A container registry that uses the AT Protocol for manifest storage and S3 for blob storage.
atcr.io
docker
container
atproto
go
clean up old docs add quotas prelim spec
+1289
-1901
-826
docs/API_KEY_MIGRATION.md
-826
docs/API_KEY_MIGRATION.md
···
1
-
# API Key Migration Plan
2
-
3
-
## Overview
4
-
5
-
Replace the session token system (used only by credential helper) with API keys that link to OAuth sessions. This simplifies authentication while maintaining all use cases.
6
-
7
-
## Current State
8
-
9
-
### Three Separate Auth Systems
10
-
11
-
1. **Session Tokens** (`pkg/auth/session/`)
12
-
- JWT-like tokens: `<base64_claims>.<base64_signature>`
13
-
- Created after OAuth callback, shown to user to copy
14
-
- User manually pastes into credential helper config
15
-
- Validated in `/auth/token` and `/auth/exchange`
16
-
- 30-day TTL
17
-
- **Problem:** Awkward UX, requires manual copy/paste
18
-
19
-
2. **UI Sessions** (`pkg/appview/session/`)
20
-
- Cookie-based (`atcr_session`)
21
-
- Random session ID, server-side store
22
-
- 24-hour TTL
23
-
- **Keep this - works well**
24
-
25
-
3. **App Password Auth** (via PDS)
26
-
- Direct `com.atproto.server.createSession` call
27
-
- No AppView involvement until token request
28
-
- **Keep this - essential for non-UI users**
29
-
30
-
## Target State
31
-
32
-
### Two Auth Methods
33
-
34
-
1. **API Keys** (NEW - replaces session tokens)
35
-
- Generated in UI after OAuth login
36
-
- Format: `atcr_<32_bytes_base64>`
37
-
- Linked to server-side OAuth refresh token
38
-
- Multiple keys per user (laptop, CI/CD, etc.)
39
-
- Revocable without re-auth
40
-
41
-
2. **App Passwords** (KEEP)
42
-
- Direct PDS authentication
43
-
- Works without UI/OAuth
44
-
45
-
### UI Sessions (UNCHANGED)
46
-
- Cookie-based for web UI
47
-
- Separate system, no changes needed
48
-
49
-
---
50
-
51
-
## Implementation Plan
52
-
53
-
### Phase 1: API Key System
54
-
55
-
#### 1.1 Create API Key Store (`pkg/appview/apikey/store.go`)
56
-
57
-
```go
58
-
package apikey
59
-
60
-
import (
61
-
"crypto/rand"
62
-
"encoding/base64"
63
-
"encoding/json"
64
-
"fmt"
65
-
"os"
66
-
"sync"
67
-
"time"
68
-
"golang.org/x/crypto/bcrypt"
69
-
)
70
-
71
-
// APIKey represents a user's API key
72
-
type APIKey struct {
73
-
ID string `json:"id"` // UUID
74
-
KeyHash string `json:"key_hash"` // bcrypt hash
75
-
DID string `json:"did"` // Owner's DID
76
-
Handle string `json:"handle"` // Owner's handle
77
-
Name string `json:"name"` // User-provided name
78
-
CreatedAt time.Time `json:"created_at"`
79
-
LastUsed time.Time `json:"last_used"`
80
-
}
81
-
82
-
// Store manages API keys
83
-
type Store struct {
84
-
mu sync.RWMutex
85
-
keys map[string]*APIKey // keyHash -> APIKey
86
-
byDID map[string][]string // DID -> []keyHash
87
-
filePath string // /var/lib/atcr/api-keys.json
88
-
}
89
-
90
-
// NewStore creates a new API key store
91
-
func NewStore(filePath string) (*Store, error)
92
-
93
-
// Generate creates a new API key and returns the plaintext key (shown once)
94
-
func (s *Store) Generate(did, handle, name string) (key string, keyID string, err error)
95
-
96
-
// Validate checks if an API key is valid and returns the associated data
97
-
func (s *Store) Validate(key string) (*APIKey, error)
98
-
99
-
// List returns all API keys for a DID (without plaintext keys)
100
-
func (s *Store) List(did string) []*APIKey
101
-
102
-
// Delete removes an API key
103
-
func (s *Store) Delete(did, keyID string) error
104
-
105
-
// UpdateLastUsed updates the last used timestamp
106
-
func (s *Store) UpdateLastUsed(keyHash string) error
107
-
```
108
-
109
-
**Key Generation:**
110
-
```go
111
-
func (s *Store) Generate(did, handle, name string) (string, string, error) {
112
-
// Generate 32 random bytes
113
-
b := make([]byte, 32)
114
-
if _, err := rand.Read(b); err != nil {
115
-
return "", "", err
116
-
}
117
-
118
-
// Format: atcr_<base64>
119
-
key := "atcr_" + base64.RawURLEncoding.EncodeToString(b)
120
-
121
-
// Hash for storage
122
-
keyHash, err := bcrypt.GenerateFromPassword([]byte(key), bcrypt.DefaultCost)
123
-
if err != nil {
124
-
return "", "", err
125
-
}
126
-
127
-
// Generate ID
128
-
keyID := generateUUID()
129
-
130
-
apiKey := &APIKey{
131
-
ID: keyID,
132
-
KeyHash: string(keyHash),
133
-
DID: did,
134
-
Handle: handle,
135
-
Name: name,
136
-
CreatedAt: time.Now(),
137
-
LastUsed: time.Time{}, // Never used yet
138
-
}
139
-
140
-
s.mu.Lock()
141
-
s.keys[string(keyHash)] = apiKey
142
-
s.byDID[did] = append(s.byDID[did], string(keyHash))
143
-
s.mu.Unlock()
144
-
145
-
s.save()
146
-
147
-
// Return plaintext key (only time it's available)
148
-
return key, keyID, nil
149
-
}
150
-
```
151
-
152
-
**Key Validation:**
153
-
```go
154
-
func (s *Store) Validate(key string) (*APIKey, error) {
155
-
s.mu.RLock()
156
-
defer s.mu.RUnlock()
157
-
158
-
// Try to match against all stored hashes
159
-
for hash, apiKey := range s.keys {
160
-
if err := bcrypt.CompareHashAndPassword([]byte(hash), []byte(key)); err == nil {
161
-
// Update last used asynchronously
162
-
go s.UpdateLastUsed(hash)
163
-
return apiKey, nil
164
-
}
165
-
}
166
-
167
-
return nil, fmt.Errorf("invalid API key")
168
-
}
169
-
```
170
-
171
-
#### 1.2 Add API Key Handlers (`pkg/appview/handlers/apikeys.go`)
172
-
173
-
```go
174
-
package handlers
175
-
176
-
import (
177
-
"encoding/json"
178
-
"html/template"
179
-
"net/http"
180
-
"github.com/gorilla/mux"
181
-
"atcr.io/pkg/appview/apikey"
182
-
"atcr.io/pkg/appview/middleware"
183
-
)
184
-
185
-
// GenerateAPIKeyHandler handles POST /api/keys
186
-
type GenerateAPIKeyHandler struct {
187
-
Store *apikey.Store
188
-
}
189
-
190
-
func (h *GenerateAPIKeyHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
191
-
user := middleware.GetUser(r)
192
-
if user == nil {
193
-
http.Error(w, "Unauthorized", http.StatusUnauthorized)
194
-
return
195
-
}
196
-
197
-
name := r.FormValue("name")
198
-
if name == "" {
199
-
name = "Unnamed Key"
200
-
}
201
-
202
-
key, keyID, err := h.Store.Generate(user.DID, user.Handle, name)
203
-
if err != nil {
204
-
http.Error(w, "Failed to generate key", http.StatusInternalServerError)
205
-
return
206
-
}
207
-
208
-
// Return key (shown once!)
