Problem: In my recent PR(#264), I accidentally picked overlay output
changes from the philtaken/nixos-vm-tests branch, which broke existing
overlay imports after updating the deploy-rs input.
Solution: Add backwards compatibility so that users don't have to make
changes to their nix flakes.
Problem: Currently, the only way to test deploy-rs deployments is to
actually do a deployment to an existing NixOS instance (either in VM,
or a real machine) manually. This is a bit inconvenient and one can
forget to test changes when developing/reviewing deploy-rs changes.
Solution: Add NixOS VM tests.
Problem: It's possible to use 'deploy-rs' for deploying 'darwinSystem'
configuration from 'nix-darwin' to a darwin system. However, there is no
dedicated activatiot script for darwin and thus one has to come up with
'custom' activation script.
Solution:
1) Add 'darwin' attribute to 'lib.activate' that provides a script that
should be used to activate 'darwinSystem' config with 'deploy-rs'.
2) Add a new 'examples/darwin' example that provides simple flake for
deploying configuration to a darwin target.
Use the deploy-rs from the final packages set. This can avoid rebuilding
deploy-rs when using it in a nixos config. It can use the version cached
in nixpkgs.
Also add instructions to the readme on how to craft an overlay that uses
nixpkgs deploy-rs.
This can be useful when e.g. deploying a kernel update to a target host.
You usually plan a reboot (or kexec) after that to activate the new
kernel. However you don't want to wait for services to be restarted
first since these will be "restarted" anyways on the reboot. In cases
like GitLab or the Atlassian stack this actually makes a difference.
This patch changes the following things:
* If `--boot` is provided, `nix-env -p profile-to-activate --set` is
called for each deployed profile to make sure that it is activated
automatically after a reboot.
* However, the actual activation (e.g. `switch-to-configuration switch`)
is skipped. Instead:
* For NixOS, `switch-to-configuration boot` is called to set the new
profile as default in the bootloader.
* For everything else, nothing else is done. The profile is already
the new default (and thus picked up on the next boot).
nix 2.7 renamed defaultApp and defaultPackage. Both the
old and new names are supported in 2.7, but 2.8 has removed
support for the old names, breaking the nix run invocation.
Old names are kept in this PR to keep compatibility with nix 2.6,
but could be removed if support of this version is not needed anymore.
In 5d5da48 (https://github.com/serokell/deploy-rs/pull/81), $DRY_ACTIVATE
is used in a bash script with -u. When DRY_ACTIVATE is not set (which it
is not for older profiles), the script fails. Fix this by setting a fallback.
Now that activate uses a cross-platform fs notification library, we
can build it on Darwin again (and build its deps separately, speeding
up recompiles there).
* Add xcbuild as native build input 2 ways & pass --verbose to cargo build
* Make "deploy" build on macOS
* Bump the nixpkgs version to one that ships Rust 1.47, which can
produce binaries under macOS 11 (Big Sur).
* Use darwin-specific options that let Naersk build the "deploy"
binary under macOS. (With a work-around for
https://github.com/nmattia/naersk/issues/127, which prevents this
from being much more straight-forward).
Unfortunately, the "activate" binary can't be built under macOS due to
the inotify dependency; that means the best we can do is to run the
deploy under macOS, not deploy *to* macOS.
