shell script : container vs host - containers

For Loop to start from 2nd argument behaving differently on container and host
#!/bin/bash
for i in "${#:2}"
do
echo $i
done
Call:
script.sh 129 5 6 7
Output:
Container: Alpine:Latest
#skipping 2 characters
9 5 6 7
Host: Debian GNU/Linux
#skipping 1st argument complete
5 6 7


i am not sure how you were able to run the above shell script into the alpine version.
#!/bin/bash
for i in "${#:2}"
do
echo $i
done
as in script you are using #!/bin/bash which is not supported in the alpine.
Issue could be due the different interpreters, sh and bash which one you are using. Debian comes with bash while alpine uses sh by default.
you can quickly verify it using : docker pull alpine:latest
docker run -it alpine bash while docker run -it alpine sh will work.

Related

How to use external files as arguments in Docker images [duplicate]

I am trying to mount a host directory into a Docker container so that any updates done on the host is reflected into the Docker containers.
Where am I doing something wrong. Here is what I did:
kishore$ cat Dockerfile
FROM ubuntu:trusty
RUN apt-get update
RUN apt-get -y install git curl vim
CMD ["/bin/bash"]
WORKDIR /test_container
VOLUME ["/test_container"]
kishore$ tree
.
├── Dockerfile
└── main_folder
├── tfile1.txt
├── tfile2.txt
├── tfile3.txt
└── tfile4.txt
1 directory, 5 files
kishore$ pwd
/Users/kishore/tdock
kishore$ docker build --tag=k3_s3:latest .
Uploading context 7.168 kB
Uploading context
Step 0 : FROM ubuntu:trusty
---> 99ec81b80c55
Step 1 : RUN apt-get update
---> Using cache
---> 1c7282005040
Step 2 : RUN apt-get -y install git curl vim
---> Using cache
---> aed48634e300
Step 3 : CMD ["/bin/bash"]
---> Running in d081b576878d
---> 65db8df48595
Step 4 : WORKDIR /test_container
---> Running in 5b8d2ccd719d
---> 250369b30e1f
Step 5 : VOLUME ["/test_container"]
---> Running in 72ca332d9809
---> 163deb2b1bc5
Successfully built 163deb2b1bc5
Removing intermediate container b8bfcb071441
Removing intermediate container d081b576878d
Removing intermediate container 5b8d2ccd719d
Removing intermediate container 72ca332d9809
kishore$ docker run -d -v /Users/kishore/main_folder:/test_container k3_s3:latest
c9f9a7e09c54ee1c2cc966f15c963b4af320b5203b8c46689033c1ab8872a0eakishore$ docker run -i -t k3_s3:latest /bin/bash
root#0f17e2313a46:/test_container# ls -al
total 8
drwx------ 2 root root 4096 Apr 29 05:15 .
drwxr-xr-x 66 root root 4096 Apr 29 05:15 ..
root#0f17e2313a46:/test_container# exit
exitkishore$ docker -v
Docker version 0.9.1, build 867b2a9
I don't know how to check boot2docker version
Questions, issues facing:
How do I need to link the main_folder to the test_container folder present inside the docker container?
I need to make this automatically. How do I to do that without really using the run -d -v command?
What happens if the boot2docker crashes? Where are the Docker files stored (apart from Dockerfile)?

There are a couple ways you can do this. The simplest way to do so is to use the dockerfile ADD command like so:
ADD . /path/inside/docker/container
However, any changes made to this directory on the host after building the dockerfile will not show up in the container. This is because when building a container, docker compresses the directory into a .tar and uploads that context into the container permanently.
The second way to do this is the way you attempted, which is to mount a volume. Due to trying to be as portable as possible you cannot map a host directory to a docker container directory within a dockerfile, because the host directory can change depending on which machine you are running on. To map a host directory to a docker container directory you need to use the -v flag when using docker run, e.g.,:
# Run a container using the `alpine` image, mount the `/tmp`
# directory from your host into the `/container/directory`
# directory in your container, and run the `ls` command to
# show the contents of that directory.
docker run \
-v /tmp:/container/directory \
alpine \
ls /container/directory

