gmackie/ForgeGraph

Fork 0

ci(release-agent): one nix shell for all targets, 90 min timeout #440

Merged

gmackie merged 3 commits from release-dispatch/0.1.57 into main

2026-08-27 02:09:25 +00:00

gmackie commented

2026-08-25 04:59:52 +00:00

Owner

The first Release agent run (14287) built each arch in its own nix shell — a cold nixpkgs evaluation + Go toolchain fetch every time — and hit the 30-minute job timeout, which Forgejo reports as cancelled. This enters the shell once for all three builds and raises the timeout to 90 minutes.

Also worth knowing for whoever dispatches next: a workflow_dispatch run on main is cancelled the moment anyone merges to main (run 14277 died that way), so dispatch from a release-dispatch/<version> branch — the workflow checks out origin/main itself.

🤖 Generated with Claude Code

The first `Release agent` run (14287) built each arch in its own `nix shell` — a cold nixpkgs evaluation + Go toolchain fetch every time — and hit the 30-minute job timeout, which Forgejo reports as *cancelled*. This enters the shell once for all three builds and raises the timeout to 90 minutes. Also worth knowing for whoever dispatches next: a `workflow_dispatch` run on `main` is cancelled the moment anyone merges to `main` (run 14277 died that way), so dispatch from a `release-dispatch/<version>` branch — the workflow checks out `origin/main` itself. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

gmackie added 1 commit

2026-08-25 04:59:52 +00:00

ci(release-agent): one nix shell for all targets, 90 min timeout

CI / gitleaks (pull_request) Successful in 6s

Details

CI / storybook (pull_request) Successful in 1m21s

Details

forgegraph/ci CI passed

CI / ci (pull_request) Successful in 9m32s

Details

f166b07505

The first Release agent run built each arch in its own nix shell — a cold
nixpkgs eval + toolchain fetch each time — and hit the 30 min job timeout
(reported by Forgejo as cancelled).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

gmackie commented

2026-08-25 05:03:08 +00:00

Author

Owner

Preview environment is live: https://pr-440-forgegraph.forgegraf.com

Deployed f166b075 with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

Preview environment is live: https://pr-440-forgegraph.forgegraf.com Deployed `f166b075` with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

gmackie added 1 commit

2026-08-25 06:15:05 +00:00

ci(release-agent): git --no-pager — the checkout step blocked on a pager until timeout

CI / gitleaks (pull_request) Has been cancelled

Details

CI / storybook (pull_request) Has been cancelled

Details

CI / ci (pull_request) Has been cancelled

Details

02fadfd0cd

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

gmackie commented

2026-08-25 06:18:28 +00:00

Author

Owner

Preview environment is live: https://pr-440-forgegraph.forgegraf.com

Deployed 02fadfd0 with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

Preview environment is live: https://pr-440-forgegraph.forgegraf.com Deployed `02fadfd0` with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

gmackie added 1 commit

2026-08-25 06:19:20 +00:00

ci(release-agent): export HOME before set -u — unset in runner host jobs

CI / gitleaks (pull_request) Has been cancelled

Details

CI / storybook (pull_request) Has been cancelled

Details

CI / ci (pull_request) Has been cancelled

Details

e2cacc04cf

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

gmackie commented

2026-08-25 06:21:58 +00:00

Author

Owner

Preview environment is live: https://pr-440-forgegraph.forgegraf.com

Deployed e2cacc04 with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

Preview environment is live: https://pr-440-forgegraph.forgegraf.com Deployed `e2cacc04` with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

gmackie commented

2026-08-25 06:21:58 +00:00

Author

Owner

Preview environment is live: https://pr-440-forgegraph.forgegraf.com

Deployed e2cacc04 with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

Preview environment is live: https://pr-440-forgegraph.forgegraf.com Deployed `e2cacc04` with the beta stage's environment. It redeploys on every push and is destroyed when this PR closes.

gmackie added 1 commit

2026-08-25 06:34:16 +00:00

ci(release-agent): tag via git push with the CI token — the tags API returns 401 for it

CI / gitleaks (pull_request) Successful in 7s

Details

CI / storybook (pull_request) Successful in 1m38s

Details

forgegraph/ci CI passed

CI / ci (pull_request) Successful in 9m42s

Details

425448ef2b

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

gmackie scheduled this pull request to auto merge when all checks succeed

2026-08-25 07:10:47 +00:00

gmackie commented

2026-08-26 18:28:00 +00:00

Author

Owner

Tried to rebase this onto current main to clear the conflicts. Two of the three resolved cleanly:

f166b075 (90-min timeout) — main had gained GIT_PAGER/PAGER env on the same block. Kept both: the 90-minute timeout with its comment, plus main's env.
02fadfd0 (--no-pager) — main already has the fix by another route; only the explanatory comment was new, so I kept the comment.

425448ef (tag via git push) needs you. It conflicts with a competing fix for the same bug that landed on main separately:

	approach
main	try `GIT_API_TOKEN` against the tags REST API, retry with the run's repo-scoped token on any failure, make the degraded path loud
this branch	abandon the REST endpoint entirely, `git push` the tag over https — "the tags REST endpoint answers 401 for this token (run 14331), git over https does not"

If main's retry actually works, this commit is obsolete. If the REST endpoint 401s for this token regardless of retry, this commit is the real fix and main's is dead code. Deciding needs a real release run, so I left the branch untouched rather than guess in release infrastructure.

Tried to rebase this onto current main to clear the conflicts. Two of the three resolved cleanly: - **`f166b075` (90-min timeout)** — main had gained `GIT_PAGER`/`PAGER` env on the same block. Kept both: the 90-minute timeout with its comment, plus main's env. - **`02fadfd0` (`--no-pager`)** — main already has the fix by another route; only the explanatory comment was new, so I kept the comment. **`425448ef` (tag via git push) needs you.** It conflicts with a *competing fix for the same bug* that landed on main separately: | | approach | |---|---| | main | try `GIT_API_TOKEN` against the tags REST API, retry with the run's repo-scoped token on any failure, make the degraded path loud | | this branch | abandon the REST endpoint entirely, `git push` the tag over https — "the tags REST endpoint answers 401 for this token (run 14331), git over https does not" | If main's retry actually works, this commit is obsolete. If the REST endpoint 401s for this token regardless of retry, this commit is the real fix and main's is dead code. Deciding needs a real release run, so I left the branch untouched rather than guess in release infrastructure.

gmackie force-pushed release-dispatch/0.1.57 from 425448ef2b

CI / gitleaks (pull_request) Successful in 7s

Details

CI / storybook (pull_request) Successful in 1m38s

Details

forgegraph/ci CI passed

CI / ci (pull_request) Successful in 9m42s

Details

to 42e3cacb68

CI / gitleaks (pull_request) Successful in 7s

Details

CI / storybook (pull_request) Successful in 1m45s