209
-
w.Header().Set("Content-Type", "application/json")
210
-
json.NewEncoder(w).Encode(map[string]string{
211
-
"id": keyID,
212
-
"key": key,
213
-
})
214
-
}
215
-
216
-
// ListAPIKeysHandler handles GET /api/keys
217
-
type ListAPIKeysHandler struct {
218
-
Store *apikey.Store
219
-
}
220
-
221
-
func (h *ListAPIKeysHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
222
-
user := middleware.GetUser(r)
223
-
if user == nil {
224
-
http.Error(w, "Unauthorized", http.StatusUnauthorized)
225
-
return
226
-
}
227
-
228
-
keys := h.Store.List(user.DID)
229
-
230
-
w.Header().Set("Content-Type", "application/json")
231
-
json.NewEncoder(w).Encode(keys)
232
-
}
233
-
234
-
// DeleteAPIKeyHandler handles DELETE /api/keys/{id}
235
-
type DeleteAPIKeyHandler struct {
236
-
Store *apikey.Store
237
-
}
238
-
239
-
func (h *DeleteAPIKeyHandler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
240
-
user := middleware.GetUser(r)
241
-
if user == nil {
242
-
http.Error(w, "Unauthorized", http.StatusUnauthorized)
243
-
return
244
-
}
245
-
246
-
vars := mux.Vars(r)
247
-
keyID := vars["id"]
248
-
249
-
if err := h.Store.Delete(user.DID, keyID); err != nil {
250
-
http.Error(w, "Failed to delete key", http.StatusInternalServerError)
251
-
return
252
-
}
253
-
254
-
w.WriteHeader(http.StatusNoContent)
255
-
}
256
-
```
257
-
258
-
### Phase 2: Update Token Handler
259
-
260
-
#### 2.1 Modify `/auth/token` Handler (`pkg/auth/token/handler.go`)
261
-
262
-
```go
263
-
type Handler struct {
264
-
issuer *Issuer
265
-
validator *atproto.SessionValidator
266
-
apiKeyStore *apikey.Store // NEW
267
-
defaultHoldEndpoint string
268
-
}
269
-
270
-
func (h *Handler) ServeHTTP(w http.ResponseWriter, r *http.Request) {
271
-
username, password, ok := r.BasicAuth()
272
-
if !ok {
273
-
return unauthorized
274
-
}
275
-
276
-
var did, handle, accessToken string
277
-
278
-
// 1. Check if it's an API key (NEW)
279
-
if strings.HasPrefix(password, "atcr_") {
280
-
apiKey, err := h.apiKeyStore.Validate(password)
281
-
if err != nil {
282
-
fmt.Printf("DEBUG [token/handler]: API key validation failed: %v\n", err)
283
-
return unauthorized
284
-
}
285
-
286
-
did = apiKey.DID
287
-
handle = apiKey.Handle
288
-
fmt.Printf("DEBUG [token/handler]: API key validated for DID=%s, handle=%s\n", did, handle)
289
-
290
-
// API key is linked to OAuth session
291
-
// OAuth refresher will provide access token when needed via middleware
292
-
}
293
-
// 2. Try app password (direct PDS)
294
-
else {
295
-
did, handle, accessToken, err = h.validator.CreateSessionAndGetToken(r.Context(), username, password)
296
-
if err != nil {
297
-
fmt.Printf("DEBUG [token/handler]: App password validation failed: %v\n", err)
298
-
return unauthorized
299
-
}
300
-
301
-
fmt.Printf("DEBUG [token/handler]: App password validated, DID=%s\n", did)
302
-
303
-
// Cache access token for manifest operations
304
-
auth.GetGlobalTokenCache().Set(did, accessToken, 2*time.Hour)
305
-
306
-
// Ensure profile exists
307
-
// ... existing code ...
308
-
}
309
-
310
-
// Rest of handler: validate access, issue JWT, etc.
311
-
// ... existing code ...
312
-
}
313
-
```
314
-
315
-
**Key Changes:**
316
-
- Remove session token validation (`sessionManager.Validate()`)
317
-
- Add API key check as first priority
318
-
- Keep app password as fallback
319
-
- API keys use OAuth refresher (server-side), app passwords use token cache (client-side)
320
-
321
-
#### 2.2 Remove `/auth/exchange` Endpoint
322
-
323
-
The `/auth/exchange` endpoint was only used for exchanging session tokens for registry JWTs. With API keys, this is no longer needed.
324
-
325
-
**Files to delete:**
326
-
- `pkg/auth/exchange/handler.go`
327
-
328
-
**Files to update:**
329
-
- `cmd/appview/serve.go` - Remove exchange handler registration
330
-
331
-
### Phase 3: Update UI
332
-
333
-
#### 3.1 Add API Keys Section to Settings Page
334
-
335
-
**Template** (`pkg/appview/templates/settings.html`):
336
-
337
-
```html
338
-
<!-- Add after existing profile settings -->
339
-
<section class="api-keys">
340
-
<h2>API Keys</h2>
341
-
<p>Generate API keys for Docker CLI and CI/CD. Each key is linked to your OAuth session.</p>
342
-
343
-
<!-- Generate New Key -->
344
-
<div class="generate-key">
345
-
<h3>Generate New API Key</h3>
346
-
<form id="generate-key-form">
347
-
<input type="text" id="key-name" placeholder="Key name (e.g., My Laptop)" required>
348
-
<button type="submit">Generate Key</button>
349
-
</form>
350
-
</div>
351
-
352
-
<!-- Key Generated Modal (shown once) -->
353
-
<div id="key-modal" class="modal hidden">
354
-
<div class="modal-content">
355
-
<h3>✓ API Key Generated!</h3>
356
-
<p><strong>Copy this key now - it won't be shown again:</strong></p>
357
-
<div class="key-display">
358
-
<code id="generated-key"></code>
359
-
<button onclick="copyKey()">Copy to Clipboard</button>
360
-
</div>
361
-
<div class="usage-instructions">
362
-
<h4>Using with Docker:</h4>
363
-
<pre>docker login atcr.io -u <span class="handle">{{.Profile.Handle}}</span> -p <span class="key-placeholder">[paste key here]</span></pre>
364
-
</div>
365
-
<button onclick="closeModal()">Done</button>
366
-
</div>
367
-
</div>
368
-
369
-
<!-- Existing Keys List -->
370
-
<div class="keys-list">
371
-
<h3>Your API Keys</h3>
372
-
<table>
373
-
<thead>
374
-
<tr>
375
-
<th>Name</th>
376
-
<th>Created</th>
377
-
<th>Last Used</th>
378
-
<th>Actions</th>
379
-
</tr>
380
-
</thead>
381
-
<tbody id="keys-table">
382
-
<!-- Populated via JavaScript -->
383
-
</tbody>
384
-
</table>
385
-
</div>
386
-
</section>
387
-
388
-
<script>
389
-
// Generate key
390
-
document.getElementById('generate-key-form').addEventListener('submit', async (e) => {
391
-
e.preventDefault();
392
-
const name = document.getElementById('key-name').value;
393
-
394
-
const resp = await fetch('/api/keys', {
395
-
method: 'POST',
396
-
headers: {'Content-Type': 'application/x-www-form-urlencoded'},
397
-
body: `name=${encodeURIComponent(name)}`
398
-
});
399
-
400
-
const data = await resp.json();
401
-
402
-
// Show key in modal (only time it's available)
403
-
document.getElementById('generated-key').textContent = data.key;
404
-
document.getElementById('key-modal').classList.remove('hidden');
405
-
406
-
// Refresh keys list
407
-
loadKeys();
408
-
});
409
-
410
-
// Copy key to clipboard
411
-
function copyKey() {
412
-
const key = document.getElementById('generated-key').textContent;
413
-
navigator.clipboard.writeText(key);
414
-
alert('Copied to clipboard!');
415
-
}
416
-
417
-
// Load existing keys
418
-
async function loadKeys() {
419
-
const resp = await fetch('/api/keys');
420
-
const keys = await resp.json();
421
-
422
-
const tbody = document.getElementById('keys-table');
423
-
tbody.innerHTML = keys.map(key => `
424
-
<tr>
425
-
<td>${key.name}</td>
426
-
<td>${new Date(key.created_at).toLocaleDateString()}</td>
427
-
<td>${key.last_used ? new Date(key.last_used).toLocaleDateString() : 'Never'}</td>
428
-
<td><button onclick="deleteKey('${key.id}')">Revoke</button></td>
429
-
</tr>
430
-
`).join('');
431
-
}
432
-
433
-
// Delete key
434
-
async function deleteKey(id) {
435
-
if (!confirm('Are you sure you want to revoke this key?')) return;
436
-
437
-
await fetch(`/api/keys/${id}`, { method: 'DELETE' });
438
-
loadKeys();
439
-
}
440
-
441
-
// Load keys on page load
442
-
loadKeys();
443
-
</script>
444
-
445
-
<style>
446
-
.modal.hidden { display: none; }
447
-
.modal {
448
-
position: fixed;
449
-
top: 0;
450
-
left: 0;
451
-
width: 100%;
452
-
height: 100%;
453
-
background: rgba(0,0,0,0.5);
454
-
display: flex;
455
-
align-items: center;
456
-
justify-content: center;
457
-
}
458
-
.modal-content {
459
-
background: white;
460
-
padding: 2rem;
461
-
border-radius: 8px;
462
-
max-width: 600px;
463
-
}
464
-
.key-display {
465
-
background: #f5f5f5;
466
-
padding: 1rem;
467
-
margin: 1rem 0;
468
-
border-radius: 4px;
469
-
}
470
-
.key-display code {
471
-
word-break: break-all;
472
-
font-size: 14px;
473
-
}
474
-
.usage-instructions {
475
-
margin-top: 1rem;
476
-
padding: 1rem;
477
-
background: #e3f2fd;
478
-
border-radius: 4px;
479
-
}
480
-
.usage-instructions pre {
481
-
background: #263238;
482
-
color: #aed581;
483
-
padding: 1rem;
484
-
border-radius: 4px;
485
-
overflow-x: auto;
486
-
}
487
-
.handle { color: #ffab40; }
488
-
.key-placeholder { color: #64b5f6; }
489
-
</style>
490
-
```
491
-
492
-
#### 3.2 Register API Key Routes (`cmd/appview/serve.go`)
493
-
494
-
```go
495
-
// In initializeUI() function, add:
496
-
497
-
// API key management routes (authenticated)
498
-
authRouter.Handle("/api/keys", &uihandlers.GenerateAPIKeyHandler{
499
-
Store: apiKeyStore,
500
-
}).Methods("POST")
501
-
502
-
authRouter.Handle("/api/keys", &uihandlers.ListAPIKeysHandler{
503
-
Store: apiKeyStore,
504
-
}).Methods("GET")
505
-
506
-
authRouter.Handle("/api/keys/{id}", &uihandlers.DeleteAPIKeyHandler{
507
-
Store: apiKeyStore,
508
-
}).Methods("DELETE")
509
-
```
510
-
511
-
### Phase 4: Update Credential Helper
512
-
513
-
#### 4.1 Simplify Configuration (`cmd/credential-helper/main.go`)
514
-
515
-
```go
516
-
// SessionStore becomes CredentialStore
517
-
type CredentialStore struct {
518
-
Handle string `json:"handle"`
519
-
APIKey string `json:"api_key"`
520
-
AppViewURL string `json:"appview_url"`
521
-
}
522
-
523
-
func handleConfigure(handle string) {
524
-
fmt.Println("ATCR Credential Helper Configuration")
525
-
fmt.Println("=====================================")
526
-
fmt.Println()
527
-
fmt.Println("You need an API key from the ATCR web UI.")
