fix(reaper): 认领时重置活跃时钟与预警，回收只复用本次认领后的 reaper 归档

docs/troubleshooting.md

@@ -13,6 +13,8 @@ graded for how far the in-repo Go test suite proves the behaviour:
 
 - **[GO-TESTED]** — a hermetic `*_test.go` exercises this exact path; the
   string/code is asserted in CI.
+- **[PG-TESTED]** — a `-tags pgint` test in `internal/pgint` drives it against a
+  real Postgres with the shipped migrations (CONTRIBUTING.md has the command).
 - **[CODE-ONLY]** — the code path and string exist and are real, but no unit
   test drives them (notably the Velocity Java plugin, which is not compiled or
   tested in this repo).
@@ -697,6 +699,11 @@ SMTP sink.]
   the activity clock never resets** and an actively-played world becomes
   reap-eligible after 15 days. Verify join events are flowing (§3a) — this is the
   most important reaper check. [INTEGRATION-ONLY for the live Postgres write.]
+  A claim also resets it: `ClaimServer` sets `last_active_at` to the claim and
+  clears `warned_*`, so a world reaped long ago and claimed again gets a full 15
+  days, and a later reap archives the new owner's world rather than reusing the
+  previous owner's archive (`FreshBackup` counts only `inactive_15d` archives
+  taken since the claim). [PG-TESTED `TestReclaimRestartsReaperClock`.]
 - **Unowned servers are still reaped.** A server with `owner_id=""` gets **no
   pre-deletion warning** (`maybeWarn` skips unowned), but is still reaped at 15d.
   [GO-TESTED `TestReapUnownedServerStillReaped`.] Claim or exempt servers you

internal/api/pgrepo.go

@@ -417,6 +417,12 @@ func (p *PGRepo) ServerResources(ctx context.Context, name string) (ResourceSpec
 // user for two DIFFERENT ownerless servers serialize instead of both passing the
 // gate (audit #4); the row is additionally taken FOR UPDATE so concurrent claims
 // of the SAME server still resolve to exactly one winner.
+//
+// A claim starts the reaper's clock afresh: last_active_at moves to the claim
+// and the pre-reap warnings clear. A world reaped before keeps its release time
+// as last_active_at, so without the reset a new owner who configures it from
+// the panel before anyone joins would lose it on the next reaper run, with no
+// warning and no archive of their own.
 func (p *PGRepo) ClaimServer(ctx context.Context, name, userID string) (bool, error) {
 	tx, err := p.db.BeginTx(ctx, nil)
 	if err != nil {
@@ -470,7 +476,8 @@ func (p *PGRepo) ClaimServer(ctx context.Context, name, userID string) (bool, er
 	}
 
 	res, err := tx.ExecContext(ctx,
-		`UPDATE servers SET owner_id = $2, claimed_at = now() WHERE name = $1 AND owner_id IS NULL AND deleted_at IS NULL`,
+		`UPDATE servers SET owner_id = $2, claimed_at = now(), last_active_at = now(), warned_3d_at = NULL, warned_1d_at = NULL
+		 WHERE name = $1 AND owner_id IS NULL AND deleted_at IS NULL`,
 		name, userID)
 	if err != nil {
 		return false, err

internal/api/repo.go

@@ -244,6 +244,8 @@ type Repo interface {
 	// authoritative gate: a claim that would exceed a cap → ErrQuotaExceeded (403),
 	// and two concurrent claims by one user cannot both pass (audit #4).
 	// QuotaCheck remains the advisory pre-check for the handler's fast-path 403.
+	// A successful claim resets last_active_at to now and clears warned_*, so the
+	// reaper counts idleness from the claim.
 	ClaimServer(ctx context.Context, name, userID string) (bool, error)
 	// UserInAllowlist reports whether the user's linked UUID is on the server
 	// allowlist (spec §9.4).

