question: (basic) best practice for using herokuish to create a runnable image

[tony-kerz]

hi,

apologies in advance for beginner question (feel free to redirect me to more appropriate forum).

i have been using tutum/buildstep to build a runnable image like so as part of a home-grown build pipeline using jenkins, git and a docker runtime platform.

i recently wanted to use an iojs runtime in a node app, and it turns out that requires a more recent node buildpack than the one baked into the version of tutum/buildstep on dockerhub.

trying to gain some background context, i did some research and came across this thread about preferring herokuish over buildstep.

along those lines, i would like to use herokuish directly without tutum or progrium buildsteps, but i'm not clear on exactly how to accomplish that.

for example would i use:

1. some form of docker run gliderlabs/herokuish at the command line, or
2. FROM gliderlabs/herokuish with some other commands in a Dockerfile

any guidance or examples would be appreciated

best,
tony.

[progrium]

You'd use docker run. I suppose the readme doesn't explain this well? I'll need to write better docs for end users.

[tony-kerz]

i toyed with docker run, but it wasn't clear to me how to get to a built docker image that way.

i was able to get the desired effect with a Dockerfile that looks like this:

FROM gliderlabs/herokuish

RUN mkdir -p /app
ADD . /app
RUN /build

ENTRYPOINT ["/exec", "node", "/app/my-app.js"]

and running:

docker build -t me/my-image .
docker run --rm me/my-image

(in my case, my app is just a node command-line app)

does that seem like a reasonable approach?

[progrium]

Yeah that works alright, too.

On Wed, Aug 12, 2015 at 8:58 AM, tony kerz notifications@github.com wrote:

i toyed with docker run, but it wasn't clear to me how to get to a built
docker image that way.

i was able to get the desired effect with a Dockerfile that looks like
this:

FROM gliderlabs/herokuish

RUN mkdir -p /app
ADD . /app
RUN /build

ENTRYPOINT ["/exec", "node", "/app/my-app.js"]

and running:

docker build -t me/my-image .
docker run --rm me/my-image

(in my case, my app is just a node command-line app)

does that seem like a reasonable approach?

—
Reply to this email directly or view it on GitHub
#62 (comment)
.

Jeff Lindsay
http://progrium.com

[msabramo]

What would the docker run approach look like?

[msabramo]

docker run --rm -v $(pwd):/tmp/app gliderlabs/herokuish /bin/herokuish buildpack build foobar

seems to execute without any errors, but where is the new image?

[Haybu]

docker run --rm -v $(pwd):/tmp/app gliderlabs/herokuish /bin/herokuish buildpack build foobar

works all good, where is the built image gets saved ?

[josegonzalez]

If you do not docker commit the change, its simply in the ephemeral container that is running. You can either specify a name --name or use docker commit to save the image id.

[joshmanders]

The built container is the one you run, you added the --rm flag to the run command, which means after the container exited, to remove it.