The user of this question was using Docker version 0.9.1, build 867b2a9, I will give you an answer for docker version >= 17.06.
What you want, keep local directory synchronized within container directory, is accomplished by mounting the volume with type bind. This will bind the source (your system) and the target (at the docker container) directories. It's almost the same as mounting a directory on linux.
According to Docker documentation, the appropriate command to mount is now mount instead of -v. Here's its documentation:
--mount: Consists of multiple key-value pairs, separated by commas. Each key/value pair takes the form of a <key>=<value> tuple. The --mount syntax is more verbose than -v or --volume, but the order of the keys is not significant, and the value of the flag is easier to understand.
The type of the mount, which can be bind, volume, or tmpfs. (We are going to use bind)
The source of the mount. For bind mounts, this is the path to the file or directory on the Docker daemon host. May be specified as source or src.
The destination takes as its value the path where the file or directory will be mounted in the container. May be specified as destination, dst, or target.
So, to mount the the current directory (source) with /test_container (target) we are going to use:
docker run -it --mount src="$(pwd)",target=/test_container,type=bind k3_s3
If these mount parameters have spaces you must put quotes around them. When I know they don't, I would use `pwd` instead:
docker run -it --mount src=`pwd`,target=/test_container,type=bind k3_s3
You will also have to deal with file permission, see this article.

you can use -v option from cli, this facility is not available via Dockerfile
docker run -t -i -v <host_dir>:<container_dir> ubuntu /bin/bash
where host_dir is the directory from host which you want to mount.
you don't need to worry about directory of container if it doesn't exist docker will create it.
If you do any changes in host_dir from host machine (under root privilege) it will be visible to container and vice versa.

2 successive mounts:
I guess many posts here might be using two boot2docker, the reason you don't see anything is that you are mounting a directory from boot2docker, not from your host.
You basically need 2 successive mounts:
the first one to mount a directory from your host to your system
the second to mount the new directory from boot2docker to your container like this:
1) Mount local system on boot2docker
sudo mount -t vboxsf hostfolder /boot2dockerfolder
2) Mount boot2docker file on linux container
docker run -v /boot2dockerfolder:/root/containerfolder -i -t imagename
Then when you ls inside the containerfolder you will see the content of your hostfolder.

Is it possible that you use docker on OS X via boot2docker or something similar.
I've made the same experience - the command is correct but nothing (sensible) is mounted in the container, anyway.
As it turns out - it's already explained in the docker documentation. When you type docker run -v /var/logs/on/host:/var/logs/in/container ... then /var/logs/on/host is actually mapped from the boot2docker VM-image, not your Mac.
You'll have to pipe the shared folder through your VM to your actual host (the Mac in my case).

For those who wants to mount a folder in current directory:
docker run -d --name some-container -v ${PWD}/folder:/var/folder ubuntu

I'm just experimenting with getting my SailsJS app running inside a Docker container to keep my physical machine clean.
I'm using the following command to mount my SailsJS/NodeJS application under /app:
cd my_source_code_folder
docker run -it -p 1337:1337 -v $(pwd):/app my_docker/image_with_nodejs_etc

[UPDATE] As of ~June 2017, Docker for Mac takes care of all the annoying parts of this where you have to mess with VirtualBox. It lets you map basically everything on your local host using the /private prefix. More info here. [/UPDATE]
All the current answers talk about Boot2docker. Since that's now deprecated in favor of docker-machine, this works for docker-machine:
First, ssh into the docker-machine vm and create the folder we'll be mapping to:
docker-machine ssh $MACHINE_NAME "sudo mkdir -p \"$VOL_DIR\""
Now share the folder to VirtualBox:
WORKDIR=$(basename "$VOL_DIR")
vboxmanage sharedfolder add "$MACHINE_NAME" --name "$WORKDIR" --hostpath "$VOL_DIR" --transient
Finally, ssh into the docker-machine again and mount the folder we just shared:
docker-machine ssh $MACHINE_NAME "sudo mount -t vboxsf -o uid=\"$U\",gid=\"$G\" \"$WORKDIR\" \"$VOL_DIR\""
Note: for UID and GID you can basically use whatever integers as long as they're not already taken.
This is tested as of docker-machine 0.4.1 and docker 1.8.3 on OS X El Capitan.

