83
votes

I'm trying to setup a private docker registry using the image taken from: https://github.com/docker/docker-registry

Just by running:
docker run -p 5000:5000 registry

I can pull/push from/to this repository only from localhost, but if i try to access it from another machine (using a private address on the same LAN) it fails with an error message:

*2014/11/03 09:49:04 Error: Invalid registry endpoint https ://10.0.0.26:5000/v1/': 
Get https:// 10.0.0.26:5000/v1/_ping: Forbidden. If this private 
registry supports only HTTP or HTTPS with an unknown CA certificate,
please add `--insecure-registry 10.0.0.26:5000` to the daemon's 
arguments. In the case of HTTPS, if you have access to the registry's
CA certificate, no need for the flag; simply place the CA certificate 
at /etc/docker/certs.d/10.0.0.26:5000/ca.crt*

What drives me crazy is that I can access it successfully using: curl 10.0.0.26:5000 and/or curl 10.0.0.26:5000/v1/search

I also don't understand where and how I should pass the --insecure-registry flag.

15
many of the answers seem to be out of date for docker 1.12 but see vikas027 answer which is great for docker 1.12 (latest ATOW) - danday74
On Ubuntu, the Docker documentation and this answer worked well for me. - Batandwa

15 Answers

73
votes

OK - I found the solution to this - after a day of digging.

For docker below 1.12.1:

It turns out that the new client version refuses to work with a private registry without SSL.

To fix this - the daemon on the client machine should be launched with the insecure flag:

Just type:

sudo service docker stop # to stop the service

and then

sudo docker -d --insecure-registry 10.0.0.26:5000

(replace the 10.0.0.26 with your own ip address).

I would expect the docker guys to add this option to the pull/push command line...

Edit - altenantively - you can add the flag to DOCKER_OPTS env variable inside /etc/default/docker... and then sudo service docker restart

Edit again - It seems that the docker guys are on it - and a fix will come soon: https://github.com/docker/docker/pull/8935

For docker 1.12.1:

Please follow below the answer of vikas027 (valid for centos)

28
votes

This is what worked for me on CentOS 7.2 and Docker 1.12.1 (latest as on date). My private registry v2 was running on 192.168.1.88:5000, change it accordingly. This also works if you have multiple registries, just keep on adding --insecure-registry IP:Port

$ sudo vim /usr/lib/systemd/system/docker.service
#ExecStart=/usr/bin/dockerd 
ExecStart=/usr/bin/dockerd --insecure-registry 192.168.1.88:5000
$
$ sudo systemctl stop docker
$ sudo systemctl daemon-reload
$ systemctl start docker
24
votes

Edit the config file "/etc/default/docker"

sudo vi /etc/default/docker

add the line at the end of file

DOCKER_OPTS="$DOCKER_OPTS --insecure-registry=192.168.2.170:5000"

(replace the 192.168.2.170 with your own ip address)

and restart docker service

sudo service docker restart

12
votes

I found the following to be very useful as it discusses how the Docker service itself is configured. https://docs.docker.com/articles/systemd/

Along with this article on the systemctl command https://www.digitalocean.com/community/tutorials/how-to-use-systemctl-to-manage-systemd-services-and-units

I used the following series of commands in a Centos 7 based container with a registry image obtained by "docker pull registry:2.1.1"

sudo mkdir -p /etc/systemd/system/docker.service.d
cd /etc/systemd/system/docker.service.d
sudo touch override.conf
sudo nano override.conf

And inside the override.conf added the following.

[Service]
ExecStart=
ExecStart=/usr/bin/docker -d -H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock --insecure-registry 10.2.3.4:5000

Note the first, blank, ExecStart= clears anything that is already in place so be sure to add anything from the /usr/lib/systemd/system/docker.service ExecStart= statement that you wish to retain.

If you don't specify the -d(daemon) option you'll get a "Please specify only one -H" error.

After issuing the following series of commands I can see my overrides in place.

sudo systemctl stop docker
sudo systemctl daemon-reload
sudo systemctl start docker
sudo systemctl status docker

docker.service - Docker Application Container Engine
   Loaded: loaded (/usr/lib/systemd/system/docker.service; enabled)
  Drop-In: /etc/systemd/system/docker.service.d
           └─override.conf
   Active: active (running) since Thu 2015-09-17 13:37:34 AEST; 7s ago
     Docs: https://docs.docker.com
 Main PID: 5697 (docker)
   CGroup: /system.slice/docker.service
           └─5697 /usr/bin/docker -d -H tcp://0.0.0.0:2375 -H unix:///var/run/docker.sock --insecure-registry 10.2.3.4:5000

NOTE: The information provided by Loaded: and Drop-In: lines in the status message, the are useful for checking what's happening with a pre-existing docker daemon to work.

NOTE: Also have a look in the Loaded: docker.service file for an EnvironmentFile= for further clues.

