feat(federation): reset-cleanup panel — informational detach copy, real server-side Dismiss, Keep removed (detach spec §4.6)

This commit is contained in:
Jannis Braun
2026-07-02 19:06:21 +02:00
parent 42ad5e141d
commit 5ad8aefaff
5 changed files with 162 additions and 67 deletions
+1 -1
View File
@@ -536,7 +536,7 @@ Manages: federation peers list, pending approval requests (inbound + outbound),
- **Reset cleanup section** (`ResetCleanup`, instance-epoch self-healing §6.4) — the highest-priority attention surface, rendered above the peer list; returns `null` when there is nothing to clean up. Fetches `api.federation.peers()` + `api.federation.resetEvents()` and subscribes to `onFederationPeerResetDetected` (the `federation_peer_reset_detected` WS event) to refetch live. Two stacked surfaces:
- **Reset-detected banner** — one persistent accent-rose banner per peer with `status === 'needs_attention' && needsAttentionReason === 'peer_reset_detected'`, distinguishing a wiped-and-reinstalled peer from a generic auth-failure peer. Its **Re-peer** button runs the existing one-click flow in order: `api.federation.resetPeer(id)` **then** `api.federation.initiatePeering({ remoteOrigin })` — resetting the stale local record *before* the fresh handshake so activation heals stale friendships/DMs against the new incarnation (warning-variant ConfirmDialog). **The outcome is now surfaced honestly, not always as success:** if `initiatePeering` resolves with `verified === false` (or the returned peer is `needs_attention`), the toast is a **warning** — "Re-peer incomplete — {peer} still holds stale peering for you. Its admin must reset their side, then Re-peer again." — because the handshake could not be cryptographically verified (the remote still holds a conflicting row; see `federation.md` "Trust re-establishment contract"). If it rejects with `409 PEER_EXISTS_RESET_REQUIRED`, the same "ask the remote admin to reset their side" warning is shown. Only a verified activation shows the green "Re-peering initiated" success toast. This means the common one-side reset recovers in one click, while a bidirectional-stale peering tells the admin exactly that the **other** side must reset once.
- **Orphaned-accounts list** (from `GET /reset-events`, per origin with `orphanedAccounts.length > 0`)each real account frozen by the reset quarantine, shown with its owned-spaces / membership / message counts. Per-row actions: **Keep** (default no-op resting/frozen state — the account stays `federationHomeOrphaned = 1`) and **Remove** (danger ConfirmDialog → `api.admin.deleteUser(id)`, i.e. the existing `DELETE /api/admin/users/:id` full purge). A Remove on a space-owning account returns the existing `409 { ownedSpaces }`; the UI surfaces a "transfer ownership first" toast rather than deleting.
- **Detached-accounts card** (from `GET /reset-events`, per origin with `orphanedAccounts.length > 0 && acknowledgedAt === null` — acknowledged events are filtered client-side, the endpoint keeps returning them for audit) — neutral-tier (`bg-white/[0.02]`, no rose/urgency styling; the rose banner is reserved for the actionable reset-detected surface). Copy is informational, not urgent-cleanup: "{origin} was reset — N replicated identities auto-cleaned, N accounts with local content detached. Detached accounts keep working locally — owners keep access with their existing password." Each detached real account is shown with its owned-spaces / membership / message counts. Actions: a per-account **Remove** (danger ConfirmDialog → `api.admin.deleteUser(id)`, i.e. the existing `DELETE /api/admin/users/:id` full purge, for accounts that truly are abandoned; a Remove on a space-owning account returns the existing `409 { ownedSpaces }` → "transfer ownership first" toast rather than deleting) and a per-event **Dismiss — keep all detached accounts** footer button that calls `api.federation.acknowledgeResetEvent(origin)` (`POST /api/federation/reset-events/acknowledge`) then re-fetches — a real, server-side acknowledgement (not the old client-only "Keep") that hides the card and drops the event from the badge count without touching any account. The badge counts unacknowledged events-with-orphans plus reset-detected peers.
