I'm on the part of the tutorial where it talks about data persistence.
First, I run this command to put a random number into a text file within an ubuntu image:
docker run -d ubuntu bash -c "shuf -i 1-10000 -n 1 -o /data.txt && tail -f /dev/null"
I think I understand this line pretty well.
Next, the instructions ask me to start a new container (the same image) and I will see that the file is not the same:
docker run -it ubuntu ls /
However, when I run the above command, I get the following error:
/ ls: cannot access 'C:/Program Files/Git/': No such file or directory
I'm running Windows 10 using Git Bash, and this is being done through VS Code.
For now, I've gotten around this issue by re-running the exact command (docker run -d ubuntu bash -c "shuf -i 1-10000 -n 1 -o /data.txt && tail -f /dev/null"), but I would like to know why the docker run -it ubuntu ls / instructions failed, and what the solution is?
I managed to solve the issue so I am posting the solution here in case people come across the same issue in the future: git bash changes absolute paths so it is something that should be disabled.
Put this into .bashrc to correct the way paths are handled:
# Workaround for Docker for Windows in Git Bash.
docker()
{
(export MSYS_NO_PATHCONV=1; "docker.exe" "$#")
}
Unfortunately, this doesn't work in scenarios where docker run is called from npm scripts, etc. Volume mapping will still break.
See here to continue exploring the issue and seeing possible workarounds
Without making the title too long, here is the scenario...
I have two scripts:
The first script (host_startup.sh) checks whether a website is up. When it finally posts, then it opens the website in the default browser:
URL=http://localhost:9876
until contents=$(wget -q --spider --no-check-certificate "$URL")
do
sleep 1
done
xdg-open ${URL} &
The second script (run.sh) starts the host_startup script and then starts up a Docker container which serves up a webpage at the aforementioned address:
(../../host_startup.sh) &
docker run --rm -it \
-p 9786:9786 \
company:image
Note that the docker command runs with the --rm flag. However, when I run the run.sh script, and Ctrl+C the process, the Docker image is still running... Specifically, docker ps shows the container still.
I would like Ctrl+C to stop the container and clean it up.
I thought that the container would stop and cleanup because I put the host_startup.sh script in the background, NOT the docker run command...
Please tell me how I can achieve the desired behavior.
Appearantly "Bash does not forward signals like SIGTERM to processes it is currently waiting on".
So you could modify your run.sh to:
(../../host_startup.sh) &
exec docker run --rm -it \
-p 9786:9786 \
company:image
Which would replace the fork of the shell with the docker process instad of waiting for it.
Based on Suart P. Bentley's answer on the unix stackexchange
Alternatively you could manually listen to and act on signals using a modified version of cuonglm's answer to the same question.
I am new to Docker and trying to understand but I have noticed the Visual Studio does a lot of 'magic' behind the scenes. I have managed to figure out all my questions about the docker run command VS uses when you debug an ASP.NET Core app with Docker support except one.
docker run
-dt
-v "C:\Users\jnhaf\vsdbg\vs2017u5:/remote_debugger:rw"
-v "D:\ProtoTypes\WebAppDockerOrNot\WebAppDockerOrNot:/app"
-v "C:\Users\jnhaf\AppData\Roaming\ASP.NET\Https:/root/.aspnet/https:ro"
-v "C:\Users\jnhaf\AppData\Roaming\Microsoft\UserSecrets:/root/.microsoft/usersecrets:ro"
-v "C:\Users\jnhaf\.nuget\packages\:/root/.nuget/fallbackpackages2"
-v "C:\Program Files\dotnet\sdk\NuGetFallbackFolder:/root/.nuget/fallbackpackages"
-e "DOTNET_USE_POLLING_FILE_WATCHER=1"
-e "ASPNETCORE_ENVIRONMENT=Development"
-e "ASPNETCORE_URLS=https://+:443;http://+:80"
-e "ASPNETCORE_HTTPS_PORT=44328"
-e "NUGET_PACKAGES=/root/.nuget/fallbackpackages2"
-e "NUGET_FALLBACK_PACKAGES=/root/.nuget/fallbackpackages;/root/.nuget/fallbackpackages2"
-p 4800:80
-p 44328:443
--entrypoint tail webappdockerornot:dev -f /dev/null
The final argument --entrypoint tail webappdockerornot:dev -f /dev/null is the one that confuses me. I get that VS is overriding the entry point setup in the Dockerfile but what I do not understand nor can find online is what tail webappdockerornot:dev and the -f /dev/null. I figured out that webappdockerornot:dev is the docker image but can someone explain how this argument works or provide a link to something that explains it.