11
votes

Ok. Here is how I got it to work. If you see this error in docker 1.3.2 or above, do this

go to /etc/sysconfig/docker

other_args="--insecure-registry 10.0.0.26:5000"

and run

sudo service docker restart

9
votes

use the following command replacing {YOUR_REGISTRY} with your registry

boot2docker ssh "echo $'EXTRA_ARGS=\"--insecure-registry {YOUR_REGISTRY}\"' | sudo tee -a /var/lib/boot2docker/profile && sudo /etc/init.d/docker restart"
7
votes

edit docker.service file, add --insecure-registry x.x.x.x after -d flag, restart docker

this is the only thing that worked for me, the DOCKER_OPTS didn't have any effect

4
votes

Docker 1.12.1

For CentOS 7.2

/usr/lib/systemd/system/docker.service
#ExecStart=/usr/bin/dockerd
ExecStart=/usr/bin/dockerd --insecure-registry my-docker-registry.com:5000

For ubuntu 16.04

/lib/systemd/system/docker.service
#ExecStart=/usr/bin/dockerd -H fd://
ExecStart=/usr/bin/dockerd --insecure-registry my-docker-registry.com:5000 -H fd://

sudo systemctl stop docker
sudo systemctl daemon-reload
sudo systemctl start docker

It seems the --insecure-registry option may be used both with and without the "=" between it and the registry ID.

2
votes

I found that docker client version and registry docker version has to match up, else you would run into connectivity issues, despite having everything in place.

2
votes

Two step solution(without --insecure-registry):

  1. Download public key from your registry
  2. Put it into /etc/docker/certs.d/$HOSTNAME/ directory

 

mkdir -p /etc/docker/certs.d/10.0.0.26:5000
echo -n | openssl s_client -connect 10.0.0.26:5000 | sed -ne '/-BEGIN CERTIFICATE-/,/-END CERTIFICATE-/p' > /etc/docker/certs.d/10.0.0.26:5000/registry.crt

Now your docker will trust your self-signed certificate.

1
votes

This is based on the answer from vikas027 on Centos 7 and Docker 1.12

Since I am behind a proxy my full solution was ...

/etc/systemd/system/docker.service.d/http-proxy.conf

[Service]

Environment="FTP_PROXY={{MY_PROXY}}"
Environment="ftp_proxy={{MY_PROXY}}"

Environment="HTTPS_PROXY={{MY_PROXY}}"
Environment="https_proxy={{MY_PROXY}}"

Environment="HTTP_PROXY={{MY_PROXY}}"
Environment="http_proxy={{MY_PROXY}}"

Environment="NO_PROXY=localhost,127.0.0.1,{{MY_INSECURE_REGISTRY_IP}}"
Environment="no_proxy=localhost,127.0.0.1,{{MY_INSECURE_REGISTRY_IP}}"

/usr/lib/systemd/system/docker.service

ExecStart=/usr/bin/dockerd --insecure-registry {{MY_INSECURE_REGISTRY_IP}}:5000

and dont forget to restart :)

sudo systemctl daemon-reload; sudo systemctl restart docker;
0
votes

Setting Local insecure registry in docker along with proxy:

1) in ubuntu add the following flag --insecure-registry IP:port under DOCKER_OPTS in file /etc/default/docker

1.1) configure no_proxy env variable to bypass local IP/hostname/domainname...as proxy can throw a interactive msg ...like continue and this intermediate msg confuses docker client and finally timesout...

1.2) if domainname is configured...then don't forget to update /etc/hosts file if not using DNS.

1.3) in /etc/default/docker set the env variables http_proxy and https_proxy...as it enables to download images from outside company hubs. format http_proxy=http://username:password@proxy:port

2) restart the docker service...if installed as service, use sudo service docker restart

3) restart the registry container [sudo docker run -p 5000:5000 registry:2 ]

4) tag the required image using sudo docker tag imageid IP:port/imagename/tagname ifany

5) push the image ...sudo docker push ip:port/imagename

6) If u want to pull the image from another machine say B without TLS/SSL,then in B apply setps 1,1.1 and 2. If these changes are not done in machine B...pull will fail.

0
votes

To save you hassle, why don't you just use the FREE private docker registry service provided by gitlab - works great

https://about.gitlab.com/2016/05/23/gitlab-container-registry/

Their registry is secure so you won't have any issues

0
votes

Ubuntu 16.04

Create (does not exist) file /etc/systemd/system/docker.service.d/registry.conf with contents:

[Service]
#You need the below or you 'ExecStart=' or you will get and error 'Service has more than one ExecStart= setting, which is only allowed'
ExecStart=
ExecStart=/usr/bin/dockerd -H fd:// --insecure-registry 10.20.30.40:5000

then

sudo systemctl stop docker
sudo systemctl daemon-reload
sudo systemctl start docker
0
votes

In addition to the above answers, I am adding what worked in "docker for mac" for me:

  1. Click on the docker whale icon from mac tray on top right corner of your screen.
  2. Click on Preferences -> Daemon.
  3. Add your IP and port to the insecure registries.
  4. Restart the Daemon.

enter image description here