Docker - Error response from daemon: client is new

问题:

After creating a new machine with Docker Machine, I'm getting the following error:

$ docker ps
Error response from daemon: client is newer than server(client API version 1.21, server API version: 1.19)

How can I fix this?

回答1:

docker-machine upgrade <your-machine>

will do the trick. This can happen - as it did for me - even if you're not using RCs and your machine was newly created. It would be due to an ISO cache issue. The error is commented in this thread.

If the docker client is 1.9.x and the server is running docker 1.8.x, the error message is observed.

回答2:

If someone happens to get this error, but is not using docker-machine, there is another way to resolve the issue by specifying an older API version in an environment variable on the client side:

export DOCKER_API_VERSION=<version>

for example:

export DOCKER_API_VERSION=1.19

and retrying the docker command.

Reference.

回答3:

On ubuntu distrib it happens after an apt update, if docker run as a service. The client is updated but the old version of the server is still running.

In this case just do a:

sudo service docker restart

回答4:

If you upgrade your docker client you will not be able to use old docker-machine VMs. Actually you can force an upgrade with docker-machine upgrade vm-name, but if you are working with other distros (except boot2docker) as docker host like coreos, atomic, rancheros it may fail.

I am using Carina to manage more than one docker client version.

just install with $ curl -sL https://download.getcarina.com/dvm/latest/install.sh | sh (linux and mac installation) and run dvm ls to list installed versions, dvm install 1.9.0 to install versions and dvm use 1.8.1 to change your client version.

Very easy and powerfull. =)

Atention, for your safety always take a look inside scripts downloaded from internet before running them.

Hope it helps

回答5:

Apparently this error is due to docker-machine "falling asleep". Here is the magic command solving it for me:

$ docker-machine restart default

Or replace default by the name of your machine. The result is:

$ docker-machine restart default
Restarting "default"...
Restarted machines may have new IP addresses. You may need to re-run the `docker-machine env` command.

As suggested, you may need to:

$ docker-machine env default
export DOCKER_TLS_VERIFY="1"
export DOCKER_HOST="tcp://....."
export DOCKER_CERT_PATH=".../.docker/machine/machines/default"
export DOCKER_MACHINE_NAME="default"
# Run this command to configure your shell:
# eval $(docker-machine env default)

And so you finally run that command:

$ eval $(docker-machine env default)

From now it should work.

回答6:

I had the same issue with 1.10.0-rc4, and it turned out that boot2docker wasn't updated properly when installing a newer Docker Toolbox. I had to remove boot2docker manually:

sudo rm -rf /usr/local/share/boot2docker

You can also delete the rest of the toolbox:

sudo rm -rf /usr/local/bin/docker*
sudo rm -rf /Applications/Docker

And finally install Docker Toolbox of the desired version.

Please see official uninstall script for on GitHub for OS X here.