The previous method used an ephemeral container, which was slightly more convoluted and required cleanup. The new method also allows us to set an initial WORKDIR, which is useful for ensuring 'docker exec' works as expected.
Previously, we would always set the port mapping during a dockerfile build, making it difficult for users to override mappings. We also only _sometimes_ updated the detected port mapping, further confusing issues when users were migrating from Dockerfile to Buildpacks for builds.
Now, we always detect the port mapping during the build process, and only use that detected port mapping if an override is not specified. This greatly simplifies the experience around port mapping, as now a user can create an app, set a port mapping, and that first deploy will respect the port mapping without an additional deploy.
The builder always has the best context for what the app should be listening on, and thus we can always specify a "default" port mapping at this stage. Users can override this map as desired later.
This change also results in the removal of a ton of internal code that is now centralized in the ports plugin.
Closes#4067
During an app build, we now auto-detect ports based on the source code. This is usually http:80:5000, with Dockerfile-based deploys having their ports extracted from the docker image or Dockerfile. Additionally, we add an https:443 mapping for any detected http:80 mapping when there is an ssl certificate, and all http port mappings are transformed to https mappings for Dockerfile-based deploys.
While the ports aren't currently consumed, a future refactor will provide the ability to fallback to the new detected ports when there is no user-specified port mapping.
The shorthand is more prevalent in this codebase, and is something that bash supports, so we should just use the same thing everywhere.
Note that we do not use shorthand redirect in Makefile as shell parsing is a bit different in Make and the shorthand redirect doesn't seem to be properly supported, causing CI errors.
By removing the image cleanup registration, we avoid cases where we delete gliderlabs/herokuish, which may cause issues for users in airgapped installations (where the image may not be pullable from a registry).
The previous lever for manipulating this was the DOKKU_IMAGE environment variable. While this is all well and good, it only works for CNB and is yet another `DOKKU_*` environment variable that need not exist.
While this does not add support for manipulating the CNB stack just yet, the groundwork is set for the future.
Closes#4306
Rather than waiting for the next deploy, take advantage of the retire system to retire these containers immediately.
Note that since the retire process happens out of band, the containers may stick around for up to 5 minutes, after which point they will be removed.
Customization of the wait time to retire - currently 60 seconds - is up for future debate. The containers ultimately haven't been useful for debugging, so keeping them around for longer won't help in most cases, and folks can disable the dokku-retire service (or cron) if that is desired.
Containers can be attached:
- after they are created, but before they are started
- after a successful deploy, but before the proxy reloads
This allows folks to have flexibility around when they would like a container to be made available to a network.
The IMAGE_TAG is not always available, while the IMAGE_SOURCE_TYPE is. Since the point of these triggers was to unify the argument passed and make it easier to interact with them, this version makes more sense.
Closes#3780
This PR uses the new syntax for commands introduced in Docker 1.13, making it a bit easier to understand just what a particular command is trying to do.
This also pushes the long-form syntax for docker command flags, which are also easier to understand at a glance.
This will allow those triggers to insert and extract files from the sourcecode work directory. Note that changes are not committed to this directory, nor will they be reflected in the committed codebase.
This is a "one-shot" ability, meaning that it is highly discouraged for multiple plugins to implement the triggers.
The `builder-create-dokku-image` trigger is meant to modify the base DOKKU_IMAGE that will be used to build an application. A good use of the hook is to inject system packages by:
- calculating an image tag from the set of system packages
- checking that the image doesnt already exist
- creating the image if it does not
As the output is sent to stdout, we have to find an alternative way of fetching the new DOKKU_IMAGE. The `builder-dokku-image` trigger can output an image tag based on the calculated image tag. If no tag is output, then the DOKKU_IMAGE is used as is.
Using this pattern, users can efficiently modify the base image in use for an app without duplicating the work on every push. This is especially useful for package installation, where packages are required for app building, but may take a long time and can be error prone.
A builder plugin implements some method for 'building' artifacts from a given source code archive. This refactor moves the majority of the image building code into separate plugins, allowing us to implement other builders, but also simplifying the common pugin codebase.
Also move the pre-build-buildpack hook from the build-env plugin to builder-herokuish.
