feat(scanner): background jobs declare acting identity for RLS

Phase 4 of the RLS plan: detached contexts (context.Background()) that
reach media_items now carry identities — with the policies live, a bare
context made governed reads raise (fresh connections: 42704 unrecognized
parameter; reused pool connections: 22P02 empty-string cast after the
LOCAL GUC reverted — the plan's A2 signature).

- scanner jobs (API-triggered, mtime-poll, initial scan): the job's
  Context now carries the acting admin's identity — the library owner,
  first-admin fallback (the job's UserID already encoded this trust
  decision; the Context just needed to carry it). performInitialScan
  uses the scanner's configured admin.
- websocket initial-sync read (GetUserMediaItemsForSync): runs as the
  connection's own user.
- hash backfill: runs as first admin (server self-heal over all media).

Verified on the dev stack: forced rescan of the 1199-file Audiobooks
library completed with 0 errors under RLS (previously every SHA lookup
and create raised); hash-backfill refusals gone. The sync queue
processor needs no identity — sync_queue is not an RLS-governed table.
Known remaining detached reads (conflicts/analytics handlers using
context.Background()) are HTTP-path triage, enumerated by the full
suite next.
This commit is contained in:
John O'Keefe committed 2026-10-29 09:08:54 -04:00
1 parent 922b92a8b6
commit 385a5d9dba
4 files changed
+36 -8

