gmackie/ForgeGraph
1
0
Fork 0
Code Issues 8 Pull requests 9 Projects Releases Packages Wiki Activity Actions

fix(nix): refresh the registry pnpmDeps hash — npm.forgegraf.com could not build #491

Closed
gmackie wants to merge 0 commits from fix/registry-pnpm-deps-hash into main
pull from: fix/registry-pnpm-deps-hash
merge into: gmackie:main
Conversation 1 Commits - Files changed -
gmackie commented 2026-08-27 19:12:32 +00:00
Owner
Copy link

One of 8 critical deploy_failed alerts, and the one that is ours.

npm-registry has not deployed since 2026-08-13. Every attempt dies the same way:

error: hash mismatch in fixed-output derivation forgegraph-registry-pnpm-deps.drv
  specified: sha256-cSpsy3dC16ypMlnRe7hSygIEGukW+vw/ez415BnlH9g=
       got:  sha256-KPYVfGFQcIJpoXeea+uJpzxs7I4AWW1Ub15oHyCN69k=

The flake's pnpmDeps hash still describes an older pnpm-lock.yaml. That matters more than one red app: npm.forgegraf.com is the registry serving every @forgegraph package — agent releases and @forgegraph/check-events publish through it — so a stale hash here quietly blocks the publish path while everything else looks green.

The replacement hash is the build's own reported got: at rev 145d0cd17f, which is still main's head with pnpm-lock.yaml untouched since (verified: no commit between that rev and main touches the lockfile), so it is current rather than already-drifted.

Diagnosis of the other seven, for the record — they are six unrelated causes, not one outage:

  • playtrek, playtrek-engine — same ERR_PNPM_NO_OFFLINE_TARBALL / pnpmDeps drift, in their own repo
  • jobs-pulse — wrangler deploy --name jobs-pulse-onebox exits 1
  • gt-ops-edge — the frappe-edge-proxy systemd unit fails to start
  • linear-clone — native better-sqlite3 install fails
  • latchflow-beta — cp: cannot stat wrangler.beta.jsonc: the file is gone
  • check-events-demo — mine, from the fork-bomb build; retires on its own

🤖 Generated with Claude Code

One of 8 critical `deploy_failed` alerts, and the one that is ours. **npm-registry has not deployed since 2026-08-13.** Every attempt dies the same way: ``` error: hash mismatch in fixed-output derivation forgegraph-registry-pnpm-deps.drv specified: sha256-cSpsy3dC16ypMlnRe7hSygIEGukW+vw/ez415BnlH9g= got: sha256-KPYVfGFQcIJpoXeea+uJpzxs7I4AWW1Ub15oHyCN69k= ``` The flake's `pnpmDeps` hash still describes an older `pnpm-lock.yaml`. That matters more than one red app: npm.forgegraf.com is the registry serving every `@forgegraph` package — agent releases and `@forgegraph/check-events` publish through it — so a stale hash here quietly blocks the publish path while everything else looks green. The replacement hash is the build's own reported `got:` at rev `145d0cd17f`, which is still main's head with `pnpm-lock.yaml` untouched since (verified: no commit between that rev and main touches the lockfile), so it is current rather than already-drifted. **Diagnosis of the other seven, for the record** — they are six unrelated causes, not one outage: - `playtrek`, `playtrek-engine` — same `ERR_PNPM_NO_OFFLINE_TARBALL` / pnpmDeps drift, in their own repo - `jobs-pulse` — `wrangler deploy --name jobs-pulse-onebox` exits 1 - `gt-ops-edge` — the frappe-edge-proxy systemd unit fails to start - `linear-clone` — native `better-sqlite3` install fails - `latchflow-beta` — `cp: cannot stat wrangler.beta.jsonc`: the file is gone - `check-events-demo` — mine, from the fork-bomb build; retires on its own 🤖 Generated with [Claude Code](https://claude.com/claude-code)
gmackie closed this pull request 2026-08-27 19:32:52 +00:00
gmackie reopened this pull request 2026-08-27 19:44:30 +00:00
gmackie commented 2026-08-27 19:44:53 +00:00
Author
Owner
Copy link

Superseded by a fresh PR from the same branch — see below.

This PR's record is stuck on a stale cached merge-base. It was opened when the branch pointed at 8cf344f2 (a commit from 2026-08-17 that is already in main, which is why this PR showed 0 files for so long). I have now pushed the actual fix, but Forgejo still reports merge_base: 8cf344f2 and renders a bogus 696 files / +74300 −43101 diff. Git disagrees:

merge-base(head, main):  e641b37d   (== main)
files in head...main:    1

Closing and reopening did not clear the cache, so the PR is reopened cleanly rather than left displaying a diff that would be alarming and unreviewable.

The fix itself is on fix/registry-pnpm-deps-hash at 99c34804 — one line in flake.nix, and the hash was computed rather than transcribed (built .#registry on hetzner-worker with a sentinel hash and took what nix reported). Worth noting the value quoted in this PR's original description differs from the computed one by a single character — …KPYVfGFW… vs …KPYVfGFQ… — which would have reproduced the same deploy failure while looking right.

Superseded by a fresh PR from the same branch — see below. This PR's record is stuck on a **stale cached merge-base**. It was opened when the branch pointed at `8cf344f2` (a commit from 2026-08-17 that is already in `main`, which is why this PR showed 0 files for so long). I have now pushed the actual fix, but Forgejo still reports `merge_base: 8cf344f2` and renders a bogus **696 files / +74300 −43101** diff. Git disagrees: ``` merge-base(head, main): e641b37d (== main) files in head...main: 1 ``` Closing and reopening did not clear the cache, so the PR is reopened cleanly rather than left displaying a diff that would be alarming and unreviewable. **The fix itself is on `fix/registry-pnpm-deps-hash` at `99c34804`** — one line in `flake.nix`, and the hash was computed rather than transcribed (built `.#registry` on hetzner-worker with a sentinel hash and took what nix reported). Worth noting the value quoted in this PR's original description differs from the computed one by a single character — `…KPYVfGFW…` vs `…KPYVfGFQ…` — which would have reproduced the same deploy failure while looking right.
gmackie closed this pull request 2026-08-27 19:44:54 +00:00
All checks were successful
CI / gitleaks (pull_request) Successful in 6s
Required
Details
CI / ci (pull_request) Successful in 9m9s
Required
Details
forgegraph/ci CI passed

Pull request closed

This pull request cannot be reopened because the branch was deleted.
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels   
community
Community process, governance, contribution UX

  
docs
Documentation improvements

  
forgegraph:public-update
Publish approved PR lifecycle updates to the ForgeGraph public feed.

  
good first issue
Suitable for first-time contributors

  
help wanted
Extra attention is needed; maintainers welcome a PR

No labels
community
docs
forgegraph:public-update
good first issue
help wanted
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
gmackie/ForgeGraph!491
No description provided.