Releases: register new releases at 0% rollout

[adamshiervani]

A newly registered release went to 10% of devices at once. It now starts at 0%: registered, visible in the table, served to nobody until an operator raises it. A 0% row is also ignored when picking the latest release, so registering one does not interrupt a rollout in progress.

 sync (tick, POST /releases/sync, interactive script default)
   new stable version in R2
-    → Release row @10%     # 10% of devices update on their next check
+    → Release row @0%      # no device updates until the rollout is raised

 Retrieve(deviceId, sku)
-  latest   = newest version
+  latest   = newest version with percentage > 0
   fallback = newest version at 100%
   bucket < latest.percentage → latest, else fallback

0.5.3 @100   0.5.4 @50   0.5.5 @0 (just registered)

              before        after
bucket 8      0.5.3         0.5.4      ← 0.5.5 shadowed the running rollout
bucket 95     0.5.3         0.5.3

Version-constrained requests, which skip rollout, still see 0% rows.

 src/release-sync.ts        DEFAULT_ROLLOUT_PERCENTAGE 10 → 0
 src/releases.ts            getLatestRelease filters rolloutPercentage > 0
 scripts/sync-releases.ts   prompt default follows the constant
 test/                      expectations 10 → 0; one test for the shadowing case

Code only, no migration. The schema default stays at 10; every insert sets the value.

---

Note

Medium Risk
Changes default OTA behavior for newly registered releases and how devices pick the “latest” rollout target; incorrect rollout selection could affect firmware delivery until an operator raises percentages.

Overview
Newly synced releases start at 0% rollout instead of 10%, so registering a version no longer pushes it to any devices until an operator raises the percentage.

Retrieve (background OTA checks) now treats “latest” as the newest release with rolloutPercentage > 0, so a 0% row stays visible in the DB but does not replace an older release that is still rolling out. Version-pinned requests are unchanged and can still resolve 0% rows.

Tests expect 0% on sync-created rows and add coverage for the case where a newer 0% release must not shadow an in-progress rollout.

^{Reviewed by Cursor Bugbot for commit 2c64c11. Bugbot is set up for automated code reviews on this repo. Configure here.}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-21T13:35:30.280521Z	2c64c11	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 2235e014d5

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Keep zero-percent releases from shadowing active rollouts

When an older release is still rolling out (for example, v2 at 50%), syncing v3 with this new 0% default makes v3 the result of getLatestRelease; Retrieve rejects it for every device and getDefaultRelease only considers 100% rows, so v2 immediately stops being offered and clients fall back to v1. Registering an inactive release therefore interrupts any in-progress rollout; release selection should ignore 0% releases or retain the previous staged candidate until the new release is raised.

Useful? React with 👍 / 👎.

[adamshiervani]

Fixed in 2c64c11: getLatestRelease ignores rows at 0%, so a registered but unreleased version does not displace an older release mid-rollout. Test added: 2.0.0 @100, 2.1.0 @50, 2.2.0 @0; bucket 8 still gets 2.1.0, bucket 95 gets 2.0.0.

[adamshiervani]

@codex review

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Keep it up!

Reviewed commit: 2c64c11f7d

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[Copilot]

Copilot review overview

🟢 Approval recommended

The behavior matches the stated rollout policy and has focused regression coverage.

Review effort: Balanced
Findings: None

What changed in this PR

Registers new releases at 0% rollout so they remain inactive until explicitly promoted.

Changes:

• Default new releases to 0%.
• Exclude 0% releases from latest-rollout selection.
• Add focused regression coverage.

File	Description
test/​sync-releases.test.ts	Updates synchronization expectations to 0%.
test/​releases.test.ts	Tests that 0% releases do not shadow active rollouts.
src/​releases.ts	Filters inactive releases from latest selection.
src/​release-sync.ts	Changes the default rollout percentage to 0%.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.