The docker-local scheduler now sends `SIGTERM` to old containers immediately after a successful deploy via `docker container kill --signal=SIGTERM`, rather than waiting `wait-to-retire` seconds before signaling. This matches Heroku's graceful-shutdown contract and lets applications begin draining in-flight work as soon as proxy traffic switches. The existing `wait-to-retire` grace period and `stop-timeout-seconds` hard-stop continue to apply unchanged as the authoritative cleanup path.
Fresh apt installs now drop a catch-all server block at `/etc/nginx/conf.d/00-default-vhost.conf` that uses `ssl_reject_handshake on` and `return 444` to drop requests with unknown Host headers. Conflicting upstream nginx default vhosts are renamed to `*.dokku-disabled` rather than deleted, preserving any local edits. The new `dokku/install_default_site` debconf flag opts out of the install. Upgrades leave existing nginx config untouched.
Custom nginx.conf.sigil templates that reference DOKKU_APP_WEB_LISTENERS
may now receive an empty value when rendered for apps without running web
processes. The migration guide documents how to handle this with a
conditional in the template.
The buildpacks entry in tests/apps/dockerfile/app.json caused
the post-extract trigger to write a .buildpacks file during
Dockerfile-based deploys, breaking builder detection. Remove
the entry and add Go unit tests for getBuildpacks/validBuildpackURL,
a bats integration test for deploying with app.json buildpacks,
and documentation for the new feature.
Some files - those maintained by external organizations - have a very light treatment and solely point to the upstream documentation to reduce any issues creating examples/documentation for them that may differ in the future.
Closes#7315
The ingress-nginx ingress implementation is the standard ingress in the Kubernetes community, and it doesn't make sense for us to stray from that just to utilize the k3s default.
In the future, we might drop k3s, but this works well for now.
Rather than require a heavy chown operation across various paths, just chown the files already in the built image during the release process. This ensures we can skip not-only the chown process during the container start that herokuish injects, but also the one that Dokku runs which modifies mounted container paths as well during the pre-deploy.
Note that users will need to ensure any mounted volumes don't have permissions reset by other processes or containers won't be able to access them.
Building/testing for ARM does not happen often - the only runtime environment is Raspberry PI, which supports ARM64 - and complicates support for a ton of features. Aside from that, CI runs are much longer for ARM Dokku images, often reaching 15-20 minutes or just timing out completely.
Rather than support an architecture that doesn't have much usage by maintainers and has a lot of maintenance burden, we're removing the platform.
AMD64 and ARM64 continue to be supported, but ARM in CI is not currently supported and doesn't get local testing either, so deprecating it now and removing it in the future makes most sense for the project.