528
-
fmt.Println()
529
-
530
-
appViewURL := os.Getenv("ATCR_APPVIEW_URL")
531
-
if appViewURL == "" {
532
-
appViewURL = defaultAppViewURL
533
-
}
534
-
535
-
// Auto-open settings page
536
-
settingsURL := appViewURL + "/settings"
537
-
fmt.Printf("Opening settings page: %s\n", settingsURL)
538
-
fmt.Println("Log in and generate an API key if you haven't already.")
539
-
fmt.Println()
540
-
541
-
if err := oauth.OpenBrowser(settingsURL); err != nil {
542
-
fmt.Printf("Could not open browser. Please visit: %s\n\n", settingsURL)
543
-
}
544
-
545
-
// Prompt for credentials
546
-
if handle == "" {
547
-
fmt.Print("Enter your ATProto handle (e.g., alice.bsky.social): ")
548
-
fmt.Scanln(&handle)
549
-
} else {
550
-
fmt.Printf("Using handle: %s\n", handle)
551
-
}
552
-
553
-
fmt.Print("Enter your API key (from settings page): ")
554
-
var apiKey string
555
-
fmt.Scanln(&apiKey)
556
-
557
-
// Validate key format
558
-
if !strings.HasPrefix(apiKey, "atcr_") {
559
-
fmt.Fprintf(os.Stderr, "Invalid API key format. Key should start with 'atcr_'\n")
560
-
os.Exit(1)
561
-
}
562
-
563
-
// Save credentials
564
-
creds := &CredentialStore{
565
-
Handle: handle,
566
-
APIKey: apiKey,
567
-
AppViewURL: appViewURL,
568
-
}
569
-
570
-
if err := saveCredentials(getCredentialsPath(), creds); err != nil {
571
-
fmt.Fprintf(os.Stderr, "Error saving credentials: %v\n", err)
572
-
os.Exit(1)
573
-
}
574
-
575
-
fmt.Println()
576
-
fmt.Println("✓ Configuration complete!")
577
-
fmt.Println("You can now use docker push/pull with atcr.io")
578
-
}
579
-
580
-
func handleGet() {
581
-
var serverURL string
582
-
fmt.Fscanln(os.Stdin, &serverURL)
583
-
584
-
// Load credentials
585
-
creds, err := loadCredentials(getCredentialsPath())
586
-
if err != nil {
587
-
fmt.Fprintf(os.Stderr, "Error loading credentials: %v\n", err)
588
-
fmt.Fprintf(os.Stderr, "Please run: docker-credential-atcr configure\n")
589
-
os.Exit(1)
590
-
}
591
-
592
-
// Return credentials for Docker
593
-
// Docker will send these as Basic Auth to /auth/token
594
-
response := Credentials{
595
-
ServerURL: serverURL,
596
-
Username: creds.Handle,
597
-
Secret: creds.APIKey, // API key as password
598
-
}
599
-
600
-
json.NewEncoder(os.Stdout).Encode(response)
601
-
}
602
-
```
603
-
604
-
**File Rename:**
605
-
- `~/.atcr/session.json` → `~/.atcr/credentials.json`
606
-
607
-
### Phase 5: Remove Session Token System
608
-
609
-
#### 5.1 Delete Session Token Files
610
-
611
-
**Files to delete:**
612
-
- `pkg/auth/session/handler.go`
613
-
- `pkg/auth/exchange/handler.go`
614
-
615
-
#### 5.2 Update OAuth Server (`pkg/auth/oauth/server.go`)
616
-
617
-
**Remove session token creation:**
618
-
```go
619
-
// OLD (delete this):
620
-
sessionToken, err := s.sessionManager.Create(did, handle)
621
-
if err != nil {
622
-
s.renderError(w, fmt.Sprintf("Failed to create session token: %v", err))
623
-
return
624
-
}
625
-
626
-
// Check if this is a UI login...
627
-
if cookie, err := r.Cookie("oauth_return_to"); err == nil && s.uiSessionStore != nil {
628
-
// UI flow...
629
-
} else {
630
-
// Render success page with session token (for credential helper)
631
-
s.renderSuccess(w, sessionToken, handle)
632
-
}
633
-
```
634
-
635
-
**NEW (replace with):**
636
-
```go
637
-
// Check if this is a UI login
638
-
if cookie, err := r.Cookie("oauth_return_to"); err == nil && s.uiSessionStore != nil {
639
-
// Create UI session
640
-
uiSessionID, err := s.uiSessionStore.Create(did, handle, sessionData.HostURL, 24*time.Hour)
641
-
// ... set cookie, redirect ...
642
-
} else {
643
-
// Non-UI flow: redirect to settings to get API key
644
-
s.renderRedirectToSettings(w, handle)
645
-
}
646
-
```
647
-
648
-
**Add redirect to settings template:**
649
-
```go
650
-
func (s *Server) renderRedirectToSettings(w http.ResponseWriter, handle string) {
651
-
tmpl := template.Must(template.New("redirect").Parse(`
652
-
<!DOCTYPE html>
653
-
<html>
654
-
<head>
655
-
<title>Authorization Successful - ATCR</title>
656
-
<meta http-equiv="refresh" content="3;url=/settings">
657
-
</head>
658
-
<body>
659
-
<h1>✓ Authorization Successful!</h1>
660
-
<p>Redirecting to settings page to generate your API key...</p>
661
-
<p>If not redirected, <a href="/settings">click here</a>.</p>
662
-
</body>
663
-
</html>
664
-
`))
665
-
w.Header().Set("Content-Type", "text/html")
666
-
tmpl.Execute(w, nil)
667
-
}
668
-
```
669
-
670
-
#### 5.3 Update Server Constructor
671
-
672
-
```go
673
-
// Remove sessionManager parameter
674
-
func NewServer(app *App) *Server {
675
-
return &Server{
676
-
app: app,
677
-
}
678
-
}
679
-
```
680
-
681
-
#### 5.4 Update Registry Initialization (`cmd/appview/serve.go`)
682
-
683
-
```go
684
-
// REMOVE session manager creation:
685
-
// sessionManager, err := session.NewManagerWithPersistentSecret(secretPath, 30*24*time.Hour)
686
-
687
-
// Create API key store
688
-
apiKeyStorePath := filepath.Join(filepath.Dir(storagePath), "api-keys.json")
689
-
apiKeyStore, err := apikey.NewStore(apiKeyStorePath)
690
-
if err != nil {
691
-
return fmt.Errorf("failed to create API key store: %w", err)
692
-
}
693
-
694
-
// OAuth server doesn't need session manager anymore
695
-
oauthServer := oauth.NewServer(oauthApp)
696
-
oauthServer.SetRefresher(refresher)
697
-
if uiSessionStore != nil {
698
-
oauthServer.SetUISessionStore(uiSessionStore)
699
-
}
700
-
701
-
// Token handler gets API key store instead of session manager
702
-
if issuer != nil {
703
-
tokenHandler := token.NewHandler(issuer, apiKeyStore, defaultHoldEndpoint)
704
-
tokenHandler.RegisterRoutes(mux)
705
-
706
-
// Remove exchange handler registration (no longer needed)
707
-
}
708
-
```
709
-
710
-
---
711
-
712
-
## Migration Path
713
-
714
-
### For Existing Users
715
-
716
-
**Option 1: Smooth Migration (Recommended)**
717
-
1. Keep session token validation temporarily with deprecation warning
718
-
2. When session token is used, log warning and return special response header
719
-
3. Docker client shows warning: "Session tokens deprecated, please regenerate API key"
720
-
4. Remove session token support in next major version
721
-
722
-
**Option 2: Hard Cutover**
723
-
1. Deploy new version with API keys
724
-
2. Session tokens stop working immediately
725
-
3. Users must reconfigure: `docker-credential-atcr configure`
726
-
4. Cleaner but disruptive
727
-
728
-
### Rollout Plan
729
-
730
-
**Week 1: Deploy API Keys**
731
-
- Add API key system
732
-
- Keep session token validation
733
-
- Add deprecation notice to OAuth callback
734
-
735
-
**Week 2-4: Migration Period**
736
-
- Monitor API key adoption
737
-
- Email users about migration
738
-
- Provide migration guide
739
-
740
-
**Week 5: Remove Session Tokens**
741
-
- Delete session token code
742
-
- Force users to API keys
743
-
744
-
---
745
-
746
-
## Testing Plan
747
-
748
-
### Unit Tests
749
-
750
-
1. **API Key Store**
751
-
- Test key generation (format, uniqueness)