We can break down that command line a little differently as
docker run \
... some other arguments ... \
--entrypoint tail \
webappdockerornot:dev \
-f /dev/null
and match this against a general form
docker run [OPTIONS] [IMAGENAME:TAG] [CMD]
So the --entrypoint tail option sets the entry point to tail, and the "command" part is -f /dev/null. When Docker actually launches the container, it passes the command as additional arguments to the entrypoint. In the end, the net effect of this is
Ignore what the Dockerfile said to do; after setting up the container runtime environment, run tail -f /dev/null instead.
which in turn is a common way to launch a container that doesn't do anything but also stays running. Then you can use docker exec and similar debugging-oriented tools to do things inside the container.
I am running the following command from my Jenkinsfile. However, I get the error "The input device is not a TTY".
docker run -v $PWD:/foobar -it cloudfoundry/cflinuxfs2 /foobar/script.sh
Is there a way to run the script from the Jenkinsfile without doing interactive mode?
I basically have a file called script.sh that I would like to run inside the Docker container.
Remove the -it from your cli to make it non interactive and remove the TTY. If you don't need either, e.g. running your command inside of a Jenkins or cron script, you should do this.
Or you can change it to -i if you have input piped into the docker command that doesn't come from a TTY. If you have something like xyz | docker ... or docker ... <input in your command line, do this.
Or you can change it to -t if you want TTY support but don't have it available on the input device. Do this for apps that check for a TTY to enable color formatting of the output in your logs, or for when you later attach to the container with a proper terminal.
Or if you need an interactive terminal and aren't running in a terminal on Linux or MacOS, use a different command line interface. PowerShell is reported to include this support on Windows.
What is a TTY? It's a terminal interface that supports escape sequences, moving the cursor around, etc, that comes from the old days of dumb terminals attached to mainframes. Today it is provided by the Linux command terminals and ssh interfaces. See the wikipedia article for more details.
To see the difference of running a container with and without a TTY, run a container without one: docker run --rm -i ubuntu bash. From inside that container, install vim with apt-get update; apt-get install vim. Note the lack of a prompt. When running vim against a file, try to move the cursor around within the file.
For docker run DON'T USE -it flag
(as said BMitch)
And it's not exactly what you are asking, but would be also useful for others:
For docker-compose exec use -T flag!
The -T key would help people who are using docker-compose exec! (It disable pseudo-tty allocation)
For example:
docker-compose -f /srv/backend_bigdata/local.yml exec -T postgres backup
or
docker-compose exec -T mysql mysql -uuser_name -ppassword database_name < dir/to/db_backup.sql
For those who struggle with this error and git bash on Windows, just use PowerShell where -it works perfectly.
If you are using git bash on windows, you just need to put
winpty
before your 'docker line' :
winpty docker exec -it some_container bash
In order for docker to allocate a TTY (the -t option) you already need to be in a TTY when docker run is called. Jenkins executes its jobs not in a TTY.
Having said that, the script you are running within Jenkins you may also want to run locally. In that case it can be really convenient to have a TTY allocated so you can send signals like ctrl+c when running it locally.
To fix this make your script optionally use the -t option, like so:
test -t 1 && USE_TTY="-t"
docker run ${USE_TTY} ...
when using 'git bash',
1) I execute the command:
docker exec -it 726fe4999627 /bin/bash
I have the error:
the input device is not a TTY. If you are using mintty, try prefixing the command with 'winpty'
2) then, I execute the command:
winpty docker exec -it 726fe4999627 /bin/bash
I have another error:
OCI runtime exec failed: exec failed: container_linux.go:344: starting container process caused "exec: \"D:/Git/usr/bin/
bash.exe\": stat D:/Git/usr/bin/bash.exe: no such file or directory": unknown
3) third, I execute the:
winpty docker exec -it 726fe4999627 bash
it worked.
when I using 'powershell', all worked well.
Using docker-compose exec -T fixed the problem for me via Jenkins
docker-compose exec -T containerName php script.php
Same Case Here, I am running the following command throw .sh script(bash) and python .py
However, I get the same error "The input device is not a TTY".
in my case, I'm trying to take the dump from a running container of my "production" env with authentication and passing with some arguments,
then take the output of .bak file of my mssql database container.
Remove -it from the command. If you want to keep it interactive then keep -i.
you can check my .sh file and a long command taking dump.
if using windows, try with cmd , for me it works. check if docker is started.
My Jenkins pipeline step shown below failed with the same error.
steps {
echo 'Building ...'
sh 'sh ./Tools/build.sh'
}
In my "build.sh" script file "docker run" command output this error when it was executed by Jenkins job. However it was working OK when the script ran in the shell terminal.The error happened because of -t option passed to docker run command that as I know tries to allocate terminal and fails if there is no terminal to allocate.