internal/pgint/reaper_test.go

@@ -0,0 +1,141 @@
+//go:build pgint
+
+package pgint
+
+import (
+	"context"
+	"database/sql"
+	"testing"
+	"time"
+
+	"felis.lolicon.best/internal/backup"
+	"felis.lolicon.best/internal/reaper"
+)
+
+// reclaimCluster knows one server; every other row in the shared schema reads
+// as a server whose CRD is gone, which the reaper skips.
+type reclaimCluster struct {
+	name       string
+	deletedPVC []string
+}
+
+func (c *reclaimCluster) Inspect(_ context.Context, name string) (reaper.ServerCRD, error) {
+	if name != c.name {
+		return reaper.ServerCRD{}, reaper.ErrNotFound
+	}
+	return reaper.ServerCRD{PVC: "world-" + name + "-0"}, nil
+}
+
+func (c *reclaimCluster) DeletePVC(_ context.Context, pvc string) error {
+	c.deletedPVC = append(c.deletedPVC, pvc)
+	return nil
+}
+
+func (c *reclaimCluster) Stop(context.Context, string) error { return nil }
+
+type reclaimArchiver struct{ archived []string }
+
+func (a *reclaimArchiver) Archive(_ context.Context, server, _ string) (backup.ArchiveRef, int64, error) {
+	ref := "/archives/" + server + "-new.tar.gz"
+	a.archived = append(a.archived, ref)
+	return backup.ArchiveRef(ref), 42, nil
+}
+
+func (a *reclaimArchiver) Restore(context.Context, backup.ArchiveRef, string) error { return nil }
+func (a *reclaimArchiver) Delete(context.Context, backup.ArchiveRef) error        { return nil }
+
+// TestReclaimRestartsReaperClock: a world reaped weeks ago and claimed by a new
+// owner is not reaped again on the next run, and when it does go idle the reap
+// archives the new owner's world instead of reusing the previous owner's
+// archive (data-durability-5).
+func TestReclaimRestartsReaperClock(t *testing.T) {
+	ctx := context.Background()
+	st := reaper.NewPGStore(db)
+	name := "reclaim-" + suffix(t)
+	if _, err := db.ExecContext(ctx,
+		`INSERT INTO servers (name, cached_cpu_milli, cached_memory_mb, cached_storage_mb) VALUES ($1, 100, 128, 1)`,
+		name); err != nil {
+		t.Fatalf("seed server: %v", err)
+	}
+	first := newUser(t, "user", "reclaim-a")
+	if ok, err := repo.ClaimServer(ctx, name, first.ID); err != nil || !ok {
+		t.Fatalf("first claim = (%v, %v)", ok, err)
+	}
+
+	// The first owner went idle and the reaper took the world 20 days ago.
+	reapedAt := time.Now().Add(-20 * reaper.Day)
+	if _, err := db.ExecContext(ctx,
+		`UPDATE servers SET last_active_at = $2, claimed_at = $3 WHERE name = $1`,
+		name, reapedAt.Add(-16*reaper.Day), reapedAt.Add(-60*reaper.Day)); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := db.ExecContext(ctx,
+		`INSERT INTO world_backups (id, server_name, former_owner, backup_ref, size_bytes, reason, status, created_at, expires_at)
+		 VALUES ($1, $2, $3, $4, 7, 'inactive_15d', 'present', $5, $6)`,
+		"bk-"+suffix(t), name, first.ID, "/archives/"+name+"-old.tar.gz", reapedAt, reapedAt.Add(90*reaper.Day)); err != nil {
+		t.Fatalf("seed the old archive: %v", err)
+	}
+	if err := st.ReleaseWorld(ctx, name, reapedAt); err != nil {
+		t.Fatalf("ReleaseWorld: %v", err)
+	}
+	// Leftover warning stamps must not survive into the next ownership.
+	if _, err := db.ExecContext(ctx,
+		`UPDATE servers SET warned_3d_at = $2, warned_1d_at = $2 WHERE name = $1`, name, reapedAt); err != nil {
+		t.Fatal(err)
+	}
+
+	second := newUser(t, "user", "reclaim-b")
+	before := time.Now()
+	if ok, err := repo.ClaimServer(ctx, name, second.ID); err != nil || !ok {
+		t.Fatalf("second claim = (%v, %v)", ok, err)
+	}
+	var lastActive time.Time
+	var w3, w1 sql.NullTime
+	if err := db.QueryRowContext(ctx,
+		`SELECT last_active_at, warned_3d_at, warned_1d_at FROM servers WHERE name = $1`, name).Scan(&lastActive, &w3, &w1); err != nil {
+		t.Fatal(err)
+	}
+	if lastActive.Before(before.Add(-time.Minute)) {
+		t.Fatalf("last_active_at = %v after the claim at %v: the claim did not restart the clock", lastActive, before)
+	}
+	if w3.Valid || w1.Valid {
+		t.Fatalf("warnings survived the claim: 3d=%v 1d=%v", w3, w1)
+	}
+	// The previous owner's archive is not this owner's world, whatever since says.
+	if f, ok, err := st.FreshBackup(ctx, name, reapedAt.Add(-30*reaper.Day)); err != nil || ok {
+		t.Fatalf("FreshBackup = (%+v, %v, %v): the old archive counted as the new world's", f, ok, err)
+	}
+
+	cl := &reclaimCluster{name: name}
+	ar := &reclaimArchiver{}
+	now := time.Now()
+	r := &reaper.Reaper{Cfg: reaper.DefaultConfig(), Store: st, Cluster: cl, Archiver: ar,
+		Now: func() time.Time { return now }}
+	if _, err := r.RunOnce(ctx); err != nil {
+		t.Fatalf("RunOnce: %v", err)
+	}
+	if len(cl.deletedPVC) != 0 || len(ar.archived) != 0 {
+		t.Fatalf("the reaper took a world claimed moments ago: deleted %v, archived %v", cl.deletedPVC, ar.archived)
+	}
+
+	// Sixteen idle days after the claim it goes, with a fresh archive.
+	now = time.Now().Add(16 * reaper.Day)
+	if _, err := r.RunOnce(ctx); err != nil {
+		t.Fatalf("RunOnce after 16 days: %v", err)
+	}
+	if len(ar.archived) != 1 || len(cl.deletedPVC) != 1 {
+		t.Fatalf("after 16 idle days: archived %v, deleted %v; want one new archive, then the delete", ar.archived, cl.deletedPVC)
+	}
+	var owner sql.NullString
+	var newest string
+	if err := db.QueryRowContext(ctx, `SELECT owner_id FROM servers WHERE name = $1`, name).Scan(&owner); err != nil {
+		t.Fatal(err)
+	}
+	if err := db.QueryRowContext(ctx,
+		`SELECT backup_ref FROM world_backups WHERE server_name = $1 ORDER BY created_at DESC LIMIT 1`, name).Scan(&newest); err != nil {
+		t.Fatal(err)
+	}
+	if owner.Valid || newest != ar.archived[0] {
+		t.Fatalf("owner = %v, newest backup = %q; want released and %q", owner, newest, ar.archived[0])
+	}
+}