752
-
- Test key validation (correct/incorrect keys)
753
-
- Test bcrypt hashing
754
-
- Test key listing/deletion
755
-
756
-
2. **Token Handler**
757
-
- Test API key authentication
758
-
- Test app password authentication
759
-
- Test invalid credentials
760
-
- Test key format validation
761
-
762
-
### Integration Tests
763
-
764
-
1. **Full Auth Flow**
765
-
- UI login → OAuth → API key generation
766
-
- Credential helper → API key → registry JWT
767
-
- App password → registry JWT
768
-
769
-
2. **Docker Client Tests**
770
-
- `docker login -u handle -p api_key`
771
-
- `docker login -u handle -p app_password`
772
-
- `docker push` with API key
773
-
- `docker pull` with API key
774
-
775
-
### Security Tests
776
-
777
-
1. **Key Security**
778
-
- Verify bcrypt hashing (not plaintext storage)
779
-
- Test key shown only once
780
-
- Test key revocation
781
-
- Test unauthorized key access
782
-
783
-
2. **OAuth Security**
784
-
- Verify API key links to correct OAuth session
785
-
- Test expired refresh token handling
786
-
- Test multiple keys for same user
787
-
788
-
---
789
-
790
-
## Files Changed
791
-
792
-
### New Files
793
-
- `pkg/appview/apikey/store.go` - API key storage and validation
794
-
- `pkg/appview/handlers/apikeys.go` - API key HTTP handlers
795
-
- `docs/API_KEY_MIGRATION.md` - This document
796
-
797
-
### Modified Files
798
-
- `pkg/auth/token/handler.go` - Add API key validation, remove session token
799
-
- `pkg/auth/oauth/server.go` - Remove session token creation, redirect to settings
800
-
- `pkg/appview/handlers/settings.go` - Add API key management UI
801
-
- `pkg/appview/templates/settings.html` - Add API key section
802
-
- `cmd/credential-helper/main.go` - Simplify to use API keys
803
-
- `cmd/appview/serve.go` - Initialize API key store, remove session manager
804
-
805
-
### Deleted Files
806
-
- `pkg/auth/session/handler.go` - Session token system
807
-
- `pkg/auth/exchange/handler.go` - Exchange endpoint (no longer needed)
808
-
809
-
---
810
-
811
-
## Advantages
812
-
813
-
✅ **Simpler Auth:** Two methods instead of three (API keys + app passwords)
814
-
✅ **Better UX:** No manual copy/paste of session tokens
815
-
✅ **Multiple Keys:** Users can have laptop key, CI key, etc.
816
-
✅ **Revocable:** Revoke individual keys without re-auth
817
-
✅ **Server-Side OAuth:** Refresh tokens stay on server, not in client files
818
-
✅ **Familiar Pattern:** Matches AWS ECR, GitHub tokens, etc.
819
-
820
-
## Backward Compatibility
821
-
822
-
⚠️ **Breaking Change:** Session tokens will stop working
823
-
✅ **App passwords:** Still work (no changes)
824
-
✅ **UI sessions:** Still work (separate system)
825
-
826
-
**Migration Required:** Users with session tokens must run `docker-credential-atcr configure` again to get API keys.
-281
docs/OAUTH.md
-281
docs/OAUTH.md
···
1
-
# ATCR OAuth Implementation
2
-
3
-
## Overview
4
-
5
-
ATCR now supports ATProto OAuth authentication via Docker credential helpers. This allows users to authenticate with their ATProto identity (Bluesky account) and use Docker push/pull commands seamlessly.
6
-
7
-
## Architecture
8
-
9
-
### Components
10
-
11
-
1. **OAuth Client** (`pkg/auth/oauth/`)
12
-
- Full ATProto OAuth implementation with DPoP support
13
-
- Uses `authelia.com/client/oauth2` for OAuth + PAR
14
-
- Uses `github.com/AxisCommunications/go-dpop` for DPoP proof generation
15
-
- Automatic authorization server discovery
16
-
- PKCE support for security
17
-
18
-
2. **Credential Helper** (`cmd/credential-helper/`)
19
-
- Standalone binary: `docker-credential-atcr`
20
-
- Implements Docker credential helper protocol
21
-
- Manages OAuth flow with browser
22
-
- Stores tokens securely in `~/.atcr/oauth-token.json`
23
-
24
-
3. **Registry Integration**
25
-
- `/auth/exchange` endpoint exchanges OAuth tokens for registry JWTs
26
-
- Existing `/auth/token` endpoint for standard Docker auth
27
-
28
-
## Dependencies
29
-
30
-
- `authelia.com/client/oauth2` - OAuth client with PAR support (2⭐, Authelia-backed)
31
-
- `github.com/AxisCommunications/go-dpop` - DPoP implementation (10⭐, RFC 9449 compliant)
32
-
- `github.com/golang-jwt/jwt/v5` - JWT library (transitive, 11k+⭐)
33
-
34
-
## Usage
35
-
36
-
### Setup
37
-
38
-
1. Build the credential helper:
39
-
```bash
40
-
go build -o docker-credential-atcr ./cmd/credential-helper
41
-
```
42
-
43
-
2. Install it in your PATH:
44
-
```bash
45
-
sudo mv docker-credential-atcr /usr/local/bin/
46
-
```
47
-
48
-
3. Configure Docker to use it by editing `~/.docker/config.json`:
49
-
```json
50
-
{
51
-
"credsStore": "atcr"
52
-
}
53
-
```
54
-
55
-
### Configuration
56
-
57
-
Run the OAuth flow:
58
-
```bash
59
-
docker-credential-atcr configure
60
-
```
61
-
62
-
This will:
63
-
1. Prompt for your ATProto handle (e.g., `alice.bsky.social`)
64
-
2. Open your browser for OAuth authorization
65
-
3. Store the OAuth token and DPoP key in `~/.atcr/oauth-token.json`
66
-
67
-
### Using with Docker
68
-
69
-
Once configured, use Docker normally:
70
-
71
-
```bash
72
-
# Push an image
73
-
docker push atcr.io/alice/myapp:latest
74
-
75
-
# Pull an image
76
-
docker pull atcr.io/alice/myapp:latest
77
-
```
78
-
79
-
The credential helper automatically:
80
-
1. Loads your stored OAuth token
81
-
2. Refreshes it if expired
82
-
3. Exchanges it for a registry JWT
83
-
4. Provides the JWT to Docker
84
-
85
-
## How It Works
86
-
87
-
### OAuth Flow
88
-
89
-
1. **User runs** `docker-credential-atcr configure`
90
-
2. **Resolve identity**: alice.bsky.social → DID → PDS endpoint
91
-
3. **Discover auth server**: GET `{pds}/.well-known/oauth-authorization-server`
92
-
4. **Generate DPoP key**: ECDSA P-256 key pair
93
-
5. **PAR request**: POST to PAR endpoint with DPoP header + PKCE challenge
94
-
6. **Open browser**: User authorizes on their PDS
95
-
7. **Receive code**: Callback to `localhost:8888/callback`
96
-
8. **Exchange code**: POST to token endpoint with DPoP header + PKCE verifier
97
-
9. **Save tokens**: Store OAuth token + DPoP key + DID/handle
98
-
99
-
### Docker Push/Pull Flow
100
-
101
-
1. **Docker needs credentials** for `atcr.io`
102
-
2. **Calls credential helper**: `docker-credential-atcr get`
103
-
3. **Helper loads token** from `~/.atcr/oauth-token.json`
104
-
4. **Refresh if needed**: Uses refresh token + DPoP if expired
105
-
5. **Exchange for registry JWT**: POST to `/auth/exchange` with OAuth token + handle
106
-
6. **Registry validates token**: Calls `getSession` on PDS to validate token
107
-
7. **Registry issues JWT**: Creates registry JWT with validated DID/handle
108
-
8. **Return to Docker**: `{"Username": "oauth2", "Secret": "<jwt>"}`
109
-
9. **Docker uses JWT**: For authentication to registry API
110
-
111
-
## Security
112
-
113
-
### DPoP (Demonstrating Proof-of-Possession)
114
-
115
-
Every OAuth request includes a DPoP proof:
116
-
- Unique JWT signed with ECDSA private key
117
-
