π End-to-end encrypted cloud for everything.
Dart 2faandroidauthye2eeencryptionend-to-end-encryptionfluttergolanggoogle-photosgoogle-photos-alternativeiosopensourcephotosprivacyprivacy-toolstypescriptzero-knowledge
Remove file count rollout document
Neeraj Guptac85b7d30d6d3
diff --git a/server/docs/file-count-initialization.md b/server/docs/file-count-initialization.mddeleted file mode 100644index 2f76b02e82..0000000000--- a/server/docs/file-count-initialization.md+++ /dev/null@@ -1,63 +0,0 @@-# Initialize existing users' file counts--`POST /admin/user/init-file-counts` initializes one user's paired-NULL counters.-It uses the existing admin authentication and the server's primary database.-Nothing runs automatically on deployment; client count reads remain source-based.--`GET /admin/user?id=...` (or `?email=...`) also returns top-level-`photosFileCount` and `lockerFileCount` from `usage`. Both are `-1` until the pair-is initialized, including when the usage row is missing; an initialized empty-library returns `0`. These are individual-user counts, not family totals, and-reading them does not trigger initialization. Existing response fields are unchanged.--Call only after **all count-aware writers are deployed and old processes have-stopped**, including the stale-membership invalidation path. For the first cohort,-submit users with `user_id < 200` one at a time, pausing between requests. The API-accepts an explicit user ID, not a range, so cohort selection and pacing stay with-the caller. A small ID range does not imply small libraries.--Request body:--```json-{"userID": 1}-```--- HTTP 200 with `{"initialized": true}`: both counters were published.-- HTTP 200 with `{"initialized": false}`: already initialized or the source version- changed; retry later if the counters are still NULL.-- HTTP 200 with `initialized: false` and `reason`: inconsistent history; leave the- counters NULL and investigate. Repeating the request does not repair the data.-- HTTP 504: the database work exceeded its 10-second budget. Stop or defer large- accounts if DB load or latency rises. Rerunning a request is safe, including- when a response is lost after publication.-- Other errors use the normal API error handling. A missing usage row returns- HTTP 404 and is never created by this endpoint.--Search server logs for `file count initialization`; each attempt records the-admin, target user, result, and any error. There is no new metric or per-user-metric label.--Each attempt reads the version, distinct active owned-file counts, and-eligibility in one SQL snapshot. It then publishes both counters only if they-are still NULL and the version is unchanged. Counting takes no file or membership-row locks; only publication briefly locks that user's usage row. Count-changing-writers that commit before publication cause this attempt to be skipped; writers-that follow it maintain the initialized counts. Storage and source version are-unchanged by publication.--Unsafe histories stay NULL:--- An owned file has no active owned membership and no unrestored Trash row.- Trash was added later: older cleanup used missing collection references to- delete objects directly, so missing references do not establish Trash state.-- An unrestored Trash row conflicts with active owned membership or file ownership.-- An owned file has active cross-app or unsupported-app memberships.-- An active Locker membership's actual file ownership disagrees with the legacy- `f_owner_id` predicate. NULL denormalized owners on old Photos rows are allowed.--Do not reset source versions or write zero counts to bypass a failed check.-Before expanding the cohort or switching reads, check the existing-`museum_file_count_comparisons_total` match/mismatch metric and `file count mismatch`-logs during real source reads and file activity. No observations is not successful-validation. The trusted-client same-app assumption still applies; this endpoint-does not add guards to AddFiles/MoveFiles.