#### StoragePanel
+1 -1
View File
@@ -480,7 +480,7 @@ This slice is intentionally separate from `instanceStore` because the data is pe
Modeled on the peering-approval surface above, the FederationPanel's `ResetCleanup` component (`admin.md` "FederationPanel") is the admin surface for a factory-reset peer. It fetches `api.federation.peers()` + `api.federation.resetEvents()` (`GET /api/federation/reset-events`) and subscribes to `onFederationPeerResetDetected(cb)` — the client handler for the `federation_peer_reset_detected` admin WS event (`useWebSocket.ts`) — to refetch live. Two surfaces:
- **Reset-detected banner** — one persistent accent-rose banner per peer with `status === 'needs_attention' && needsAttentionReason === 'peer_reset_detected'` (the `needsAttentionReason` field distinguishes a reset from a generic auth-failure, and now also `'repeer_incomplete'`). **Re-peer** runs `resetPeer(id)` **then** `initiatePeering({ remoteOrigin })` — reset-before-handshake so activation heals the stale graph against the new incarnation. **The result is surfaced honestly:** `initiatePeering` now returns `{ peer, verified }`; when `verified === false` (or the peer comes back `needs_attention`), or when it rejects with `409 PEER_EXISTS_RESET_REQUIRED`, the toast is a **warning** telling the admin the remote still holds stale peering and its admin must reset the **other** side, then Re-peer again — rather than a false success. A cryptographically-verified activation shows the success toast. The common one-side reset recovers in one click; a bidirectional-stale case names the side that must act. See `federation.md` "Trust re-establishment contract".
- **Orphaned-accounts list** — real accounts frozen by the server-side reset quarantine (`FederationOrphanedAccount`: owned-spaces / membership / message counts). **Keep** is the no-op frozen resting state; **Remove** reuses `api.admin.deleteUser(id)` (`DELETE /api/admin/users/:id`, full purge). A Remove on a space owner surfaces the existing `409 { ownedSpaces }` as a "transfer ownership first" toast instead of deleting.
- **Detached-accounts card** — informational, neutral-tier surface (no rose/urgency styling) for the reset incarnation's real accounts that now operate as sovereign local accounts (`FederationOrphanedAccount`: owned-spaces / membership / message counts). Copy: detached accounts keep working locally and owners sign in with their existing password. Cards render only for unacknowledged events (`orphanedAccounts.length > 0 && acknowledgedAt === null`; the endpoint still returns acknowledged events for audit). Per-account **Remove** reuses `api.admin.deleteUser(id)` (`DELETE /api/admin/users/:id`, full purge) for genuinely-abandoned accounts — a Remove on a space owner surfaces the existing `409 { ownedSpaces }` as a "transfer ownership first" toast instead of deleting. A per-event **Dismiss** footer calls `api.federation.acknowledgeResetEvent(origin)` (`POST /api/federation/reset-events/acknowledge`) then re-fetches — a real server-side acknowledgement (replacing the old client-only "Keep") that hides the card and removes the event from the badge count without touching any account.
---
+1 -1
View File
@@ -370,7 +370,7 @@ Detection (`markPeerReset`) only snapshots + journals + notifies; it destroys no
**Login self-heal epoch guard (design §6.3a).** The federated password self-heal (`auth.ts` §4) now gates re-hashing on the home instance's current epoch, read via the authenticated `fetchPeerEpoch(peer)` (HMAC-signed both ways): no baseline on record → allow (legacy); baseline differs from the fetched epoch → refuse; epoch can't be determined (`fetchPeerEpoch` null — 404/unreachable/bad-sig/desynced secret) → **fail closed/refuse**; match → allow. Closes the *pre*-re-peer hijack (a reset home accepting a new same-name user's password); the universal quarantine freeze closes the post-re-peer window. Full three-way in `auth.md` §4.
**Reset-events admin surface (`GET /api/federation/reset-events`).** Admin-only, read-only. Returns the durable `federation_reset_events` journal joined with each origin's current orphaned real accounts (`federation_home_orphaned = 1`, `homeInstanceMatch`), each with `ownedSpaces`, `spaceMemberCount`, and authored-`messageCount` for disposition. Response type `FederationResetEventsResponse` (`{ events: FederationResetEvent[] }`, each event carrying `orphanedAccounts: FederationOrphanedAccount[]`). Disposition actions reuse existing endpoints — one-click Re-peer (`/peers/:id/reset``/peer/initiate`) and full-purge Remove (`DELETE /api/admin/users/:id`, owns-spaces → transfer first). See `admin.md` "FederationPanel" and `client-federation.md` §8.