- Contains HTTP method, URL, timestamp, nonce
118
-
- Public key (JWK) included in JWT header
119
-
- Binds the token to the specific client
120
-
121
-
### PKCE (Proof Key for Code Exchange)
122
-
123
-
- Code verifier generated locally
124
-
- Code challenge sent in authorization request
125
-
- Verifier sent in token exchange
126
-
- Prevents authorization code interception
127
-
128
-
### Token Storage
129
-
130
-
- Tokens stored in `~/.atcr/oauth-token.json`
131
-
- File permissions: 0600 (owner read/write only)
132
-
- DPoP key stored in PEM format
133
-
- Refresh tokens for long-term access
134
-
135
-
## Implementation Details
136
-
137
-
### Code Structure
138
-
139
-
```
140
-
pkg/auth/oauth/
141
-
├── client.go # OAuth client with DPoP
142
-
├── discovery.go # Authorization server discovery
143
-
├── metadata.go # Client metadata document
144
-
├── storage.go # Token persistence
145
-
└── transport.go # DPoP HTTP transport
146
-
147
-
pkg/auth/atproto/
148
-
├── session.go # ATProto session validation (Basic auth)
149
-
└── validator.go # OAuth token validation via getSession
150
-
151
-
cmd/credential-helper/
152
-
├── main.go # Docker credential helper protocol
153
-
├── oauth.go # OAuth flow orchestration
154
-
└── token.go # Token management
155
-
156
-
pkg/auth/exchange/
157
-
└── handler.go # OAuth → Registry JWT exchange
158
-
```
159
-
160
-
### Key Classes
161
-
162
-
**OAuth Client** (`pkg/auth/oauth/client.go`)
163
-
- `NewClient()` - Create client with DPoP key
164
-
- `InitializeForHandle()` - Discover auth server
165
-
- `AuthorizeURL()` - Generate authorization URL with PAR + PKCE
166
-
- `Exchange()` - Exchange code for token with DPoP
167
-
- `RefreshToken()` - Refresh expired token with DPoP
168
-
169
-
**DPoP Transport** (`pkg/auth/oauth/transport.go`)
170
-
- Implements `http.RoundTripper`
171
-
- Automatically adds DPoP header to all requests
172
-
- Handles nonce management and retries
173
-
- Used by OAuth client for all HTTP requests
174
-
175
-
**Token Store** (`pkg/auth/oauth/storage.go`)
176
-
- Persists OAuth tokens and DPoP key
177
-
- PEM encoding for private key
178
-
- Expiration checking
179
-
- Secure file permissions
180
-
181
-
**Token Validator** (`pkg/auth/atproto/validator.go`)
182
-
- `ValidateToken()` - Validate token via PDS getSession
183
-
- `ValidateTokenWithResolver()` - Auto-resolve PDS from handle
184
-
- Returns validated DID and handle
185
-
- Used by registry to verify OAuth tokens
186
-
187
-
## Testing
188
-
189
-
### Manual Testing
190
-
191
-
1. Configure the helper:
192
-
```bash
193
-
./docker-credential-atcr configure
194
-
# Enter handle: alice.bsky.social
195
-
# Browser opens for authorization
196
-
# Token saved to ~/.atcr/oauth-token.json
197
-
```
198
-
199
-
2. Test credential retrieval:
200
-
```bash
201
-
echo '{"ServerURL": "atcr.io"}' | ./docker-credential-atcr get
202
-
# Should return: {"Username":"oauth2","Secret":"<jwt>"}
203
-
```
204
-
205
-
3. Test with Docker:
206
-
```bash
207
-
docker push atcr.io/alice/test:latest
208
-
```
209
-
210
-
### Integration Testing
211
-
212
-
TODO: Add automated tests for:
213
-
- OAuth flow with mock PDS
214
-
- DPoP proof generation
215
-
- Token exchange
216
-
- Credential helper protocol
217
-
218
-
## Security Features
219
-
220
-
### OAuth Token Validation
221
-
222
-
The registry validates ATProto OAuth tokens by calling `com.atproto.server.getSession` on the user's PDS. This ensures:
223
-
- Token is valid and not expired
224
-
- Token belongs to the claimed user
225
-
- User's DID and handle are extracted from the PDS response
226
-
- No trust in client-provided identity information
227
-
228
-
**Flow:**
229
-
1. Client sends OAuth token + handle to `/auth/exchange`
230
-
2. Registry resolves handle → PDS endpoint
231
-
3. Registry calls `{pds}/xrpc/com.atproto.server.getSession` with token
232
-
4. PDS validates token and returns session info (DID, handle)
233
-
5. Registry uses validated DID/handle to issue registry JWT
234
-
235
-
## Future Improvements
236
-
237
-
1. **Token refresh in background**
238
-
- Proactively refresh before expiry
239
-
- Reduce latency on Docker commands
240
-
241
-
3. **Multiple account support**
242
-
- Store tokens for multiple handles
243
-
- Allow selecting which account to use
244
-
245
-
4. **Revocation support**
246
-
- Implement token revocation
247
-
- Clean up on logout
248
-
249
-
5. **Better error messages**
250
-
- User-friendly OAuth error handling
251
-
- Guide users through common issues
252
-
253
-
## Troubleshooting
254
-
255
-
### "Failed to resolve identity"
256
-
- Check internet connection
257
-
- Verify handle is correct (e.g., `alice.bsky.social`)
258
-
- Ensure PDS is accessible
259
-
260
-
### "Authorization timed out"
261
-
- Complete authorization within 5 minutes
262
-
- Check if browser opened correctly
263
-
- Try running `configure` again
264
-
265
-
### "Token expired"
266
-
- Credential helper should auto-refresh
267
-
- If persistent, run `configure` again
268
-
- Check `~/.atcr/oauth-token.json` permissions
269
-
270
-
### "Failed to exchange token"
271
-
- Ensure registry is running
272
-
- Check `/auth/exchange` endpoint is accessible
273
-
- Verify token hasn't been revoked
274
-
275
-
## References
276
-
277
-
- [ATProto OAuth Specification](https://atproto.com/specs/oauth)
278
-
- [RFC 9449: DPoP](https://datatracker.ietf.org/doc/html/rfc9449)
279
-
- [RFC 9126: PAR](https://datatracker.ietf.org/doc/html/rfc9126)
280
-
- [RFC 7636: PKCE](https://datatracker.ietf.org/doc/html/rfc7636)
281
-
- [Docker Credential Helpers](https://github.com/docker/docker-credential-helpers)
+1289
docs/QUOTAS.md
+1289
docs/QUOTAS.md
···
1
+
# ATCR Quota System
2
+
3
+
This document describes ATCR's storage quota implementation, inspired by Harbor's proven approach to per-project blob tracking with deduplication.
4
+
5
+
## Table of Contents
6
+
7
+
- [Overview](#overview)
8
+
- [Harbor's Approach (Reference Implementation)](#harbors-approach-reference-implementation)
9
+
- [Storage Options](#storage-options)
10
+
- [Quota Data Model](#quota-data-model)
11
+
- [Push Flow (Detailed)](#push-flow-detailed)
12
+
- [Delete Flow](#delete-flow)
13
+
- [Garbage Collection](#garbage-collection)
14
+
- [Quota Reconciliation](#quota-reconciliation)
15
+
- [Configuration](#configuration)
16
+
- [Trade-offs & Design Decisions](#trade-offs--design-decisions)
17
+
- [Future Enhancements](#future-enhancements)
18
+
19
+
## Overview
20
+
21
+
ATCR implements per-user storage quotas to:
22
+
1. **Limit storage consumption** on shared hold services
23
+
2. **Track actual S3 costs** (what new data was added)
24
+
3. **Benefit from deduplication** (users only pay once per layer)
25
+
4. **Provide transparency** (show users their storage usage)
26
+
27
+
**Key principle:** Users pay for layers they've uploaded, but only ONCE per layer regardless of how many images reference it.