Using command-line :
docker run -it --name <WHATEVER> -p <LOCAL_PORT>:<CONTAINER_PORT> -v <LOCAL_PATH>:<CONTAINER_PATH> -d <IMAGE>:<TAG>
Using docker-compose.yaml :
version: '2'
services:
cms:
image: <IMAGE>:<TAG>
ports:
- <LOCAL_PORT>:<CONTAINER_PORT>
volumes:
- <LOCAL_PATH>:<CONTAINER_PATH>
Assume :
IMAGE: k3_s3
TAG: latest
LOCAL_PORT: 8080
CONTAINER_PORT: 8080
LOCAL_PATH: /volume-to-mount
CONTAINER_PATH: /mnt
Examples :
First create /volume-to-mount. (Skip if exist)
$ mkdir -p /volume-to-mount
docker-compose -f docker-compose.yaml up -d
version: '2'
services:
cms:
image: ghost-cms:latest
ports:
- 8080:8080
volumes:
- /volume-to-mount:/mnt
Verify your container :
docker exec -it CONTAINER_ID ls -la /mnt

docker run -v /host/directory:/container/directory -t IMAGE-NAME /bin/bash
docker run -v /root/shareData:/home/shareData -t kylemanna/openvpn /bin/bash
In my system I've corrected the answer from nhjk, it works flawless when you add the -t flag.

On Mac OS, to mount a folder /Users/<name>/projects/ on your mac at the root of your container:
docker run -it -v /Users/<name>/projects/:/projects <container_name> bash
ls /projects

If the host is windows 10 then instead of forward slash, use backward slash -
docker run -it -p 12001:80 -v c:\Users\C\Desktop\dockerStorage:/root/sketches
Make sure the host drive is shared (C in this case). In my case I got a prompt asking for share permission after running the command above.

For Windows 10 users, it is important to have the mount point inside the C:/Users/ directory. I tried for hours to get this to work. This post helped but it was not obvious at first as the solution for Windows 10 is a comment to an accepted answer. This is how I did it:
docker run -it -p 12001:80 -v //c/Users/C/Desktop/dockerStorage:/root/sketches \
<your-image-here> /bin/bash
Then to test it, you can do echo TEST > hostTest.txt inside your image. You should be able to see this new file in the local host folder at C:/Users/C/Desktop/dockerStorage/.

As of Docker 18-CE, you can use docker run -v /src/path:/container/path to do 2-way binding of a host folder.
There is a major catch here though if you're working with Windows 10/WSL and have Docker-CE for Windows as your host and then docker-ce client tools in WSL. WSL knows about the entire / filesystem while your Windows host only knows about your drives. Inside WSL, you can use /mnt/c/projectpath, but if you try to docker run -v ${PWD}:/projectpath, you will find in the host that /projectpath/ is empty because on the host /mnt means nothing.
If you work from /c/projectpath though and THEN do docker run -v ${PWD}:/projectpath and you WILL find that in the container, /projectpath will reflect /c/projectpath in realtime. There are no errors or any other ways to detect this issue other than seeing empty mounts inside your guest.
You must also be sure to "share the drive" in the Docker for Windows settings.

Jul 2015 update - boot2docker now supports direct mounting. You can use -v /var/logs/on/host:/var/logs/in/container directly from your Mac prompt, without double mounting

I've been having the same issue.
My command line looked like this:
docker run --rm -i --name $NAME -v `pwd`:/sources:z $NAME
The problem was with 'pwd'. So I changed that to $(pwd):
docker run --rm -i --name $NAME -v $(pwd):/sources:z $NAME

How do I link the main_folder to the test_container folder present inside the docker container?
Your command below is correct, unless your on a mac using boot2docker(depending on future updates) in which case you may find the folder empty. See mattes answer for a tutorial on correcting this.
docker run -d -v /Users/kishore/main_folder:/test_container k3_s3:latest
I need to make this run automatically, how to do that without really
using the run -d -v command.
You can't really get away from using these commands, they are intrinsic to the way docker works. You would be best off putting them into a shell script to save you writing them out repeatedly.
What happens if boot2docker crashes? Where are the docker files stored?
If you manage to use the -v arg and reference your host machine then the files will be safe on your host.
If you've used 'docker build -t myimage .' with a Dockerfile then your files will be baked into the image.
Your docker images, i believe, are stored in the boot2docker-vm. I found this out when my images disappeared when i delete the vm from VirtualBox. (Note, i don't know how Virtualbox works, so the images might be still hidden somewhere else, just not visible to docker).