**Reset-events admin surface (`GET /api/federation/reset-events`).** Admin-only. Returns the durable `federation_reset_events` journal (each event carrying a nullable `acknowledgedAt`) joined with each origin's current detached real accounts (`federation_home_orphaned = 1`, `homeInstanceMatch`), each with `ownedSpaces`, `spaceMemberCount`, and authored-`messageCount` for disposition. Response type `FederationResetEventsResponse` (`{ events: FederationResetEvent[] }`, each event carrying `orphanedAccounts: FederationOrphanedAccount[]`). The endpoint returns **all** events (including acknowledged ones, for audit); the client filters to `acknowledgedAt === null`. Disposition actions — one-click Re-peer (`/peers/:id/reset``/peer/initiate`), full-purge Remove (`DELETE /api/admin/users/:id`, owns-spaces → transfer first) for genuinely-abandoned detached accounts, and a non-destructive **Dismiss** (`POST /api/federation/reset-events/acknowledge` with `{ origin }`, idempotent — sets `acknowledged_at`) that hides the card without touching the accounts (detached accounts keep working locally). See `admin.md` "FederationPanel" and `client-federation.md` §8.
**`needsAttentionReason` on the peer API.** `GET /api/federation/peers` returns `needsAttentionReason: 'auth_failures' | 'peer_reset_detected' | 'repeer_incomplete' | null` per peer, so the admin UI distinguishes a reset-detected peer (persistent Reset-cleanup banner + one-click Re-peer) from a generic auth-failure peer (plain "Reset Peering") and from a peer whose Re-peer could not be cryptographically verified (`repeer_incomplete` — see "Trust re-establishment contract"). `repeer_incomplete` is a nullable-TEXT value only; no schema migration.
@@ -1,9 +1,10 @@
import { describe, it, expect, vi, beforeEach } from 'vitest';
import { render, screen, fireEvent, waitFor } from '@testing-library/react';
const { peers, resetEvents, resetPeer, initiatePeering, deleteUser, addToast } = vi.hoisted(() => ({
const { peers, resetEvents, acknowledgeResetEvent, resetPeer, initiatePeering, deleteUser, addToast } = vi.hoisted(() => ({
peers: vi.fn(),
resetEvents: vi.fn(),
acknowledgeResetEvent: vi.fn(),
resetPeer: vi.fn(),
initiatePeering: vi.fn(),
deleteUser: vi.fn(),
@@ -21,6 +22,7 @@ vi.mock('../../../api/client', async () => {
peers,
approvalRequests: vi.fn().mockResolvedValue({ requests: [] }),
resetEvents,
acknowledgeResetEvent,
resetPeer,
initiatePeering,
},
@@ -73,16 +75,21 @@ function orphanedAccount(overrides: Record<string, unknown> = {}) {
};
}
function resetEvent(accounts: ReturnType<typeof orphanedAccount>[]) {
function resetEvent(
accounts: ReturnType<typeof orphanedAccount>[],
overrides: Record<string, unknown> = {},
) {
return {
origin: 'https://peer.example',
deadEpoch: 'epoch-old',
newEpoch: 'epoch-new',
detectedAt: Date.now(),
resolvedAt: null,
acknowledgedAt: null,
stubCount: 3,
orphanedAccountCount: accounts.length,
orphanedAccounts: accounts,
...overrides,
};
}
@@ -90,6 +97,7 @@ describe('FederationPanel — Reset cleanup', () => {
beforeEach(() => {
peers.mockReset();
resetEvents.mockReset();
acknowledgeResetEvent.mockReset();
resetPeer.mockReset();
initiatePeering.mockReset();
deleteUser.mockReset();
@@ -272,4 +280,80 @@ describe('FederationPanel — Reset cleanup', () => {
'warning',
);
});
it('renders the detached-accounts card with Dismiss + Remove and informational copy, no Keep/frozen', async () => {
peers.mockResolvedValue({ peers: [] });
resetEvents.mockResolvedValue({ events: [resetEvent([orphanedAccount()])] });
render(<FederationPanel />);
// Both real actions are present.
await screen.findByRole('button', { name: /Dismiss/ });
expect(screen.getByRole('button', { name: 'Remove' })).toBeInTheDocument();
// Informational detach copy — not urgent-cleanup language.
expect(screen.getAllByText(/detached/i).length).toBeGreaterThan(0);
expect(screen.getByText(/existing password/i)).toBeInTheDocument();
// The fake client-only Keep/frozen affordance is fully gone.