28
+
29
+
### Example Scenario
30
+
31
+
```
32
+
Alice pushes myapp:v1 (layers A, B, C - each 100MB)
33
+
→ Alice's quota: +300MB (all new layers)
34
+
35
+
Alice pushes myapp:v2 (layers A, B, D)
36
+
→ Layers A, B already claimed by Alice
37
+
→ Layer D is new (100MB)
38
+
→ Alice's quota: +100MB (only D is new)
39
+
→ Total: 400MB
40
+
41
+
Bob pushes his-app:latest (layers A, E)
42
+
→ Layer A already exists in S3 (uploaded by Alice)
43
+
→ Bob claims it for first time → +100MB to Bob's quota
44
+
→ Layer E is new → +100MB to Bob's quota
45
+
→ Bob's quota: 200MB
46
+
47
+
Physical S3 storage: 500MB (A, B, C, D, E)
48
+
Claimed storage: 600MB (Alice: 400MB, Bob: 200MB)
49
+
Deduplication savings: 100MB (layer A shared)
50
+
```
51
+
52
+
## Harbor's Approach (Reference Implementation)
53
+
54
+
Harbor is built on distribution/distribution (same as ATCR) and implements quotas as middleware. Their approach:
55
+
56
+
### Key Insights from Harbor
57
+
58
+
1. **"Shared blobs are only computed once per project"**
59
+
- Each project tracks which blobs it has uploaded
60
+
- Same blob used in multiple images counts only once per project
61
+
- Different projects claiming the same blob each pay for it
62
+
63
+
2. **Quota checked when manifest is pushed**
64
+
- Blobs upload first (presigned URLs, can't intercept)
65
+
- Manifest pushed last → quota check happens here
66
+
- Can reject manifest if quota exceeded (orphaned blobs cleaned by GC)
67
+
68
+
3. **Middleware-based implementation**
69
+
- distribution/distribution has NO built-in quota support
70
+
- Harbor added it as request preprocessing middleware
71
+
- Uses database (PostgreSQL) or Redis for quota storage
72
+
73
+
4. **Per-project ownership model**
74
+
- Blobs are physically deduplicated globally
75
+
- Quota accounting is logical (per-project claims)
76
+
- Total claimed storage can exceed physical storage
77
+
78
+
### References
79
+
80
+
- Harbor Quota Documentation: https://goharbor.io/docs/1.10/administration/configure-project-quotas/
81
+
- Harbor Source: https://github.com/goharbor/harbor (see `src/controller/quota`)
82
+
83
+
## Storage Options
84
+
85
+
The hold service needs to store quota data somewhere. Two options:
86
+
87
+
### Option 1: S3-Based Storage (Recommended for BYOS)
88
+
89
+
Store quota metadata alongside blobs in the same S3 bucket:
90
+
91
+
```
92
+
Bucket structure:
93
+
/docker/registry/v2/blobs/sha256/ab/abc123.../data ← actual blobs
94
+
/atcr/quota/did:plc:alice.json ← quota tracking
95
+
/atcr/quota/did:plc:bob.json
96
+
```
97
+
98
+
**Pros:**
99
+
- ✅ No separate database needed
100
+
- ✅ Single S3 bucket (better UX - no second bucket to configure)
101
+
- ✅ Quota data lives with the blobs
102
+
- ✅ Hold service stays relatively stateless
103
+
- ✅ Works with any S3-compatible service (Storj, Minio, Upcloud, Fly.io)
104
+
105
+
**Cons:**
106
+
- ❌ Slower than local database (network round-trip)
107
+
- ❌ Eventual consistency issues
108
+
- ❌ Race conditions on concurrent updates
109
+
- ❌ Extra S3 API costs (GET/PUT per upload)
110
+
111
+
**Performance:**
112
+
- Each blob upload: 1 HEAD (blob exists?) + 1 GET (quota) + 1 PUT (update quota)
113
+
- Typical latency: 100-200ms total overhead
114
+
- For high-throughput registries, consider SQLite
115
+
116
+
### Option 2: SQLite Database (Recommended for Shared Holds)
117
+
118
+
Local database in hold service:
119
+
120
+
```bash
121
+
/var/lib/atcr/hold-quota.db
122
+
```
123
+
124
+
**Pros:**
125
+
- ✅ Fast local queries (no network latency)
126
+
- ✅ ACID transactions (no race conditions)
127
+
- ✅ Efficient for high-throughput registries
128
+
- ✅ Can use foreign keys and joins
129
+
130
+
**Cons:**
131
+
- ❌ Makes hold service stateful (persistent volume needed)
132
+
- ❌ Not ideal for ephemeral BYOS deployments
133
+
- ❌ Backup/restore complexity
134
+
- ❌ Multi-instance scaling requires shared database
135
+
136
+
**Schema:**
137
+
```sql
138
+
CREATE TABLE user_quotas (
139
+
did TEXT PRIMARY KEY,
140
+
quota_limit INTEGER NOT NULL DEFAULT 10737418240, -- 10GB
141
+
quota_used INTEGER NOT NULL DEFAULT 0,
142
+
updated_at TIMESTAMP
143
+
);
144
+
145
+
CREATE TABLE claimed_layers (
146
+
did TEXT NOT NULL,
147
+
digest TEXT NOT NULL,
148
+
size INTEGER NOT NULL,
149
+
claimed_at TIMESTAMP,
150
+
PRIMARY KEY(did, digest)
151
+
);
152
+
```
153
+
154
+
### Recommendation
155
+
156
+
- **BYOS (user-owned holds):** S3-based (keeps hold service ephemeral)
157
+
- **Shared holds (multi-user):** SQLite (better performance and consistency)
158
+
- **High-traffic production:** SQLite or PostgreSQL (Harbor uses this)
159
+
160
+
## Quota Data Model
161
+
162
+
### Quota File Format (S3-based)
163
+
164
+
```json
165
+
{
166
+
"did": "did:plc:alice123",
167
+
"limit": 10737418240,
168
+
"used": 5368709120,
169
+
"claimed_layers": {
170
+
"sha256:abc123...": 104857600,
171
+
"sha256:def456...": 52428800,
172
+
"sha256:789ghi...": 209715200
173
+
},
174
+
"last_updated": "2025-10-09T12:34:56Z",
175
+
"version": 1
176
+
}
177
+
```
178
+
179
+
**Fields:**
180
+
- `did`: User's ATProto DID
181
+
- `limit`: Maximum storage in bytes (default: 10GB)
182
+
- `used`: Current storage usage in bytes (sum of claimed_layers)
183
+
- `claimed_layers`: Map of digest → size for all layers user has uploaded
184
+
- `last_updated`: Timestamp of last quota update
185
+
- `version`: Schema version for future migrations
186
+
187
+
### Why Track Individual Layers?
188
+
189
+
**Q: Can't we just track a counter?**
190
+
191
+
**A: We need layer tracking for:**
192
+
193
+
1. **Deduplication detection**
194
+
- Check if user already claimed a layer → free upload
195
+
- Example: Updating an image reuses most layers
196
+
197
+
2. **Accurate deletes**
198
+
- When manifest deleted, only decrement unclaimed layers
199
+
- User may have 5 images sharing layer A - deleting 1 image doesn't free layer A
200
+
201
+
3. **Quota reconciliation**
202
+
- Verify quota matches reality by listing user's manifests
203
+
- Recalculate from layers in manifests vs claimed_layers map
204
+
205
+
4. **Auditing**
206
+
- "Show me what I'm storing"
207
+
- Users can see which layers consume their quota
208
+
209
+
## Push Flow (Detailed)
210
+
211
+
### Step-by-Step: User Pushes Image
212
+
213
+
```
214
+
┌──────────┐ ┌──────────┐ ┌──────────┐
215
+
│ Client │ │ Hold │ │ S3 │
216
+
│ (Docker) │ │ Service │ │ Bucket │
217
+
└──────────┘ └──────────┘ └──────────┘
218
+
│ │ │
219
+
│ 1. PUT /v2/.../blobs/ │ │
220
+
│ upload?digest=sha256:abc│ │
221
+
├───────────────────────────>│ │
222
+
│ │ │
223
+
│ │ 2. Check if blob exists │
224
+
│ │ (Stat/HEAD request) │
225
+
│ ├───────────────────────────>│
226
+
│ │<───────────────────────────┤
227
+
│ │ 200 OK (exists) or │
228
+
│ │ 404 Not Found │
229
+
│ │ │
230
+
│ │ 3. Read user quota │
231
+
│ │ GET /atcr/quota/{did} │
232
+
│ ├───────────────────────────>│
233
+
│ │<───────────────────────────┤
234
+
│ │ quota.json │
235
+
│ │ │
236
+
│ │ 4. Calculate quota impact │
237
+
│ │ - If digest in │
238
+
│ │ claimed_layers: 0 │
239
+
│ │ - Else: size │
240
+
│ │ │
241
+
│ │ 5. Check quota limit │
242
+
│ │ used + impact <= limit? │
243
+
│ │ │
244
+
│ │ 6. Update quota │
245
+
│ │ PUT /atcr/quota/{did} │
246
+
