Stops nginx from routing to stale container IPs after host reboots, Docker upgrades without live-restore, or runtime container restarts. The bundled `nginx.conf.sigil` can use an nginx `resolver` directive against the local `coredns-docker` socket and a variable-based `proxy_pass` keyed on `<app>.<process-type>.docker`, so the upstream is resolved at request time honoring DNS TTL. `coredns-docker` becomes a hard `Depends:` of the dokku deb and is bundled in the Docker image under runit. Closes#8313.
The wrapper config used to run `nginx -t` against an app's rendered `nginx.conf` declared no `user` directive, so nginx fell back to its compile-time default and chowned every `proxy_cache_path` directory to that user, including directories that already existed. On Debian and Ubuntu that default is `nobody`, so deploying an app shipping a custom `nginx.conf.sigil` took the running server's live cache directory away for the length of the build and turned every request the cache would serve into an HTTP 500, healing only once the deploy's own reload restored the ownership. The wrapper now declares the same `user` as the host's main nginx config, which is also exposed to custom `validate-config` templates as `$.NGINX_USER`.
On newer Ubuntu Noble, cron (among other things) has a transitory dependency on systemd, forcing our systemd/runit detection to fail. This change allows overriding the detected init on Ubuntu systems, ensuring we always do the right thing when interacting with init.
In most cases, it should be there, but it appears that system upgrades from Ubuntu 16.04 sometimes do not have usr-merge setup. As such, we need to fallback to the /bin/systemctl path if it exists.
This will allow users to switch to a docker label-based nginx proxy server transparently without rebuilds (after the first one), which would mimic how the caddy, haproxy, and traefik proxy implementations work.
See https://github.com/dokku/nginx-docker-proxy for an example server implementation.
The previous method was a bit more difficult for users to interact with as they needed to ensure the file was at a certain path in the built image. This change divorces the file from the built artifact, better aligning the nginx.conf.sigil handling with the rest of Dokku's monorepo support.
A side effect of this is now building proxy configurations does not require access to the built image.
Closes#5207
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.
The default for an app is empty, meaning it will fallback to the global value (which defaults to true). This will allow users to properly override the global value, regardless of what it is.
Unfortunately, Docker does not persist container IPs on boot, so containers may come up with the wrong IP, resulting in proxying to the incorrect app after a server restart. Instead, we should delete the configs and let ps:restore rebuild the proxy config correctly.
Closes#3932
This standardization makes it somewhat easier to read stacktraces as the command names are all uniform, so it will be slightly easier to scan trace output.
This enables the HSTS header by default when an SSL certificate is in use. HSTS options can also be managed via the nginx:set command, which also exposes the ability to disable HSTS for your application.
This change allows supporting the openresty package as installed on Ubuntu 18.04 from the official openresty package repository, while also returning general nginx support.
Closes#3675
This will allow background processes to cleanup these files if they were unable to be cleaned up during a run of the dokku binary, as is the case for files copied by the docker binary when running in docker.