expect(screen.queryByRole('button', { name: 'Keep' })).not.toBeInTheDocument();
expect(screen.queryByText(/frozen/i)).not.toBeInTheDocument();
// No "orphaned" urgency wording in the detached-accounts copy.
expect(screen.queryByText(/with local content orphaned/i)).not.toBeInTheDocument();
});
it('does not render an acknowledged event and excludes it from the badge count', async () => {
peers.mockResolvedValue({ peers: [] });
resetEvents.mockResolvedValue({
events: [resetEvent([orphanedAccount()], { acknowledgedAt: 1234 })],
});
render(<FederationPanel />);
// Give effects a chance to run, then assert the whole section stays absent.
await waitFor(() => expect(resetEvents).toHaveBeenCalled());
await waitFor(() =>
expect(screen.queryByText('Reset Cleanup')).not.toBeInTheDocument(),
);
expect(screen.queryByRole('button', { name: 'Remove' })).not.toBeInTheDocument();
expect(screen.queryByRole('button', { name: /Dismiss/ })).not.toBeInTheDocument();
});
it('dismisses an event via the acknowledge API and re-fetches', async () => {
peers.mockResolvedValue({ peers: [] });
// First load: unacknowledged. After acknowledge, re-fetch returns it acknowledged.
resetEvents
.mockResolvedValueOnce({ events: [resetEvent([orphanedAccount()])] })
.mockResolvedValue({ events: [resetEvent([orphanedAccount()], { acknowledgedAt: 1234 })] });
acknowledgeResetEvent.mockResolvedValue({ success: true });
render(<FederationPanel />);
const dismissBtn = await screen.findByRole('button', { name: /Dismiss/ });
fireEvent.click(dismissBtn);
await waitFor(() =>
expect(acknowledgeResetEvent).toHaveBeenCalledWith('https://peer.example'),
);
// fetchAll re-runs after acknowledge (peers + resetEvents both hit twice).
await waitFor(() => expect(resetEvents).toHaveBeenCalledTimes(2));
// The card disappears once the re-fetch marks the event acknowledged.
await waitFor(() =>
expect(screen.queryByRole('button', { name: /Dismiss/ })).not.toBeInTheDocument(),
);
});
it('surfaces an error toast when dismiss fails', async () => {
peers.mockResolvedValue({ peers: [] });
resetEvents.mockResolvedValue({ events: [resetEvent([orphanedAccount()])] });
acknowledgeResetEvent.mockRejectedValue(new Error('Network down'));
render(<FederationPanel />);
fireEvent.click(await screen.findByRole('button', { name: /Dismiss/ }));
await waitFor(() =>
expect(addToast).toHaveBeenCalledWith('Network down', 'warning'),
);
});
});
@@ -815,15 +815,21 @@ function PendingApprovals({ onCountChange }: { onCountChange?: (count: number) =
// ─── Reset Cleanup ──────────────────────────────────────────────────────────
//
// Highest-priority admin attention surface for the instance-epoch self-healing
// flow (§6.4). Two stacked surfaces:
// Admin attention surface for the instance-epoch self-healing flow (§6.4) and
// the orphaned-account detach flow (detach spec §4.6). Two stacked surfaces:
// 1. A persistent accent-rose banner per peer detected as reset
// (status === 'needs_attention' && needsAttentionReason === 'peer_reset_detected'),
// with a one-click Re-peer (resetPeer → initiatePeering) that triggers the
// server-side heal on activation.
// 2. Per-origin lists of the dead incarnation's orphaned real accounts with
// Keep (no-op resting/frozen state) and Remove (full purge via the existing
// admin delete) actions.
// server-side heal on activation. This one is genuinely actionable, so it
// keeps the rose/danger styling.