Had the same problem. Found this in the docker documentation:
Note: The host directory is, by its nature, host-dependent. For this reason, you can’t mount a host directory from Dockerfile, the VOLUME instruction does not support passing a host-dir, because built images should be portable. A host directory wouldn’t be available on all potential hosts.
So, mounting a read/write host directory is only possible with the -v parameter in the docker run command, as the other answers point out correctly.

I found that any directory laying under system directive like /var, /usr, /etc could not be mount under the container.
The directive should be at user's space -v switch instructs docker daemon to mount local directory to the container, for example:
docker run -t -d -v /{local}/{path}:/{container}/{path} --name {container_name} {imagename}

Here's an example with a Windows path:
docker run -P -it --name organizr --mount src="/c/Users/MyUserName/AppData/Roaming/DockerConfigs/Organizr",dst=/config,type=bind organizrtools/organizr-v2:latest
As a side note, during all of this hair pulling, having to wrestle with figuring out, and retyping paths over and over and over again, I decided to whip up a small AutoHotkey script to convert a Windows path to a "Docker Windows" formatted path. This way all I have to do is copy any Windows path that I want to use as a mount point to the clipboard, press the "Apps Key" on the keyboard, and it'll format it into a path format that Docker appreciates.
For example:
Copy this to your clipboard:
C:\Users\My PC\AppData\Roaming\DockerConfigs\Organizr
press the Apps Key while the cursor is where you want it on the command-line, and it'll paste this there:
"/c/Users/My PC/AppData/Roaming/DockerConfigs/Organizr"
Saves a lot to time for me. Here it is for anyone else who may find it useful.
; --------------------------------------------------------------------------------------------------------------
;
; Docker Utility: Convert a Windows Formatted Path to a Docker Formatter Path
; Useful for (example) when mounting Windows volumes via the command-line.
;
; By: J. Scott Elblein
; Version: 1.0
; Date: 2/5/2019
;
; Usage: Cut or Copy the Windows formatted path to the clipboard, press the AppsKey on your keyboard
; (usually right next to the Windows Key), it'll format it into a 'docker path' and enter it
; into the active window. Easy example usage would be to copy your intended volume path via
; Explorer, place the cursor after the "-v" in your Docker command, press the Apps Key and
; then it'll place the formatted path onto the line for you.
;
; TODO:: I may or may not add anything to this depending on needs. Some ideas are:
;
; - Add a tray menu with the ability to do some things, like just replace the unformatted path
; on the clipboard with the formatted one rather than enter it automatically.
; - Add 'smarter' handling so the it first confirms that the clipboard text is even a path in
; the first place. (would need to be able to handle Win + Mac + Linux)
; - Add command-line handling so the script doesn't need to always be in the tray, you could
; just pass the Windows path to the script, have it format it, then paste and close.
; Also, could have it just check for a path on the clipboard upon script startup, if found
; do it's job, then exit the script.
; - Add an 'all-in-one' action, to copy the selected Windows path, and then output the result.
; - Whatever else comes to mind.
;
; --------------------------------------------------------------------------------------------------------------
#NoEnv
SendMode Input
SetWorkingDir %A_ScriptDir%
AppsKey::
; Create a new var, store the current clipboard contents (should be a Windows path)
NewStr := Clipboard
; Rip out the first 2 chars (should be a drive letter and colon) & convert the letter to lowercase
; NOTE: I could probably replace the following 3 lines with a regexreplace, but atm I'm lazy and in a rush.
tmpVar := SubStr(NewStr, 1, 2)
StringLower, tmpVar, tmpVar
; Replace the uppercase drive letter and colon with the lowercase drive letter and colon
NewStr := StrReplace(NewStr, SubStr(NewStr, 1, 2), tmpVar)
; Replace backslashes with forward slashes
NewStr := StrReplace(NewStr, "\", "/")
; Replace all colons with nothing
NewStr := StrReplace(NewStr, ":", "")
; Remove the last char if it's a trailing forward slash
NewStr := RegExReplace(NewStr, "/$")
; Append a leading forward slash if not already there
if RegExMatch(NewStr, "^/") == 0
NewStr := "/" . NewStr
; If there are any spaces in the path ... wrap in double quotes
if RegExMatch(NewStr, " ") > 0
NewStr := """" . NewStr . """"
; Send the result to the active window
SendInput % NewStr