internal/reaper/pgstore.go

@@ -50,9 +50,14 @@ func (s *PGStore) ListActiveServers(ctx context.Context) ([]Candidate, error) {
 }
 
 func (s *PGStore) FreshBackup(ctx context.Context, server string, since time.Time) (Fresh, bool, error) {
-	const q = `SELECT backup_ref, offsite_at IS NOT NULL FROM world_backups
+	// Only the reaper's own archives count, and only those taken since the
-		WHERE server_name = $1 AND status = 'present' AND created_at >= $2
+	// current owner claimed the server: a manual backup may predate a panel edit
-		ORDER BY offsite_at IS NOT NULL DESC, created_at DESC LIMIT 1`
+	// that did not move last_active_at, and an archive from before the claim is
+	// the previous owner's world.
+	const q = `SELECT b.backup_ref, b.offsite_at IS NOT NULL FROM world_backups b
+		WHERE b.server_name = $1 AND b.status = 'present' AND b.reason = 'inactive_15d' AND b.created_at >= $2
+		  AND b.created_at >= COALESCE((SELECT s.claimed_at FROM servers s WHERE s.name = $1 AND s.deleted_at IS NULL), '-infinity')
+		ORDER BY b.offsite_at IS NOT NULL DESC, b.created_at DESC LIMIT 1`
 	var f Fresh
 	switch err := s.db.QueryRowContext(ctx, q, server, since).Scan(&f.Ref, &f.Offsite); {
 	case err == sql.ErrNoRows:

internal/reaper/reaper.go

@@ -168,8 +168,9 @@ type Store interface {
 	// ListActiveServers returns every servers row with deleted_at IS NULL.
 	ListActiveServers(ctx context.Context) ([]Candidate, error)
 
-	// FreshBackup reports an existing present backup for server whose world is
+	// FreshBackup reports an existing present reaper archive (reason
-	// still current — created at or after since (the world's last_active_at),
+	// inactive_15d) for server whose world is still current — created at or
+	// after since (the world's last_active_at) and after the current claim,
 	// preferring one already copied off-site. It makes a reap idempotent across
 	// a DeletePVC failure, and across the wait for the off-site copy: the retry
 	// reuses the archive instead of writing a duplicate.

internal/reaper/reaper_test.go

@@ -104,6 +104,7 @@ func (c *fakeCluster) Stop(_ context.Context, name string) error {
 
 type fakeBackup struct {
 	id, server, ref    string
+	reason             string
 	size               int64
 	status             string // present | deleted
 	createdAt, expires time.Time
@@ -137,7 +138,7 @@ func (s *fakeStore) ListActiveServers(context.Context) ([]Candidate, error) {
 func (s *fakeStore) FreshBackup(_ context.Context, server string, since time.Time) (Fresh, bool, error) {
 	var found *fakeBackup
 	for _, b := range s.backups {
-		if b.server == server && b.status == "present" && !b.createdAt.Before(since) {
+		if b.server == server && b.status == "present" && b.reason == ReasonInactive && !b.createdAt.Before(since) {
 			if found == nil || (b.offsite && !found.offsite) {
 				found = b
 			}
@@ -154,7 +155,7 @@ func (s *fakeStore) InsertBackup(_ context.Context, rec BackupRecord) error {
 		return s.insertErr
 	}
 	s.backups = append(s.backups, &fakeBackup{
-		id: rec.ID, server: rec.ServerName, ref: rec.BackupRef, size: rec.SizeBytes,
+		id: rec.ID, server: rec.ServerName, ref: rec.BackupRef, reason: rec.Reason, size: rec.SizeBytes,
 		status: "present", createdAt: s.clock, expires: rec.ExpiresAt,
 	})
 	s.rec.add("insert")
@@ -433,6 +434,21 @@ func TestReapDeletePVCFailureIsIdempotent(t *testing.T) {
 	}
 }
 
+// A manual backup taken after the last join is not the reaper's archive: the
+// owner may have edited the world from the panel since, which does not move
+// last_active_at. The reap writes its own archive.
+func TestReapDoesNotReuseManualBackup(t *testing.T) {
+	r, st, _, ar := newReaper(DefaultConfig(),
+		Candidate{Name: "eta", OwnerID: "user-7", LastActiveAt: idleBy(20 * Day)})
+	st.backups = []*fakeBackup{
+		{id: "man", server: "eta", ref: "ref-man", reason: "manual", size: 5, status: "present", createdAt: idleBy(10 * Day), expires: testNow.Add(80 * Day)},
+	}
+	sum := mustRun(t, r)
+	if sum.WorldsReaped != 1 || ar.archives != 1 {
+		t.Fatalf("summary = %+v, archives = %d: want the reap to archive afresh", sum, ar.archives)
+	}
+}
+
 // With an off-site bucket configured the reaper archives an idle world but
 // keeps it until the archive's copy is confirmed; the next run reuses that
 // archive and deletes the world, without archiving it again.