Repository navigation
build: add cf recovery deploy for website - #71
Closed
WillieCubed wants to merge 4 commits into
Closed
WillieCubed wants to merge 4 commits into
WillieCubed wants to merge 4 commits into
Conversation
Manual recovery deployments use a typed Cloudflare configuration for the built Worker and static assets. Production and preview settings are checked against the Wrangler fallback used by the versioned candidate gate and local tests. Co-authored-by: Codex <noreply@openai.com>
WillieCubed
had a problem deploying
to
worker-preview
September 30, 2026 03:00 — with 
GitHub Actions
Error
GitHub Actions
Error
Validation builds the site before the Cloudflare dry run, so missing Worker or asset output fails the required repository check. Co-authored-by: Codex <noreply@openai.com>
Production and preview recovery deployments use CLOUDFLARE_ACCOUNT_ID. Keep the account identifier out of the typed configuration while the existing database bindings and release gate remain pinned. Co-authored-by: Codex <noreply@openai.com>
WillieCubed
had a problem deploying
to
worker-preview
September 30, 2026 05:09 — with
GitHub Actions
Failure
GitHub Actions
Failure
WillieCubed
had a problem deploying
to
worker-preview
September 30, 2026 05:23 — with
GitHub Actions
Failure
GitHub Actions
Failure
Keep the recurring-event assertion in its dedicated suite. Its title and heading are ready when each page loads, allowing preview checks to finish within the test limit. Co-authored-by: Codex <noreply@openai.com>
WillieCubed
had a problem deploying
to
worker-preview
September 30, 2026 05:44 — with
GitHub Actions
Failure
GitHub Actions
Failure
Contributor
Author
|
Resolved by merged replacement #95 (eacbc6b). The recovery deploy now matches current production and preview routes, environment identity, and database isolation, while preserving all stored credentials. The original preview failure is resolved on the current baseline. Full pnpm check, both cf dry runs, hosted Worker browser acceptance, and every GitHub quality check pass. The replacement preserves linear history without force-pushing this branch. |
This branch had an error being deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
TL;DR
Add a typed
cfdeployment path for manual website recovery without changing the reviewed production candidate gate.Overview of Changes
A separate deploy package lets
cfbundle the existing compiled Worker and static site assets without Astro intercepting the build. Its production and preview modes preserve the separate D1 databases and runtime settings. Parity tests compare both modes with the Wrangler fallback, and the vendored LVBT standard makespnpm run deployselect onecftarget.pnpm check, the production dependency audit, production and previewcf deploy --dry-run, andpnpm run deploy --dry-runpass. The production dry run includes the site Worker, 409 assets, and the production database; the preview dry run uses the preview database. These commands upload nothing.The existing GitHub candidate workflow still uses Wrangler to upload a version, test its preview URL, confirm the current
maincommit, and promote that exact version. This preserves the live release gate whilecfversion receipt handling is validated. Wrangler also remains for local tests, generated types, and individual secret changes.Follow-ups
Move the candidate workflow to
cf workers versions createandcf workers deployments createonly after a live candidate receipt and exact-version promotion are proven. Do not use the manual recovery command for normal releases.