To get this working in Windows 10 I had to open the Docker Settings window from the system tray and go to the Shared Drives section.
I then checked the box next to C. Docker asked for my desktop credentials to gain authorisation to write to my Users folder.
Then I ran the docker container following examples above and also the example on that settings page, attaching to /data in the container.
docker run -v c:/Users/<user.name>/Desktop/dockerStorage:/data -other -options

boot2docker together with VirtualBox Guest Additions
How to mount /Users into boot2docker
https://medium.com/boot2docker-lightweight-linux-for-docker/boot2docker-together-with-virtualbox-guest-additions-da1e3ab2465c
tl;dr Build your own custom boot2docker.iso with VirtualBox Guest
Additions (see link) or download
http://static.dockerfiles.io/boot2docker-v1.0.1-virtualbox-guest-additions-v4.3.12.iso
and save it to ~/.boot2docker/boot2docker.iso.

Note that in Windows you'll have to provide the absolute path.
Host: Windows 10
Container: Tensorflow Notebook
Below worked for me.
docker run -t -i -v D:/projects/:/home/chankeypathak/work -p 8888:8888 jupyter/tensorflow-notebook /bin/bash

i had same issues , i was trying to mount C:\Users\ folder on docker
this is how i did it Docker Toolbox command line
$ docker run -it --name <containername> -v /c/Users:/myVolData <imagename>

You can also do this with Portainer web application for a different visual experience.
First pull the Portainer image:
docker pull portainer/portainer
Then create a volume for Portainer:
docker volume create portainer_data
Also create a Portainer container:
docker run -d -p 8000:8000 -p 9000:9000 --name=portainer --restart=always -v /var/run/docker.sock:/var/run/docker.sock -v portainer_data:/data portainer/portainer
You will be able to access the web app with your browser at this URL: "http://localhost:9000". At the first login, you will be prompted to set your Portainer admin credentials.
In the web app, follow these menus and buttons: (Container > Add container > Fill settings > Deploy Container)
I had trouble to create a "mount" volume with Portainer and I realized I had to click "bind" when creating my container's volume. Below is an illustration of the volume binding settings that worked for my container creation with a mounted volume binded to the host.
P.S.: I'm using Docker 19.035 and Portainer 1.23.1

I had the same requirement to mount host directory from container and I used volume mount command. But during testing noticed that it's creating files inside container too but after some digging found that they are just symbolic links and actual file system used form host machine.

Quoting from the Official Website:
Make sure you don’t have any previous getting-started containers running.
Run the following command from the app directory.
x86-64 Mac or Linux device:
docker run -dp 3000:3000 \
-w /app -v "$(pwd):/app" \
node:12-alpine \
sh -c "yarn install && yarn run dev"
Windows (PowerShell):
docker run -dp 3000:3000 `
-w /app -v "$(pwd):/app" `
node:12-alpine `
sh -c "yarn install && yarn run dev"
Aple silicon Mac or another ARM64 device:
docker run -dp 3000:3000 \
-w /app -v "$(pwd):/app" \
node:12-alpine \
sh -c "apk add --no-cache python2 g++ make && yarn install && yarn run dev"
Explaining:
dp 3000:3000 - same as before. Run in detached (background) mode and create a port mapping
w /app - sets the “working directory” or the current directory that the command will run from
v "$(pwd):/app" - bind mount the current directory from the host into the /app directory in the container
node:12-alpine - the image to use.
Note that this is the base image for our app from the Dockerfile sh -c "yarn install && yarn run dev" - the command.
We’re starting a shell using sh (alpine doesn’t have bash) and running yarn install to install all dependencies and then running yarn run dev. If we look in the package.json, we’ll see that the dev script is starting nodemon.

