In Play Console, closed testers keep getting the production build instead of the new beta; how do version codes decide which track's bundle a device receives?
answer
- highest number wins
- everyone is eligible for production
- listed and opted in
- superseded, shadowed, promoted
basics
~20 sA device gets the highest compatible version code from the tracks its user is eligible for, and everyone is eligible for production. If production's version code beats the beta's, closed testers get production and the beta track is superseded.
solid answer
~50 sPlay does not pick by track: a device receives the bundle with the highest compatible `versionCode` among the tracks its user is eligible for. Everyone is eligible for production; a closed or open tester is also eligible for that test track once listed and opted in; an internal tester is not eligible for closed or open tracks and receives the internal track's build. So when a hotfix gives production a higher version code than the beta, closed testers fall back to production. Play Console flags this in the rollout review: **Superseded** means the track serves nobody, **Partially shadowed** means some testers get each build and usually signals a version code mistake, and **Promoted** means the test bundles are live in production too. The fix is to rebuild the beta above production and adopt a scheme that keeps the next release line above any hotfix.
go deeper
Remember that Google Play compares version codes, the integer, and that everyone is eligible for the production track.
Explain eligibility per track, why internal testers are a special case, and how the highest compatible version code is chosen.
Diagnose the superseded beta from the rollout review's fallback statuses and fix it with a rebuilt version code and a numbering scheme that survives hotfixes.
Own the version code scheme across release lines, hotfixes and tracks so no team can strand testers on production by accident.
## The rule Play applies Google Play does not pick a build by track name. A device automatically receives the app version that meets **both** conditions: 1. It has the **highest version code** compatible with the device. 2. It is published to a track the user is **eligible** to receive. The **version code** is the integer `versionCode` of the bundle, not the human-readable version name. A string like "4.2-beta" plays no part in the choice. Compatibility matters too: only bundles that support a device's configuration compete for it, so one track can win on some devices and lose on others. ## Who is eligible for which track Everyone is eligible for **production**. To be eligible for a test track, a user must be included in that track's tester configuration **and** have opted in. | User | Eligible tracks | |---|---| | Not a tester | Production | | Open tester | Production, open testing | | Closed tester | Production, closed testing | | Open and closed tester | Production, open testing, closed testing | | Internal tester | The internal testing build; never closed or open, even when listed there | Opting in is per user and per test: someone listed on the closed track who never opened its opt-in link is eligible for production only. So if production carries a **higher** version code than the closed track, closed testers receive the production build. The beta still exists, but nobody on that track gets it. ## Reading the fallback statuses Each test track has a **fallback track**, another track its users also receive from, such as production for an open test. When you roll out, Play Console shows validation messages about that fallback: - **Shadowed**: one bundle serves some or all of another bundle's device configurations with a higher version code. - **Promoted**: all active bundles on the track are also active on the fallback track, for example the open-testing bundle is now live in production. This is the healthy end state. - **Superseded**: every active bundle on the track is fully shadowed by higher version codes on the fallback track, so the track serves **no one**. - **Partially shadowed**: at least one bundle is shadowed, so some testers get the test build and others get production. Play Console says this usually indicates an error in version code assignment. ## How collisions happen - A **hotfix** goes to production from a release branch with a version code above the beta that is still in closed testing. - The beta is built from an older branch, or a CI counter is reset, so its version code is lower than production's. - A team assumes a test track always wins and does not notice the warnings in the rollout review. - Several bundles target different device configurations, and only some get bumped, leaving the track partially shadowed. Every bundle that goes into a release also needs a **version code that has not been uploaded before**. The exception is **internal app sharing**, where version codes may repeat, which is one reason throwaway builds belong there and not on a track. ## Fixing and preventing it 1. Read the rollout review's validation messages before you start a test rollout; a **Superseded** or **Partially shadowed** status is the warning. 2. Rebuild the beta with a version code **above** the current production release and roll it out on the test track. 3. Adopt a version code scheme where the next release line always numbers above any hotfix of the current one, for example by reserving room per release. 4. When a test build is good, add **the same bundle** from the library to the production release instead of rebuilding it, so the track ends as **Promoted**. 5. If a long beta must survive several production hotfixes, give every new beta build a version code above the latest production build until it is promoted. ## Worked example The fintech app's closed track carries 4.2 at version code 4200. Production gets a security hotfix, 4.1.1, built with version code 4210 because the release branch's counter ran ahead. Closed testers immediately receive 4.1.1 from production and the beta track shows as superseded. The team rebuilds 4.2 as 4220, rolls it out on the closed track, and changes the scheme so release lines number in blocks of 100.
- What does a Partially shadowed status on the open testing track mean in practice?At least one active open-testing bundle is outranked by a higher version code in production for some device configurations. Some open testers get the test build and others get production, so test results mix two versions. Play Console notes that this usually indicates an error in version code assignment.
- Why add the tested bundle to the production release instead of rebuilding it for production?Adding the same bundle from the library ships exactly what testers validated, keeps one version code across tracks and leaves the test track Promoted. A rebuild is a different artifact that needs a new version code, so production would carry something nobody tested.
- Testers always get the test track's build, whatever production's version code is.
- Play compares versionName strings to decide which build is newer.
- An internal tester also receives closed builds when listed on that track.
- A superseded track still serves its bundle to some of its testers.
- Any version code can be reused as long as it goes to a different track.