Release runbook documents first-RC frozen-source behavior that isn't on main
Maintainers usually reply within 1 day
Nobody has claimed this yet.
Assessment
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Newbie friendliness
- 74/100
- Issue type
- Documentation
- Clarity
- Clearly specified
- Activity status
- Active
- Tech stack
- bash, github-actions, go
- Domain
- documentation, release
Research direction
Read docs/contributing/contributing-releases/release-runbook.md alongside .github/workflows/prepare-release.yaml and .github/scripts/prepare-release.sh on the current main branch. Check the release controller validation and sibling commit capture behavior referenced in the issue, then update the runbook to match the current workflow. Done when it no longer documents first-RC frozen-source behavior that is absent from main; recheck whether #13410 has merged before editing.
Written by the indexing model from the issue text.
Description
Bug information
Steps to reproduce
- Read
docs/contributing/contributing-releases/release-runbook.mdonmainat cd1225a1365119476385b52ef5e97c89153c228b:- Line 35: "To select an earlier reachable cutoff, add
-f product-commit="<full-40-character-SHA>"; preparation verifies that its release metadata and build tooling are compatible." - Line 39: "While the first-RC PR remains open, rerunning with the same type/channel retains its cutoff, sibling commits, release date, and curated Highlights/Upgrading. Normally omit
product-commiton reruns; a different cutoff for the same open preparation is rejected."
- Line 35: "To select an earlier reachable cutoff, add
- Compare with
.github/workflows/prepare-release.yamland.github/scripts/prepare-release.shon the same commit.
Observed behavior
The runbook describes behavior that main doesn't have:
- No
product-commitinput.workflow_dispatchinprepare-release.yamlhas onlyrelease-type,channel, andbackport-pr-numbers.gh workflow run ... -f product-commit=...fails because the input is unknown. The input has never existed in this workflow onmain. - Each rerun recaptures everything except curated notes:
- Cutoff: the workflow checks out
main, andprepare-release.shlines 703-704 setproductCommitto the currentHEAD. Ifmainmoved, the rerun freezes a newer commit. - Sibling commits: the "Capture sibling repository commits" step runs
capture-release-sibling-commits.shon every run, which takes the current head of each sibling'srelease/X.Y, ormainif there's none. - Release date:
prepare-release.shline 43 defaultsRELEASE_DATEto today, and the workflow doesn't pass one. - Only Highlights and Upgrading are kept, through "Read curated notes from existing release pull request" and
preserve-release-note-sections.sh.
- Cutoff: the workflow checks out
Other runbook lines describe the same missing first-RC frozen-source flow:
- Line 57 says the controller creates a signed release-App commit on top of the frozen cutoff. On
main,validate-release-controller-plan.shinstead requires a new-channel release commit to be the release PR's squash commit ("new-channel release commit must be the release PR commit"). - Line 107 and line 109 refer to a "frozen first RC", a "resolved release commit", and "frozen-source support".
The runbook came from #13390. These lines match draft #13410 ("fix(release): freeze first-RC sources through publication"), which isn't merged.
Desired behavior
The runbook describes what main does: a first-RC rerun freezes the current main head, recaptures sibling commits, and uses that day's date. Only Highlights and Upgrading carry over. There is no product-commit input. If #13410 merges first, the lines become correct and this can close.
Workaround
Follow prepare-release.yaml and prepare-release.sh on main rather than the runbook lines above. Don't pass product-commit.
System information
rad Version
N/A (release tooling docs)
Operating system
N/A
Additional context
#13697 documents a bicep-types-aws pin fix that depends on reruns recapturing sibling commits, which is how main behaves today. If #13410 merges as drafted, first-RC reruns reuse their sibling commits, so that fix would need to start a new preparation or use product-commit to choose the cutoff before setting the pin.
- Dominant language
- Go
- Stars
- 1.7k
- Forks
- 142
- Avg merge
- 1d 11h
- Merged PRs (30d)
- 229
Getting set up
Starts the project's dev container in your browser, under your own GitHub account.
- No Dockerfile or Docker Compose file
- Has a pull request template
- Read the contributing guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from radius-project/radius
-
bug triaged
Difficulty 1/5 Under an hour Newbie friendliness 90/100
radius-project/radius#13585 · 2 comments ·
Maintainers usually reply within 1 day
-
bug triaged
Difficulty 2/5 1-3 hours Newbie friendliness 80/100
radius-project/radius#13566 · 2 comments ·
Maintainers usually reply within 1 day
-
Flaky test: start-git-http-port-forward_test.sh hang case reads loop.pid before it existsPossibly taken @kachawla claimed this 3 days ago. Openbug triaged
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
radius-project/radius#13560 · 2 comments · 1 assignee ·
Maintainers usually reply within 1 day
-
CLI hides Kubernetes API errors behind empty ARM error detailsPossibly taken @sk593 claimed this 3 days ago. Openbug good first issue help wanted triaged
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
radius-project/radius#13491 · 4 comments · 1 assignee ·
Maintainers usually reply within 1 day
-
bug github-copilot triaged
Difficulty 2/5 1-3 hours Newbie friendliness 74/100
radius-project/radius#12851 · 2 comments ·
Maintainers usually reply within 1 day
All issues in radius-project/radius
Similar issues
-
agent-butler-finding bug
Difficulty 2/5 1-3 hours Newbie friendliness 82/100
jordansmall/spindrift#6414 ·
Maintainers usually reply within 1 day
-
Difficulty 2/5 1-3 hours Newbie friendliness 68/100
Icinga/icinga-kubernetes#294 ·
Maintainers usually reply within 4 days
-
Cluster details: the Control Plane version field has a different border color than its icon boxOpenkind/design sig/ui
Difficulty 1/5 Under an hour Newbie friendliness 92/100
kubermatic/dashboard#8449 ·
Maintainers usually reply within 1 day
-
needs-triage
Difficulty 2/5 1-3 hours Newbie friendliness 62/100
cli/cli#14962 · 1 comment · 1 reaction ·
Maintainers usually reply within 1 day
-
model-drift
Difficulty 2/5 1-3 hours Newbie friendliness 66/100
Maintainers usually reply within 1 day