// 2. Per-origin, informational cards for the dead incarnation's detached real
// accounts. Detachment is not a failure state: these accounts keep working
// locally and their owners sign in with the same password. The card offers a
// real, server-side Dismiss (acknowledgeResetEvent — hides the card without
// touching the accounts) and a per-account Remove (full purge via the existing
// admin delete) for the ones that truly are abandoned. Neutral tier styling —
// no urgency. Acknowledged events are filtered out client-side (the endpoint
// keeps returning them for audit).
function peerName(peer: FederationPeer): string {
if (peer.instanceName) return peer.instanceName;
@@ -853,7 +859,6 @@ function ResetCleanup() {
const [loading, setLoading] = useState(false);
const [confirmAction, setConfirmAction] = useState<ResetConfirmAction | null>(null);
const [actionLoading, setActionLoading] = useState(false);
const [keptIds, setKeptIds] = useState<Set<string>>(new Set());
const fetchAll = useCallback(async () => {
setLoading(true);
@@ -924,11 +929,6 @@ function ResetCleanup() {
const { account } = confirmAction;
await api.admin.deleteUser(account.id);
addToast(`Removed ${account.username} and all their content`, 'success', 3000);
setKeptIds((prev) => {
const next = new Set(prev);
next.delete(account.id);
return next;
});
await fetchAll();
}
} catch (err) {
@@ -962,7 +962,25 @@ function ResetCleanup() {
}
};
const eventsWithOrphans = events.filter((e) => e.orphanedAccounts.length > 0);
// Dismiss the detached-accounts card without touching the accounts — the event
// stays in the DB (acknowledged) for audit but stops surfacing to the admin.
const handleDismiss = async (origin: string) => {
setActionLoading(true);
try {
await api.federation.acknowledgeResetEvent(origin);
await fetchAll();
} catch (err) {
addToast(err instanceof Error ? err.message : 'Failed to dismiss', 'warning');
} finally {
setActionLoading(false);
}
};
// Only unacknowledged events with detached accounts surface a card. Dismissed
// (acknowledged) events are filtered out here and drop off the badge count.
const eventsWithOrphans = events.filter(
(e) => e.orphanedAccounts.length > 0 && e.acknowledgedAt === null,
);
// Render nothing when there is no reset-detected peer and no orphaned account —
// exactly as PendingApprovals returns null when empty (loading also renders null).
@@ -973,7 +991,7 @@ function ResetCleanup() {
<div className="flex items-center gap-2 mb-1.5">
<div className="text-[11px] font-semibold text-txt-tertiary uppercase tracking-wider">Reset Cleanup</div>
<span className="text-[10px] font-medium px-1.5 py-0.5 rounded-full bg-accent-rose/15 text-accent-rose">
{resetPeers.length + eventsWithOrphans.reduce((n, e) => n + e.orphanedAccounts.length, 0)}
{resetPeers.length + eventsWithOrphans.length}
</span>
</div>
@@ -1019,12 +1037,11 @@ function ResetCleanup() {
was reset {event.stubCount} replicated{' '}
{event.stubCount === 1 ? 'identity' : 'identities'} auto-cleaned,{' '}
{event.orphanedAccounts.length}{' '}
{event.orphanedAccounts.length === 1 ? 'account' : 'accounts'} with local content orphaned.
{event.orphanedAccounts.length === 1 ? 'account' : 'accounts'} with local content detached.
Detached accounts keep working locally owners keep access with their existing password.
</div>
<div className="space-y-2">
{event.orphanedAccounts.map((account) => {
const kept = keptIds.has(account.id);
return (
{event.orphanedAccounts.map((account) => (
<div key={account.id} className="bg-white/[0.02] rounded-md px-3 py-2.5">
<div className="flex items-center justify-between gap-3">
<div className="min-w-0">
@@ -1045,19 +1062,6 @@ function ResetCleanup() {
)}
</div>
<div className="flex items-center gap-2 shrink-0 ml-3">
{kept ? (
<span className="text-[11px] text-txt-tertiary italic">Kept frozen</span>
) : (
<button
type="button"
onClick={() =>
setKeptIds((prev) => new Set(prev).add(account.id))
}
className="px-3 py-1.5 text-xs font-medium text-txt-tertiary hover:text-txt-secondary bg-white/[0.04] hover:bg-white/[0.06] rounded transition-colors"
>
Keep
</button>
)}
<button
type="button"
onClick={() =>
@@ -1071,9 +1075,16 @@ function ResetCleanup() {
</div>
</div>
</div>
);
})}
))}
</div>
<button
type="button"
onClick={() => handleDismiss(event.origin)}
disabled={actionLoading}
className="mt-2 px-3 py-1.5 text-xs font-medium text-txt-tertiary hover:text-txt-secondary bg-white/[0.04] hover:bg-white/[0.06] rounded transition-colors disabled:opacity-50"
>
Dismiss keep all detached accounts
</button>
</div>
))}
</div>