Is it possible to pass a determined $MYSQL_ROOT_PASS to MySQL docker from outside docker? If so how?

So I'm writing an install script and because I haven't been able to find a solid MySQL replacement on armfh (the db must be MySQL compatible), I'm using a community on that works, however it does not initiate the db as it should. it requires me to pass the following argument.
mysql -h"db" -u"root" -p"$MYSQL_ROOT_PASSWORD" "$MYSQL_DATABASE" < /docker-entrypoint-initdb.d/1_db.sql
From inside the docker. Problem is I want this to flow naturally as a smooth install script. I've tried using the following command to pass the document and get a password prompt:
docker exec -it db bash -c "mysql -h"db" -u"root" -p"$MYSQL_ROOT_PASSWORD" "$MYSQL_DATABASE" < /docker-entrypoint-initdb.d/1_db.sql"
If also tried:
docker exec -it db bash -c "mysql -h'db' -u'root' -p'$MYSQL_ROOT_PASSWORD' '$MYSQL_DATABASE' < /docker-entrypoint-initdb.d/1_db.sql
FwIW: I used the MYSQL_ROOT_PASSWORD=$(cat /dev/urandom | tr -dc 'a-zA-Z0-9' | fold -w 64 | head -n 1) to define the password. And if I manually enter the docker send command 1 (in quotations) the db initiates.
So to summarize my question: Is it possible to pass a command like the above to activate the 1_db.sql file from outside docker?
Any help would be amazing! Thanks in advance!

Is it possible to pass a command like the above to activate the
1_db.sql file from outside docker?
you can try something like
cat 1_db.sql | docker exec -i test bash -c 'mysql -uroot -p$MYSQL_ROOT_PASSWORD $MYSQL_DATABASE'
also, remember when you try exec bash -c "mysql -uroot -p$MYSQL_ROOT_PASSWORD"it will for MYSQL_ROOT_PASSWORD in the host, not inside container, use single quotes.
determined $MYSQL_ROOT_PASS to MySQL docker from outside docker? If so
how?
docker exec -i test bash -c 'echo mysql docker password is $MYSQL_ROOT_PASSWORD'

Docker: run --net=host option changes to different shell in container

Goal: Currently trying to connect to a MySQL database on the host from the container.
Action: I have seen several answers to this problem, one being to run the container in host only mode with the --net=host option, so I can access the host using localhost/127.blah.
Result: Normally, when I execute the command
docker run -it [image_id] bash
I get a shell prompt with [user]#[container_id] (i.e., username#12345abcdef). However, when I execute the command
docker run --net=host -it [image_id] bash
it drops me to the same prompt as the host, but no longer in bold (specifically the prompt [user]#[container_id] is no longer in bold). I can tell I am in the container because of the file structure.
Question: What is happening? Am I one the right track to connecting to the database on the host? Why is it the same prompt and what is the significance of the change in font type from bold (host) to not bold (container)?
If it matters, the docker container is being created/ran inside a Vagrant machine.

The container id you normally see in your shell prompt is your container hostname. When you use --net=host, your container hostname is the same as your host's hostname. You have removed the normal network namespace that containers have. So the bash prompt with your container will look similar to the prompt on the host of you display the same fields on each. You can see the formatting of the prompt by checking the value of $PS1.
Edit: here's a comparison of different values of $PS1 from different base images (my host happens to be Debian):
$ echo $PS1
${debian_chroot:+($debian_chroot)}\u#\h:\w\$
$ docker run -it --rm debian:latest
root#4aca692dc29d:/# echo $PS1
${debian_chroot:+($debian_chroot)}\u#\h:\w\$
root#4aca692dc29d:/# exit
exit
$ docker run -it --rm ubuntu
root#b1eb8e51d672:/# echo $PS1
\[\e]0;\u#\h: \w\a\]${debian_chroot:+($debian_chroot)}\u#\h:\w\$
root#b1eb8e51d672:/# exit
exit
$ docker run -it --rm busybox
/ # echo $PS1
\w \$
/ # exit
$ docker run -it --rm centos
[root#abbfa9aa6968 /]# echo $PS1
[\u#\h \W]\$
[root#abbfa9aa6968 /]# exit
exit

