Cleanup: identify and fix users affected by the id_token email-claim regression
Les mainteneurs répondent en général sous 1 jour
Personne n'a encore pris cette issue.
Évaluation
- Difficulté
- 4/5
- Temps estimé
- 3-5 jours
- Accessibilité débutants
- 30/100
- Type d'issue
- Bug
- Clarté
- Clairement spécifiée
- Activité
- Active
- Domaine
- authentication, database
Piste de recherche
Start by reading the issue's described fallback in lib/omniauth/strategies/codebar.rb and the listed cleanup queries; run the scans only after codebar/auth#83 is deployed. Keep emails and auth IDs in private ops notes, reporting counts and member IDs in the issue. Done means affected records are safely merged or re-keyed, members 31450 and 31452 are gone, no matching garbage records remain, and a magic-link sign-in resolves to the correct member.
Rédigé par le modèle d'indexation à partir du texte de l'issue.
Description
The fix in codebar/auth#87 (PR codebar/auth#83) makes every id_token carry the user-record email and name claims. After it merges and deploys, the data damage from the regression still needs cleanup.
While the regression was live (auth release v64, 2026-10-02 10:01-16:34 UTC), a magic-link sign-in produced an id_token without an email claim. The codebar OmniAuth strategy fell back to sub (email = payload['email'] || payload['sub'] in lib/omniauth/strategies/codebar.rb) and created a planner member whose email and name were the better-auth user id: an empty account with no subscriptions or roles. Two such members are known: 31450 and 31452. Their users were dropped into the new-member flow as if they had never signed up. The rollback (v65) restored correct behaviour immediately: member 17284 has confirmed that sign-in works normally again. The window is therefore closed at 2026-10-02 16:34 UTC, and what remains is data cleanup.
1. Identify any affected users
- Find planner members whose
emailis a better-auth user id: members created in the window whoseemailmatches a 32-character opaque id, or whoseemailequals an auth user id directly. - Cross-check
auth_servicerows with providercodebarcreated in the window: a row whoseuidhas that same 32-character shape points at a garbage member. detect_duplicate_membersdoes not find these: the garbage member's email and name share nothing with the real member. Use the queries above, not the duplicate detector.- Keep the resulting user list out of this issue (emails and ids go in the private ops notes); report only counts and member ids here. Expected result: exactly 31450 and 31452 — earlier dump scans found no members of this shape before v64. The scan's real job is to confirm that and to catch anything the users created while stuck in the new-member flow.
2. Fix the identified users
For each affected user:
- If a real member exists (the usual case): merge anything the garbage member picked up (the user may have completed the new-member flow) into the real member, then delete the garbage member and its
auth_servicerow. 17284 is confirmed already: their real member is intact and in use, so only the garbage member (31452) goes. - If no real member exists (a genuinely new user who signed up during the window): re-key the member to the user's real email and name instead of deleting it.
- Confirm the user's next magic-link sign-in resolves to the same member. GitHub sign-ins already resolve via
github_idand are unaffected.
Verification before closing
- One magic-link sign-in resolves the real member on the fixed build (1.7.7 + explicit claims — different code path from the rollback that 17284 confirmed; member 17284 is the known case).
- No member has an email equal to a better-auth user id, and no
auth_servicerow has acodebaruid of that shape. - Members 31450 and 31452 are gone.
Timing: run this after codebar/auth#83 is deployed. Sign-ins after the fix can no longer create members of this shape.
- Langage dominant
- Ruby
- Étoiles
- 105
- Forks
- 206
- Merge moyen
- 1 j 3 h
- PR mergées (30 j)
- 74
Préparer son environnement
- Fournit un Dockerfile ou un fichier Docker Compose
- Aucun modèle de pull request
- Lire le guide de contribution
Par où commencer
- Lisez l'issue en entier, puis le guide de contribution du projet.
- Signalez en commentaire que vous la prenez — cela évite que deux personnes fassent le même travail.
- Forkez le dépôt et travaillez sur une branche.
- Ouvrez une pull request qui référence le numéro de l'issue.
Autres issues de codebar/planner
-
DashboardQuery eager-loads workshop_host with :sponsors in one join (same host-loss trap as #2975)Ouverte
Difficulté 2/5 1-3 heures Accessibilité débutants 78/100
Les mainteneurs répondent en général sous 1 jour
-
enhancement
Difficulté 2/5 1-3 heures Accessibilité débutants 76/100
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 2/5 1-3 heures Accessibilité débutants 68/100
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 3/5 1-2 jours Accessibilité débutants 45/100
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 5/5 Plus d'une semaine Accessibilité débutants 42/100
Les mainteneurs répondent en général sous 1 jour
Toutes les issues de codebar/planner
Issues similaires
-
[bug]: Sticky Streak DisappearedOuvertebug
Difficulté 2/5 1-3 heures Accessibilité débutants 62/100
Les mainteneurs répondent en général sous 1 jour
-
Použiť Redis pre ActionCableOuverte
Difficulté 2/5 1-3 heures Accessibilité débutants 65/100
slovensko-digital/autogram-portal#392 ·
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 2/5 1-3 heures Accessibilité débutants 65/100
voxpupuli/puppet-quadlets#122 · 5 commentaires ·
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 1/5 Moins d'une heure Accessibilité débutants 82/100
TheOdinProject/curriculum#31452 ·
Les mainteneurs répondent en général sous 1 jour
-
Difficulté 2/5 1-3 heures Accessibilité débutants 70/100