You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
bin/deploy / gemstack deploy: runs Kamal with the root .env.production.local, .env.production
and .env loaded (never .env.local), so .kamal/secrets resolves from the same .env as everything
else — Kamal doesn't read .env itself. Without arguments it deploys, refusing while there are
uncommitted changes (Kamal builds the last commit); other Kamal commands pass through
(bin/deploy setup, bin/deploy logs). gemstack generate deploy writes bin/deploy.
gemstack update updates the deploy files too (Dockerfile, config/deploy.yml, .kamal/secrets, bin/deploy…), the same way as the app's templates: unedited files are updated, edited ones can be
merged — starting from the exact template each was generated from, which .gemstack/generators.json
records — and files a release adds (bin/deploy) are created. Before, deploy files never got a
release's fixes after gemstack generate deploy.
gemstack generate deploy: the api role is the primary role, so it boots before the web role, whose
health check renders pages that call the API (the first deploy could fail before); the web role enables
kamal-proxy explicitly. config/deploy.yml shows how to serve www and the bare domain and to set the
SSH user, .kamal/secrets how to use GitHub's registry; the generator says to run bundle lock --add-checksums when Gemfile.lock lists gem checksums. The deployment guide covers these
and adding your own settings.