Docker integration test with static data at launch

I have been struggling with this issue for a while now. I want to do a performance test using a specific set of data.
To achieve this I am using Docker-compose and I have exported a couple of .sql files.
When I first build and run the container using docker-compose build and docker-compose up the dataset is fine. Then I run my test (which also inserts) data. After running my test I want to perform it again, so I relaunch the container using docker-compose up. But this time, for some reason I don't understand, the data that was inserted the last time (by my test) is still there. I get different behaviour now.
At the moment I have the following Dockerfile:
FROM mysql:5.7
ENV MYSQL_DATABASE dev_munisense1\
MYSQL_ROOT_PASSWORD pass
EXPOSE 3306
ADD docker/data/ /docker-entrypoint-initdb.d/
Because I read that the mysql Docker image runs everything in /docker-entrypoint-initdb.d/. The first time it works properly.
I have also tried what these posts suggested:
How do i migrate mysql data directory in docker container?
http://txt.fliglio.com/2013/11/creating-a-mysql-docker-container/
How to create populated MySQL Docker Image on build time
How to make a docker image with a populated database for automated tests?
And a couple of identical other posts.
None of them seem to work currently.
How can I make sure the dataset is exactly the same each time I launch the container? Without having to rebuild the image each time (this takes kind of long because of a large dataset).
Thanks in advance
EDIT:
I have also tried running the container with different arguments like:
docker-compose up --force-recreate --build mysql but this has no success. The container is rebuilt and restarted but the db is still affected by my test. Currently the only solution to my problem is to remove the entire container and image.

I managed to fix the issue (with mysql image) by doing the following:
change the mounting point of the sql storage (This is what actually caused the problem) I used the solution suggested here: How to create populated MySQL Docker Image on build time but I did it by running a sed command: RUN sed -i 's|/var/lib/mysql|/var/lib/mysql2|g' /etc/mysql/my.cnf
add my scripts to a folder inside the container
run import.sh script that inserts data using the daemon (using the wait-for-it.sh script)
remove the sql scripts
expose port like regular
The docker file looks like this (the variables are used to select different SQL files, I wanted multiple versions of the image):
FROM mysql:5.5.54
ADD https://raw.githubusercontent.com/vishnubob/wait-for-it/master/wait-for-it.sh /utils/wait-for-it.sh
COPY docker/import.sh /usr/local/bin/
RUN sed -i 's|/var/lib/mysql|/var/lib/mysql2|g' /etc/mysql/my.cnf
ARG MYSQL_DATABASE
ARG MYSQL_USER
ARG MYSQL_PASSWORD
ARG MYSQL_ROOT_PASSWORD
ARG MYSQL_ALLOW_EMPTY_PASSWORD
ARG DEVICE_INFORMATION
ARG LAST_NODE_STATUS
ARG V_NODE
ARG NETWORKSTATUS_EVENTS
ENV MYSQL_DATABASE=$MYSQL_DATABASE \
MYSQL_USER=$MYSQL_USER \
MYSQL_PASSWORD=$MYSQL_PASSWORD \
MYSQL_ROOT_PASSWORD=$MYSQL_ROOT_PASSWORD \
DEVICE_INFORMATION=$DEVICE_INFORMATION \
LAST_NODE_STATUS=$LAST_NODE_STATUS \
V_NODE=$V_NODE \
MYSQL_ALLOW_EMPTY_PASSWORD=$MYSQL_ALLOW_EMPTY_PASSWORD
#Set up tables
COPY docker/data/$DEVICE_INFORMATION.sql /usr/local/bin/device_information.sql
COPY docker/data/$NETWORKSTATUS_EVENTS.sql /usr/local/bin/networkstatus_events.sql
COPY docker/data/$LAST_NODE_STATUS.sql /usr/local/bin/last_node_status.sql
COPY docker/data/$V_NODE.sql /usr/local/bin/v_node.sql
RUN chmod 777 /usr/local/bin/import.sh && chmod 777 /utils/wait-for-it.sh && \
/bin/bash /entrypoint.sh mysqld --user='root' & /bin/bash /utils/wait-for-it.sh -t 0 localhost:3306 -- /usr/local/bin/import.sh; exit
RUN rm -f /usr/local/bin/*.sql
ENTRYPOINT ["docker-entrypoint.sh"]
EXPOSE 3306
CMD ["mysqld"]
the script looks like this:
#!/bin/bash
echo "Going to insert the device information"
mysql --user=$MYSQL_USER --password=$MYSQL_PASSWORD --database=$MYSQL_DATABASE < /usr/local/bin/device_information.sql
echo "Going to insert the last_node_status"
mysql --user=$MYSQL_USER --password=$MYSQL_PASSWORD --database=$MYSQL_DATABASE < /usr/local/bin/last_node_status.sql
echo "Going to insert the v_node"
mysql --user=$MYSQL_USER --password=$MYSQL_PASSWORD --database=$MYSQL_DATABASE < /usr/local/bin/v_node.sql
echo "Going to insert the networkstatus_events"
mysql --user=$MYSQL_USER --password=$MYSQL_PASSWORD --database=$MYSQL_DATABASE < /usr/local/bin/networkstatus_events.sql
echo "Database now has the following tables"
mysql --user=$MYSQL_USER --password=$MYSQL_PASSWORD --database=$MYSQL_DATABASE --execute="SHOW TABLES;"
So now all I have to do to start my performance tests is
#!/usr/bin/env bash
echo "Shutting down previous containers"
docker-compose -f docker-compose.yml down
docker-compose -f docker-compose-test-10k-half.yml down
docker-compose -f docker-compose-test-100k-half.yml down
docker-compose -f docker-compose-test-500k-half.yml down
docker-compose -f docker-compose-test-1m-half.yml down
echo "Launching rabbitmq container"
docker-compose up -d rabbitmq & sh wait-for-it.sh -t 0 -h localhost -p 5672 -- sleep 5;
echo "Going to execute 10k test"
docker-compose -f docker-compose-test-10k-half.yml up -d mysql_10k & sh wait-for-it.sh -t 0 -h localhost -p 3306 -- sleep 5 && ./networkstatus-event-service --env=performance-test --run-once=true;
docker-compose -f docker-compose-test-10k-half.yml stop mysql_10k
couple of more of these lines (slightly different, cause different container names)

