Because the app-json extraction happens before the build-dir is set, it doesn't end up respecting the value, and thus the user might get the wrong app.json file.
This takes advantage of path.Join()'s default of ignoring empty paths.
The URLS file is no longer necessary as urls are now generated directly from the app VHOST file.
This also moves all the url generation logic to the domains plugin.
This change also moves the referenced Procfile out to a host path once on deploy vs potentially several times, which should speed up deploys a small amount and simplify reasoning about the file.
Closes#5112
We will no longer support CentOS, Fedora, and Opensuse as installation targets. These are not actively maintained by anyone with commit rights and occasionally cause issues for users as they are not tested during the release process.
Rather than have subpar support for an untested operating system, we're removing support for them completely. Users of these operating systems should take a look migration to the docker-based installation method, which will always be tested and supported by the project.
Additionally, drop support for Debian 9 as it is now EOL.
This increases the time to list apps from .05s to .2s~.3s on the simplest check, but the increased time is worth it for actually verifyng access to an app (should that need arise).
We'll also want to add the same to app existence checks.
This ensures the value is correct, where previously it was computed based on a file - and therefore did not distinguish between git:from-archive and git:from-image.
Closes#4464
This changes restarts to drop the 'release' part that currently gets triggered, which also helps reduce the possibility that a new image layer will be added due to predeploy deploy triggers.
Note that the old method essentially halfway-rebuilt the app (which was what added the extra layer). This refactor removes that, which is a bc-break.
Additionally, scaling processes will result in deploying _only_ restarting the processes being scaled.
Closes#2184
This makes the installation a bit more secure by ensuring a user does not accidentally expose a way for unauthorized users to add new ssh keys to the system.
Additionally, this removes the extra HOSTNAME file to make the initial install process easier (that file was not modifiable by any dokku commands.
Closes#2247
This is useful when there is a service not managed by Dokku but should be exposed via the Dokku routing layer. As an example, some binaries (consul, nomad, vault) expose web uis, and are traditionally run on the host directly vs in a container.
Closes#4665
Rather than hardcode two builders, allow builders to specify a `builder-detect` trigger. This trigger can be used to specify if the builder should or should not be used for an application. Each builder takes stdin and can decide if it wants to emit it or emit it's own image source type.
If the final value is empty, then Dokku will default to herokuish (and cnb once that is stable). In addition, a future change may allow users to manually specify a builder in the case they wish to override the choice selected by Dokku.
This change enables users to build custom builder plugins and have those plugins used for building an image asset. By way of example, an enterprising user could create a `builder-lambda` based on lambci, and then pair this with a scheduler plugin that updates a lambda function on AWS. Alternatively, a user might decide they wish to place their Dockerfile in a specific directory for their applications - such as an `_infrastructure` directory - and create a plugin to override how that is detected within Dokku.
Previously, checking if an app was deployed actually checked if there were any running processes. This is not only incorrect, but also fails to take into account applications that do not have running processes and are only used for one-off commands.
This fix migrates apps to the new "deployed" property which is set in core-post-deploy. The result is a slightly faster "deployed" check that is correct and allows non-scaled apps to actually work.
Closes#4398
This will allow dropping many of the "source config/functions" calls littered over the codebase. Plugin triggers are the way to perform cross-plugin communication :)
This allows alternative plugins - such as dokku-letsencrypt - to inject scheduled cron tasks into the cron system used by Dokku. Cron systems can choose to include or not include a cron task based on the specified scheduler, and can also optionally use a third parameter to store arbitrary information.