│ ├───────────────────────────>│
247
+
│ │<───────────────────────────┤
248
+
│ │ 200 OK │
249
+
│ │ │
250
+
│ 7. Presigned URL │ │
251
+
│<───────────────────────────┤ │
252
+
│ {url: "https://s3..."} │ │
253
+
│ │ │
254
+
│ 8. Upload blob to S3 │ │
255
+
├────────────────────────────┼───────────────────────────>│
256
+
│ │ │
257
+
│ 9. 200 OK │ │
258
+
│<───────────────────────────┼────────────────────────────┤
259
+
│ │ │
260
+
```
261
+
262
+
### Implementation (Pseudocode)
263
+
264
+
```go
265
+
// cmd/hold/main.go - HandlePutPresignedURL
266
+
267
+
func (s *HoldService) HandlePutPresignedURL(w http.ResponseWriter, r *http.Request) {
268
+
var req PutPresignedURLRequest
269
+
json.NewDecoder(r.Body).Decode(&req)
270
+
271
+
// Step 1: Check if blob already exists in S3
272
+
blobPath := fmt.Sprintf("/docker/registry/v2/blobs/%s/%s/%s/data",
273
+
algorithm, digest[:2], digest)
274
+
275
+
_, err := s.driver.Stat(ctx, blobPath)
276
+
blobExists := (err == nil)
277
+
278
+
// Step 2: Read quota from S3 (or SQLite)
279
+
quota, err := s.quotaManager.GetQuota(req.DID)
280
+
if err != nil {
281
+
// First upload - create quota with defaults
282
+
quota = &Quota{
283
+
DID: req.DID,
284
+
Limit: s.config.QuotaDefaultLimit,
285
+
Used: 0,
286
+
ClaimedLayers: make(map[string]int64),
287
+
}
288
+
}
289
+
290
+
// Step 3: Calculate quota impact
291
+
quotaImpact := req.Size // Default: assume new layer
292
+
293
+
if _, alreadyClaimed := quota.ClaimedLayers[req.Digest]; alreadyClaimed {
294
+
// User already uploaded this layer before
295
+
quotaImpact = 0
296
+
log.Printf("Layer %s already claimed by %s, no quota impact",
297
+
req.Digest, req.DID)
298
+
} else if blobExists {
299
+
// Blob exists in S3 (uploaded by another user)
300
+
// But this user is claiming it for first time
301
+
// Still counts against their quota
302
+
log.Printf("Layer %s exists globally but new to %s, quota impact: %d",
303
+
req.Digest, req.DID, quotaImpact)
304
+
} else {
305
+
// Brand new blob - will be uploaded to S3
306
+
log.Printf("New layer %s for %s, quota impact: %d",
307
+
req.Digest, req.DID, quotaImpact)
308
+
}
309
+
310
+
// Step 4: Check quota limit
311
+
if quota.Used + quotaImpact > quota.Limit {
312
+
http.Error(w, fmt.Sprintf(
313
+
"quota exceeded: used=%d, impact=%d, limit=%d",
314
+
quota.Used, quotaImpact, quota.Limit,
315
+
), http.StatusPaymentRequired) // 402
316
+
return
317
+
}
318
+
319
+
// Step 5: Update quota (optimistic - before upload completes)
320
+
quota.Used += quotaImpact
321
+
if quotaImpact > 0 {
322
+
quota.ClaimedLayers[req.Digest] = req.Size
323
+
}
324
+
quota.LastUpdated = time.Now()
325
+
326
+
if err := s.quotaManager.SaveQuota(quota); err != nil {
327
+
http.Error(w, "failed to update quota", http.StatusInternalServerError)
328
+
return
329
+
}
330
+
331
+
// Step 6: Generate presigned URL
332
+
presignedURL, err := s.getUploadURL(ctx, req.Digest, req.Size, req.DID)
333
+
if err != nil {
334
+
// Rollback quota update on error
335
+
quota.Used -= quotaImpact
336
+
delete(quota.ClaimedLayers, req.Digest)
337
+
s.quotaManager.SaveQuota(quota)
338
+
339
+
http.Error(w, "failed to generate presigned URL", http.StatusInternalServerError)
340
+
return
341
+
}
342
+
343
+
// Step 7: Return presigned URL + quota info
344
+
resp := PutPresignedURLResponse{
345
+
URL: presignedURL,
346
+
ExpiresAt: time.Now().Add(15 * time.Minute),
347
+
QuotaInfo: QuotaInfo{
348
+
Used: quota.Used,
349
+
Limit: quota.Limit,
350
+
Available: quota.Limit - quota.Used,
351
+
Impact: quotaImpact,
352
+
AlreadyClaimed: quotaImpact == 0,
353
+
},
354
+
}
355
+
356
+
w.Header().Set("Content-Type", "application/json")
357
+
json.NewEncoder(w).Encode(resp)
358
+
}
359
+
```
360
+
361
+
### Race Condition Handling
362
+
363
+
**Problem:** Two concurrent uploads of the same blob
364
+
365
+
```
366
+
Time User A User B
367
+
0ms Upload layer X (100MB)
368
+
10ms Upload layer X (100MB)
369
+
20ms Check exists: NO Check exists: NO
370
+
30ms Quota impact: 100MB Quota impact: 100MB
371
+
40ms Update quota A: +100MB Update quota B: +100MB
372
+
50ms Generate presigned URL Generate presigned URL
373
+
100ms Upload to S3 completes Upload to S3 (overwrites A's)
374
+
```
375
+
376
+
**Result:** Both users charged 100MB, but only 100MB stored in S3.
377
+
378
+
**Mitigation strategies:**
379
+
380
+
1. **Accept eventual consistency** (recommended for S3-based)
381
+
- Run periodic reconciliation to fix discrepancies
382
+
- Small inconsistency window (minutes) is acceptable
383
+
- Reconciliation uses PDS as source of truth
384
+
385
+
2. **Optimistic locking** (S3 ETags)
386
+
```go
387
+
// Use S3 ETags for conditional writes
388
+
oldETag := getQuotaFileETag(did)
389
+
err := putQuotaFileWithCondition(quota, oldETag)
390
+
if err == PreconditionFailed {
391
+
// Retry with fresh read
392
+
}
393
+
```
394
+
395
+
3. **Database transactions** (SQLite-based)
396
+
```sql
397
+
BEGIN TRANSACTION;
398
+
SELECT * FROM user_quotas WHERE did = ? FOR UPDATE;
399
+
UPDATE user_quotas SET used = used + ? WHERE did = ?;
400
+
COMMIT;
401
+
```
402
+
403
+
## Delete Flow
404
+
405
+
### Manifest Deletion via AppView UI
406
+
407
+
When a user deletes a manifest through the AppView web interface:
408
+
409
+
```
410
+
┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐
411
+
│ User │ │ AppView │ │ Hold │ │ PDS │
412
+
│ UI │ │ Database │ │ Service │ │ │
413
+
└──────────┘ └──────────┘ └──────────┘ └──────────┘
414
+
│ │ │ │
415
+
│ DELETE manifest │ │ │
416
+
├─────────────────────>│ │ │
417
+
│ │ │ │
418
+
│ │ 1. Get manifest │ │
419
+
│ │ and layers │ │
420
+
│ │ │ │
421
+
│ │ 2. Check which │ │
422
+
│ │ layers still │ │
423
+
│ │ referenced by │ │
424
+
│ │ user's other │ │
425
+
│ │ manifests │ │
426
+
│ │ │ │
427
+
│ │ 3. DELETE manifest │ │
428
+
│ │ from PDS │ │
429
+
│ ├──────────────────────┼─────────────────────>│
430
+
│ │ │ │
431
+
│ │ 4. POST /quota/decrement │
432
+
│ ├─────────────────────>│ │
433
+
│ │ {layers: [...]} │ │
434
+
│ │ │ │
435
+
│ │ │ 5. Update quota │
436
+
│ │ │ Remove unclaimed │
437
+
│ │ │ layers │
438
+
│ │ │ │
439
+
│ │ 6. 200 OK │ │
440
+
│ │<─────────────────────┤ │
441
+
│ │ │ │
442
+
│ │ 7. Delete from DB │ │
443
+
│ │ │ │
444
+
│ 8. Success │ │ │
445
+
│<─────────────────────┤ │ │
446
+
│ │ │ │
447
+
```
448
+
449
+
### AppView Implementation
450
+
451
+
```go
452
+
// pkg/appview/handlers/manifest.go
453
+
454
+
func (h *ManifestHandler) DeleteManifest(w http.ResponseWriter, r *http.Request) {
455
+
did := r.Context().Value("auth.did").(string)
456
+
repository := chi.URLParam(r, "repository")
457
+
digest := chi.URLParam(r, "digest")
458
+
459
+
// Step 1: Get manifest and its layers from database
460
+
manifest, err := db.GetManifest(h.db, digest)
461
+
if err != nil {
462
+
http.Error(w, "manifest not found", 404)
463
+
return
464
+
}
465
+
466
+
layers, err := db.GetLayersForManifest(h.db, manifest.ID)
467
+
if err != nil {
468
+
http.Error(w, "failed to get layers", 500)
469
+
return
470
+
}
471
+
472
+
// Step 2: For each layer, check if user still references it
473
+
// in other manifests
474
+
layersToDecrement := []LayerInfo{}
475
+
476
+
for _, layer := range layers {
477
+
// Query: does this user have other manifests using this layer?
