It's just a remote builder, that will build whichever platform is asked
for, so let's remove the "native" part.
We'll also remove the service name from the builder name, so multiple
services can share the same builder.
Combine the two builders, as they are almost identical. The only
difference was whether the platforms were set.
The native cached builder wasn't using the context it created, so now
we do.
We'll set the driver to `docker-container` - it seems to be the default
but the Docker docs claim it is `docker`.
The integrations tests use their own registry so avoid hitting docker
hub rate limits.
This was using a self signed certificate but instead use
`--insecure-registry` to let the docker daemon use HTTP.
Allow blocks prefixed with `x-` in the configuration as a place to
declare reusable blocks with YAML anchors and aliases.
Borrowed from the Docker Compose configuration file format -
https://github.com/compose-spec/compose-spec/blob/main/spec.md#extension
Thanks to @ruyrocha for the suggestion.
Find the first registry mirror on each host. If we find any, pull the
images on one host per mirror, then do the remainder concurrently.
The initial pulls will seed the mirrors ensuring that we pull the image
from Docker Hub once each.
This works best if there is only one mirror on each host.
Setting `GITHUB_TOKEN` as in the docs results in reusing the existing
`GITHUB_TOKEN` since `gh` returns that env var if it's set:
```bash
GITHUB_TOKEN=junk gh config get -h github.com oauth_token
junk
```
Using the original env ensures that the templates will be evaluated the
same way regardless of whether envify had been previously invoked.
We didn't log boot errors if there was one role because there was no
barrier and the logging is done by the first host to close the barrier.
Let's always create the barrier to fix this.
- Add local logout to `kamal registry logout`
- Add `skip_local` and `skip_remote` options to `kamal registry` commands
- Skip local login in `kamal deploy` when `--skip-push` is used
Load the hosts from the contexts before trying to build.
If there is no context, we'll create one. If there is one but the hosts
don't match we'll re-create.
Where we just have a local context, there won't be any hosts but we
still inspect the builder to check that it exists.
Validate the Kamal configuration giving useful warning on errors.
Each section of the configuration has its own config class and a YAML
file containing documented example configuration.
You can run `kamal docs` to see the example configuration, and
`kamal docs <section>` to see the example configuration for a specific
section.
The validation matches the configuration to the example configuration
checking that there are no unknown keys and that the values are of
matching types.
Where there is more complex validation - e.g for envs and servers, we
have custom validators that implement those rules.
Additonally the configuration examples are used to generate the
configuration documentation in the kamal-site repo.
You generate them by running:
```
bundle exec bin/docs <kamal-site-checkout>
```
When cloning the git repo:
1. Try to clone
2. If there's already a build directory reset it
3. Check the clone is valid
If anything goes wrong during that process:
1. Delete the clone directory
2. Clone it again
3. Check the clone is valid
Raise any errors after that
If a primary role container is unhealthy, we might take a while to
timeout the health check poller. In the meantime if we have started the
other roles, they'll be running tow containers.
This could be a problem, especially if they read run jobs as that
doubles the worker capacity which could cause exessive load.
We'll wait for the first primary role container to boot successfully
before starting the other containers from other roles.
To speed up deployments, we'll remove the healthcheck step.
This adds some risk to deployments for non-web roles - if they don't
have a Docker healthcheck configured then the only check we do is if
the container is running.
If there is a bad image we might see the container running before it
exits and deploy it. Previously the healthcheck step would have avoided
this by ensuring a web container could boot and serve traffic first.
To mitigate this, we'll add a deployment barrier. Until one of the
primary role containers passes its healthcheck, we'll keep the barrier
up and avoid stopping the containers on the non-primary roles.
It the primary role container fails its healthcheck, we'll close the
barrier and shut down the new containers on the waiting roles.
We also have a new integration test to check we correctly handle a
a broken image. This highlighted that SSHKit's default runner will
stop at the first error it encounters. We'll now have a custom runner
that waits for all threads to finish allowing them to clean up.
Docker does not respect the .dockerignore file when building from a tar.
Instead by default we'll make a local clone into a tmp directory and
build from there. Subsequent builds will reset the clone to match the
checkout.
Compared to building directly in the repo, we'll have reproducible
builds.
Compared to using a git archive:
1. .dockerignore is respected
2. We'll have faster builds - docker can be smarter about caching the
build context on subsequent builds from a directory
To build from the repo directly, set the build context to "." in the
config.
If there are uncommitted changes, we'll warn about them either being
included or ignored depending on whether we build from the clone.