Running docker-compose down after your tests will destroy everything associated with your docker-compose.yml
Docker Compose is a container life cycle manager and by default it tries to keep everything across multiple runs. As Stas Makarov mentions, there is a VOLUME defined in the mysql image that persists the data outside of the container.

How do I set up network containers in Bluemix?

I've run two Alpine containers in Bluemix, used a link, and tried a tracert, and it's timing out. Is there something else I need to do to allow them to talk?
$ docker run -d --name net-a alpine sleep 99999
$ docker run -d --name net-b --link net-a:net-a alpine sleep 99999
$ docker exec -i net-b sh
traceroute net-a
traceroute to net-a (172.31.0.27), 30 hops max, 46 byte packets
1 instance-0055703a (172.31.0.28) 2998.949 ms !H 2999.897 ms !H 2999.970 ms !H
Same commands work fine with my local Docker engine.

This should just work - one noticeable difference vs local is that the containers take a bit longer to come up and have networking. Check that the first is in state Running before starting the second, and that the second is in state Running before the exec, perhaps?
Just did a quick test using one of my images (note some delays between running the below commands), and I'm seeing connectivity:
# cf ic run --name net-a myimage
ba11348e-2945-4aed-9ddf-8b85ec418423
# cf ic ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
ba11348e-294 myimage "" About a minute ago Running 4 seconds ago net-a
# cf ic run --name net-b --link net-a:net-a myimage
a45e9783-47f6-499b-ad46-2f49b275adbc
# cf ic ps -a
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
a45e9783-47f myimage "" 48 seconds ago Running 10 seconds ago net-b
ba11348e-294 myimage "" 2 minutes ago Running a minute ago net-a
# cf ic exec -ti net-b bash
root#instance-00c1f798:/# traceroute net-a
traceroute to net-a (172.31.0.22), 30 hops max, 60 byte packets
1 net-a (172.31.0.22) 2.398 ms 2.199 ms 2.224 ms