I've got a Docker container running Ubuntu which I did as follows:
docker run -it ubuntu /bin/bash
however it doesn't seem to have ping
. E.g.
bash: ping: command not found
Do I need to install that?
Seems a pretty basic command to be missing. I tried whereis ping
which doesn't report anything.
Docker images are pretty minimal, but you can install ping
in your official ubuntu docker image via:
apt-get update
apt-get install iputils-ping
Chances are you don't need ping
on your image, and just want to use it for testing purposes. Above example will help you out.
But if you need ping
to exist on your image, you can create a Dockerfile
or commit
the container you ran the above commands into a new image.
Commit:
docker commit -m "Installed iputils-ping" --author "Your Name <name@domain.com>" ContainerNameOrId yourrepository/imagename:tag
Dockerfile:
FROM ubuntu
RUN apt-get update && apt-get install -y iputils-ping
CMD bash
Please note there are best practices on creating docker images, like clearing apt cache files afterwards and etc.
This is the Docker Hub page for Ubuntu and this is how it is created. It only has (somewhat) bare minimum packages installed, thus if you need anything extra you need to install it yourself.
apt-get update && apt-get install -y iputils-ping
However usually you'd create a "Dockerfile" and build it:
mkdir ubuntu_with_ping
cat >ubuntu_with_ping/Dockerfile <<'EOF'
FROM ubuntu
RUN apt-get update && apt-get install -y iputils-ping
CMD bash
EOF
docker build -t ubuntu_with_ping ubuntu_with_ping
docker run -it ubuntu_with_ping
Please use Google to find tutorials and browse existing Dockerfiles to see how they usually do things :) For example image size should be minimized by running apt-get clean && rm -rf /var/lib/apt/lists/*
after apt-get install
commands.
echo -e
actually defies the POSIX sh standard, which doesn't allow it to do anything but print -e
on its output. (Even with some versions of bash, that's the default behavior). Use printf
instead: printf '%s\n' "FROM ubuntu" "RUN apt-get update && apt-get install -y iputils-ping" "CMD bash"
, and see the APPLICATION USAGE section of the above-linked standards document.
echo -e
the way you expect it to (but instead will have a standards-compliant behavior) when in POSIX mode compiled with --enable-xpg-echo-default
, or with appropriate environment variables or other runtime configuration.
-n
as a first argument, or when any backslash literals are present -- but even then, it's implementation-defined, not standard-guaranteed, so behavior is dependent on the individual shell in use).
Alternatively you can use a Docker image which already has ping installed, e.g. busybox:
docker run --rm busybox ping SERVER_NAME -c 2
apt-get iputils-ping
on the image which needs it.
Generally people pull the official image of Ubuntu/CentOS but they don't realize that these images are minimal and doesn't have any thing on the top of that.
For Ubuntu, this image is built from official rootfs tarballs provided by Canonical. Given that it is a minimal install of Ubuntu, this image only includes the C, C.UTF-8, and POSIX locales by default.
One can install net-tools (includes ifconfig, netstat), ip-utils(includes ping) andy other likes curl etc on container and can create image from container or can write Dockerfile that will install these tool while creating image.
Below is Dockerfile example, while creating image from this it will include these tools:
FROM vkitpro/ubuntu16.04
RUN apt-get update -y \
&& apt-get upgrade -y \
&& apt-get install iputils-ping -y \
&& apt-get install net-tools -y \
CMD bash
or launch container from base image and install these utilities on container and then commit to image. docker commit -m "any descriptive message" container_id image_name:lattest
That image will have all thing installed.
Sometimes, the minimal installation of Linux in Docker doesn't define the path and therefore it is necessary to call ping using ....
cd /usr/sbin
ping <ip>
I have used the statement below on debian 10
apt-get install iputils-ping
Every time you get this kind of error
bash: <command>: command not found
On a host with that command already working with this solution: dpkg -S $(which
Don't have a host with that package installed? Try this: apt-file search /bin/
Alternatively, you can run ping
on the host after entering the container's network namespace.
First, find the container's process ID on the host (this can be the shell or the application you are running inside the container). Then change to the container's network namespace (run as root
on the host):
host# PS1='container# ' nsenter -t <PID> -n
Modifying the PS1
environment variable is only used to show a different prompt while you are in the container's network namespace.
Now you can use ping
, netstat
, ifconfig
, ip
, etc, given that they are installed on the host.
container# ping <IP>
container# ip route get <IP>
....
container# exit
Keep in mind that this only changes the network namespace, the mount namespace (file system) has not changed and therefore name resolving may not work correctly (it's still using the /etc/hosts
file on the host)
Success story sharing
Temporary failure resolving 'security.ubuntu.com'
obviously because networking is not present.apt install
and ended up with "package not found", but as the answer says,apt-get
works just fine.docker exec -u 0 -it <container> /bin/bash
. Where-u 0
is user root.