard
Jul 11 11:47:07 nagato kernel: [37743.712593] EXT4-fs (dm-3): mounted filesystem with ordered data mode. Opts: discard
Jul 11 11:47:07 nagato kernel: EXT4-fs (dm-3): mounted filesystem with ordered data mode. Opts: discard
Jul 11 11:47:16 nagato docker: [584bdf56] -job pull(fedora, rawhide) = OK (0)

---/var/log/messages---

Btw, the epel7 branch has been removed as docker is available by default on RHEL7. Though not sure if docker 1.0.0 (and later) have made it into RHEL7 proper yet.

Changing bug Product and Version to Fedora and rawhide.

Comment 2 Daniel Walsh 2014-10-24 20:49:46 UTC

docker-1.3 should pick up the /etc/resolv.conf file on docker run time not at the time the daemon starts.

I think this fixes most of your problem.

Comment 3 Daniel Walsh 2014-11-20 19:33:51 UTC

Not likely to change the rest.

Fixed in docker-1.3.1