In my case I have changed the script to pass -t option only if a terminal could be detected. Here is the code after changes :
DOCKER_RUN_OPTIONS="-i --rm"
# Only allocate tty if we detect one
if [ -t 0 ] && [ -t 1 ]; then
DOCKER_RUN_OPTIONS="$DOCKER_RUN_OPTIONS -t"
fi
docker run $DOCKER_RUN_OPTIONS --name my-container-name my-image-tag
I know this is not directly answering the question at hand but for anyone that comes upon this question who is using WSL running Docker for windows and cmder or conemu.
The trick is not to use Docker which is installed on windows at /mnt/c/Program Files/Docker/Docker/resources/bin/docker.exe but rather to install the ubuntu/linux Docker. It's worth pointing out that you can't run Docker itself from within WSL but you can connect to Docker for windows from the linux Docker client.
Install Docker on Linux
sudo apt-get install apt-transport-https ca-certificates curl software-properties-common
curl -fsSL https://download.docker.com/linux/ubuntu/gpg | sudo apt-key add -
sudo add-apt-repository "deb [arch=amd64] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable"
sudo apt-get update
sudo apt-get install docker-ce
Connect to Docker for windows on the port 2375 which needs to be enabled from the settings in docker for windows.
docker -H localhost:2375 run -it -v /mnt/c/code:/var/app -w "/var/app" centos:7
Or set the docker_host variable which will allow you to omit the -H switch
export DOCKER_HOST=tcp://localhost:2375
You should now be able to connect interactively with a tty terminal session.
In Jenkins, I'm using docker-compose exec -T
eg:-
docker-compose exec -T app php artisan migrate
winpty works as long as you don't specify volumes to be mounted such as .:/mountpoint or ${pwd}:/mountpoint
The best workaround I have found is to use the git-bash plugin inside Visual Code Studio and use the terminal to start and stop containers or docker-compose.
For those using Pyinvoke see this documentation which I'll syndicate here in case the link dies:
99% of the time, adding pty=True to your run call will make things work as you were expecting. Read on for why this is (and why pty=True is not the default).
Command-line programs often change behavior depending on whether a controlling terminal is present; a common example is the use or disuse of colored output. When the recipient of your output is a human at a terminal, you may want to use color, tailor line length to match terminal width, etc.
Conversely, when your output is being sent to another program (shell pipe, CI server, file, etc) color escape codes and other terminal-specific behaviors can result in unwanted garbage.
Invoke’s use cases span both of the above - sometimes you only want data displayed directly, sometimes you only want to capture it as a string; often you want both. Because of this, there is no “correct” default behavior re: use of a pseudo-terminal - some large chunk of use cases will be inconvenienced either way.
For use cases which don’t care, direct invocation without a pseudo-terminal is faster & cleaner, so it is the default.
Instead of using -it use --tty
So your docker run should look like this:
docker run -v $PWD:/foobar --tty cloudfoundry/cflinuxfs2 /foobar/script.sh
use only -i flag than -it flag. which can help you to see what going on inside container.
docker exec -i $USER bash <<EOF
apt install nano -y
EOF
you might see the warning but it shows you output on the terminal inside docker.
I have a utility program that depends on terminal characteristics. I want to execute it inside a docker container. (the program is not a interactive program as such. It is an old program that was written that way).
docker run -i -t or docker exec -i -t should open a tty into container. But here is what happens..
user#1755e1f3f735:~/region/primer/cobol_v> kickstop
[Error] Unable to run without terminal device (tty)
user#1755e1f3f735:~/region/primer/cobol_v> tty
not a tty
When -t option to docker command (run/exec) should give a 'tty', the tty commands returns with 'not a tty'. This is puzzling.
I experienced this on a openSuse and fedora23 hosts and images, if that matters. I used 'guake', MATE (Gnome?) terminal emulators for this, with same results.
Is there any solution to this? or this is by design and have to replace/rewrite my utility?
I ran into the same issue, and found "docker exec -ti container script /dev/null" solved the problem.
After login to the container with the above command, I can use screen normally.
Reference: https://github.com/docker/docker/issues/8755
I ran some experiments and here are findings. Hope someone finds them useful.
(docker commands are not complete but just brief)
1. docker run -i -t
> tty
/dev/console
> echo $TERM
xterm
>kickstop
works!!
2. docker -d followed by docker exec -i -t
>tty
not a tty
>echo $TERM
dumb
>kickstop
[Error] Unable to run without terminal device (tty)
3. docker -d followed by docker attach
you get attached to /dev/console. No prompt (because I'm running tail -f xxx.log to keep the container alive). In fact I need to stop my application from another terminal (using docker exec) and stop the container to get back to the prompt (host shell)
4. docker start followed by docker attach
same as above