No files matched your search

+8 -1
View File
@@ -8,6 +8,8 @@ import (
"net/http" "net/http"
"github.com/google/uuid" "github.com/google/uuid"
"bookhoard/internal/rls"
"github.com/jackc/pgx/v5/pgtype" "github.com/jackc/pgx/v5/pgtype"
"github.com/labstack/echo/v5" "github.com/labstack/echo/v5"
) )
@@ -127,6 +129,11 @@ func (h *Handler) ScanLibrary(c *echo.Context) error {
jobID := uuid.New().String() jobID := uuid.New().String()
// RLS plan §6: the scan job's media reads run under the triggering
// admin's identity. h.ctx is a handler-lifetime Background context
// (outlives the request), so the identity is stamped explicitly —
// a bare context makes governed reads raise (placeholder reverting
// to '' on reused connections, or 42704 on fresh ones).
job := &services.Job{ job := &services.Job{
ID: jobID, ID: jobID,
Type: services.JobTypeScan, Type: services.JobTypeScan,
@@ -140,7 +147,7 @@ func (h *Handler) ScanLibrary(c *echo.Context) error {
"trigger": "admin", "trigger": "admin",
}, },
Status: services.JobStatusPending, Status: services.JobStatusPending,
Context: h.ctx, Context: rls.WithUser(h.ctx, pgtype.UUID{Bytes: userUUID, Valid: true}),
} }
if err := h.worker.EnqueueJob(job); err != nil { if err := h.worker.EnqueueJob(job); err != nil {
+6 -1
View File
@@ -11,6 +11,8 @@ import (
jwt "github.com/golang-jwt/jwt/v5" jwt "github.com/golang-jwt/jwt/v5"
"github.com/google/uuid" "github.com/google/uuid"
"bookhoard/internal/rls"
"github.com/gorilla/websocket" "github.com/gorilla/websocket"
"github.com/jackc/pgx/v5/pgtype" "github.com/jackc/pgx/v5/pgtype"
"github.com/labstack/echo/v5" "github.com/labstack/echo/v5"
@@ -201,7 +203,10 @@ func (h *WSHandler) getInitialState(userID string) map[string]interface{} {
pgUserID := pgtype.UUID{Bytes: userUUID, Valid: true} pgUserID := pgtype.UUID{Bytes: userUUID, Valid: true}
mediaItems, err := h.db.GetUserMediaItemsForSync(context.Background(), pgUserID) // RLS plan §6: this read is the connection's OWN user's media —
// declare that identity explicitly (context.Background() carries
// nothing through ScopedDB, and media_items is RLS-governed).
mediaItems, err := h.db.GetUserMediaItemsForSync(rls.WithUser(context.Background(), pgUserID), pgUserID)
if err != nil { if err != nil {
return map[string]interface{}{"error": "failed to fetch initial state"} return map[string]interface{}{"error": "failed to fetch initial state"}
} }
+9 -1
View File
@@ -427,7 +427,15 @@ func RegisterRoutes(cfg *Config) *handlers.Handler {
// startup scans for disk I/O. // startup scans for disk I/O.
go func() { go func() {
time.Sleep(30 * time.Second) time.Sleep(30 * time.Second)
services.NewHashBackfillService(cfg.Queries).Run(context.Background()) // RLS plan §6: this self-heal sweeps ALL media — run it as the
// first admin (the scanner's documented fallback identity);
// a bare Background context makes governed reads raise.
adminID, err := cfg.Queries.GetFirstAdmin(context.Background())
if err != nil {
log.Printf("[HASH-BACKFILL] no admin found, skipping backfill: %v", err)
return
}
services.NewHashBackfillService(cfg.Queries).Run(rls.WithUser(context.Background(), adminID))
}() }()
// Register progress routes with actual handler // Register progress routes with actual handler
+13 -5
View File
@@ -43,6 +43,8 @@ import (
epub "github.com/ArcadiaLin/go-epub" epub "github.com/ArcadiaLin/go-epub"
"github.com/fsnotify/fsnotify" "github.com/fsnotify/fsnotify"
"github.com/google/uuid" "github.com/google/uuid"
"bookhoard/internal/rls"
"github.com/jackc/pgx/v5" "github.com/jackc/pgx/v5"
"github.com/jackc/pgx/v5/pgtype" "github.com/jackc/pgx/v5/pgtype"
"github.com/nwaples/rardecode" "github.com/nwaples/rardecode"
@@ -621,17 +623,23 @@ func (s *MediaScanner) enqueueLibraryScan(rootFolder, trigger string) {
folderPaths[i] = f.FolderPath folderPaths[i] = f.FolderPath
} }
// The acting admin: the library owner, or the first admin as the
// documented fallback (existing pattern). Used BOTH for the job's
// UserID and — since RLS (plan §6) — as the identity the job's
// media reads run under via its Context.
adminIDStr := "" adminIDStr := ""
if libRow.CreatedByAdminID.Valid { adminUUID := libRow.CreatedByAdminID
adminIDStr = uuid.UUID(libRow.CreatedByAdminID.Bytes).String() if adminUUID.Valid {
adminIDStr = uuid.UUID(adminUUID.Bytes).String()
} }
if adminIDStr == "" { if !adminUUID.Valid {
fmt.Printf("[MTIME-POLL] Library has no owner, falling back to first admin\n") fmt.Printf("[MTIME-POLL] Library has no owner, falling back to first admin\n")
fallbackAdmin, err := s.db.GetFirstAdmin(context.Background()) fallbackAdmin, err := s.db.GetFirstAdmin(context.Background())
if err != nil { if err != nil {
fmt.Printf("[MTIME-POLL] Warning: no admin found in database, skipping scan\n") fmt.Printf("[MTIME-POLL] Warning: no admin found in database, skipping scan\n")
return return
} }
adminUUID = fallbackAdmin
adminIDStr = uuid.UUID(fallbackAdmin.Bytes).String() adminIDStr = uuid.UUID(fallbackAdmin.Bytes).String()
} }
@@ -650,7 +658,7 @@ func (s *MediaScanner) enqueueLibraryScan(rootFolder, trigger string) {
Type: JobTypeScan, Type: JobTypeScan,
Status: JobStatusPending, Status: JobStatusPending,
UserID: adminIDStr, UserID: adminIDStr,
Context: context.Background(), Context: rls.WithUser(context.Background(), adminUUID),
Params: map[string]any{ Params: map[string]any{
"library_id": libraryIDStr, "library_id": libraryIDStr,
"folders": folderPaths, "folders": folderPaths,
@@ -4771,7 +4779,7 @@ func (s *MediaScanner) performInitialScan(ctx context.Context) {
Type: JobTypeScan, Type: JobTypeScan,
Status: JobStatusPending, Status: JobStatusPending,
UserID: adminIDStr, UserID: adminIDStr,
Context: context.Background(), Context: rls.WithUser(ctx, s.adminID), // RLS plan §6: startup scans run as the configured admin
Params: map[string]any{ Params: map[string]any{
"library_id": libraryIDStr, "library_id": libraryIDStr,
"folders": folderPaths, "folders": folderPaths,