478
+
stillReferenced, err := db.CheckLayerReferencedByUser(
479
+
h.db, did, repository, layer.Digest, manifest.ID,
480
+
)
481
+
482
+
if err != nil {
483
+
http.Error(w, "failed to check layer references", 500)
484
+
return
485
+
}
486
+
487
+
if !stillReferenced {
488
+
// This layer is no longer used by user
489
+
layersToDecrement = append(layersToDecrement, LayerInfo{
490
+
Digest: layer.Digest,
491
+
Size: layer.Size,
492
+
})
493
+
}
494
+
}
495
+
496
+
// Step 3: Delete manifest from user's PDS
497
+
atprotoClient := atproto.NewClient(manifest.PDSEndpoint, did, accessToken)
498
+
err = atprotoClient.DeleteRecord(ctx, atproto.ManifestCollection, manifestRKey)
499
+
if err != nil {
500
+
http.Error(w, "failed to delete from PDS", 500)
501
+
return
502
+
}
503
+
504
+
// Step 4: Notify hold service to decrement quota
505
+
if len(layersToDecrement) > 0 {
506
+
holdClient := &http.Client{}
507
+
508
+
decrementReq := QuotaDecrementRequest{
509
+
DID: did,
510
+
Layers: layersToDecrement,
511
+
}
512
+
513
+
body, _ := json.Marshal(decrementReq)
514
+
resp, err := holdClient.Post(
515
+
manifest.HoldEndpoint + "/quota/decrement",
516
+
"application/json",
517
+
bytes.NewReader(body),
518
+
)
519
+
520
+
if err != nil || resp.StatusCode != 200 {
521
+
log.Printf("Warning: failed to update quota on hold service: %v", err)
522
+
// Continue anyway - GC reconciliation will fix it
523
+
}
524
+
}
525
+
526
+
// Step 5: Delete from AppView database
527
+
err = db.DeleteManifest(h.db, did, repository, digest)
528
+
if err != nil {
529
+
http.Error(w, "failed to delete from database", 500)
530
+
return
531
+
}
532
+
533
+
w.WriteHeader(http.StatusNoContent)
534
+
}
535
+
```
536
+
537
+
### Hold Service Decrement Endpoint
538
+
539
+
```go
540
+
// cmd/hold/main.go
541
+
542
+
type QuotaDecrementRequest struct {
543
+
DID string `json:"did"`
544
+
Layers []LayerInfo `json:"layers"`
545
+
}
546
+
547
+
type LayerInfo struct {
548
+
Digest string `json:"digest"`
549
+
Size int64 `json:"size"`
550
+
}
551
+
552
+
func (s *HoldService) HandleQuotaDecrement(w http.ResponseWriter, r *http.Request) {
553
+
var req QuotaDecrementRequest
554
+
if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
555
+
http.Error(w, "invalid request", 400)
556
+
return
557
+
}
558
+
559
+
// Read current quota
560
+
quota, err := s.quotaManager.GetQuota(req.DID)
561
+
if err != nil {
562
+
http.Error(w, "quota not found", 404)
563
+
return
564
+
}
565
+
566
+
// Decrement quota for each layer
567
+
for _, layer := range req.Layers {
568
+
if size, claimed := quota.ClaimedLayers[layer.Digest]; claimed {
569
+
// Remove from claimed layers
570
+
delete(quota.ClaimedLayers, layer.Digest)
571
+
quota.Used -= size
572
+
573
+
log.Printf("Decremented quota for %s: layer %s (%d bytes)",
574
+
req.DID, layer.Digest, size)
575
+
} else {
576
+
log.Printf("Warning: layer %s not in claimed_layers for %s",
577
+
layer.Digest, req.DID)
578
+
}
579
+
}
580
+
581
+
// Ensure quota.Used doesn't go negative (defensive)
582
+
if quota.Used < 0 {
583
+
log.Printf("Warning: quota.Used went negative for %s, resetting to 0", req.DID)
584
+
quota.Used = 0
585
+
}
586
+
587
+
// Save updated quota
588
+
quota.LastUpdated = time.Now()
589
+
if err := s.quotaManager.SaveQuota(quota); err != nil {
590
+
http.Error(w, "failed to save quota", 500)
591
+
return
592
+
}
593
+
594
+
// Return updated quota info
595
+
json.NewEncoder(w).Encode(map[string]any{
596
+
"used": quota.Used,
597
+
"limit": quota.Limit,
598
+
})
599
+
}
600
+
```
601
+
602
+
### SQL Query: Check Layer References
603
+
604
+
```sql
605
+
-- pkg/appview/db/queries.go
606
+
607
+
-- Check if user still references this layer in other manifests
608
+
SELECT COUNT(*)
609
+
FROM layers l
610
+
JOIN manifests m ON l.manifest_id = m.id
611
+
WHERE m.did = ? -- User's DID
612
+
AND l.digest = ? -- Layer digest
613
+
AND m.id != ? -- Exclude the manifest being deleted
614
+
```
615
+
616
+
## Garbage Collection
617
+
618
+
### Background: Orphaned Blobs
619
+
620
+
Orphaned blobs accumulate when:
621
+
1. Manifest push fails after blobs uploaded (presigned URLs bypass hold)
622
+
2. Quota exceeded - manifest rejected, blobs already in S3
623
+
3. User deletes manifest - blobs no longer referenced
624
+
625
+
**GC periodically cleans these up.**
626
+
627
+
### GC Cron Implementation
628
+
629
+
Similar to AppView's backfill worker, the hold service can run periodic GC:
630
+
631
+
```go
632
+
// cmd/hold/gc/gc.go
633
+
634
+
type GarbageCollector struct {
635
+
driver storagedriver.StorageDriver
636
+
appviewURL string
637
+
holdURL string
638
+
quotaManager *quota.Manager
639
+
}
640
+
641
+
// Run garbage collection
642
+
func (gc *GarbageCollector) Run(ctx context.Context) error {
643
+
log.Println("Starting garbage collection...")
644
+
645
+
// Step 1: Get list of referenced blobs from AppView
646
+
referenced, err := gc.getReferencedBlobs()
647
+
if err != nil {
648
+
return fmt.Errorf("failed to get referenced blobs: %w", err)
649
+
}
650
+
651
+
referencedSet := make(map[string]bool)
652
+
for _, digest := range referenced {
653
+
referencedSet[digest] = true
654
+
}
655
+
656
+
log.Printf("AppView reports %d referenced blobs", len(referenced))
657
+
658
+
// Step 2: Walk S3 blobs
659
+
deletedCount := 0
660
+
reclaimedBytes := int64(0)
661
+
662
+
err = gc.driver.Walk(ctx, "/docker/registry/v2/blobs", func(fileInfo storagedriver.FileInfo) error {
663
+
if fileInfo.IsDir() {
664
+
return nil // Skip directories
665
+
}
666
+
667
+
// Extract digest from path
668
+
// Path: /docker/registry/v2/blobs/sha256/ab/abc123.../data
669
+
digest := extractDigestFromPath(fileInfo.Path())
670
+
671
+
if !referencedSet[digest] {
672
+
// Unreferenced blob - delete it
673
+
size := fileInfo.Size()
674
+
675
+
if err := gc.driver.Delete(ctx, fileInfo.Path()); err != nil {
676
+
log.Printf("Failed to delete blob %s: %v", digest, err)
677
+
return nil // Continue anyway
678
+
}
679
+
680
+
deletedCount++
681
+
reclaimedBytes += size
682
+
683
+
log.Printf("GC: Deleted unreferenced blob %s (%d bytes)", digest, size)
684
+
}
685
+
686
+
return nil
687
+
})
688
+
689
+
if err != nil {
690
+
return fmt.Errorf("failed to walk blobs: %w", err)
691
+
}
692
+
693
+
log.Printf("GC complete: deleted %d blobs, reclaimed %d bytes",
694
+
deletedCount, reclaimedBytes)
695
+
696
+
return nil
697
+
}
698
+
699
+
// Get referenced blobs from AppView
700
+
func (gc *GarbageCollector) getReferencedBlobs() ([]string, error) {
701
+
// Query AppView for all blobs referenced by manifests
702
+
// stored in THIS hold service
703
+
url := fmt.Sprintf("%s/internal/blobs/referenced?hold=%s",
704
+
gc.appviewURL, url.QueryEscape(gc.holdURL))
705
+
706
+
resp, err := http.Get(url)
707
+
if err != nil {
708
+
return nil, err
709
+
}
710
+
defer resp.Body.Close()
711
+
712
+
var result struct {
713
+
Blobs []string `json:"blobs"`
714
+
}
715
+
716
+
if err := json.NewDecoder(resp.Body).Decode(&result); err != nil {
717
+
return nil, err
718
+
}
719
+
720
+
return result.Blobs, nil
721
+
}
722
+
```
723
+
724
+
### AppView Internal API
725
+
726
+
```go
727
+
// pkg/appview/handlers/internal.go
728
+
729
+
// Get all referenced blobs for a specific hold
730
+
func (h *InternalHandler) GetReferencedBlobs(w http.ResponseWriter, r *http.Request) {
731
+
holdEndpoint := r.URL.Query().Get("hold")
732
+
if holdEndpoint == "" {
733
+
http.Error(w, "missing hold parameter", 400)
734
+
return
735
+
}
736
+
737
+
// Query database for all layers in manifests stored in this hold
738
+
query := `
739
+
SELECT DISTINCT l.digest
740
+
FROM layers l
741
+
JOIN manifests m ON l.manifest_id = m.id
742
+
WHERE m.hold_endpoint = ?
743
+
`
744
+
745
+
rows, err := h.db.Query(query, holdEndpoint)
746
+
if err != nil {
747
+
http.Error(w, "database error", 500)
748
+
return
749
+
}
750
+
defer rows.Close()
751
+
752
+
blobs := []string{}
753
+
for rows.Next() {
754
+
var digest string
755
+
if err := rows.Scan(&digest); err != nil {
756
+
continue
757
+
}
758
+
blobs = append(blobs, digest)
759
+
}
760
+
761
+
json.NewEncoder(w).Encode(map[string]any{
762
+
"blobs": blobs,
763
+
"count": len(blobs),
764
+
"hold": holdEndpoint,
765
+
})
766
+
}
767
+
```
768
+
769
+
### GC Cron Schedule
770
+
771
+
```go
772
+
// cmd/hold/main.go
773
+
774
+
func main() {
775
+
// ... service setup ...
776
+
777
+
// Start GC cron if enabled
778
+
if os.Getenv("GC_ENABLED") == "true" {
779
+
gcInterval := 24 * time.Hour // Daily by default
780
+
781
+
go func() {
782
+
ticker := time.NewTicker(gcInterval)
783
+
defer ticker.Stop()
784
+
785
+
for range ticker.C {
786
+
if err := garbageCollector.Run(context.Background()); err != nil {
787
+
log.Printf("GC error: %v", err)
788
+
}
789
+
}
790
+
}()
791
+
792
+
log.Printf("GC cron started: runs every %v", gcInterval)
793
+
}
794
+
795
+
// Start server...
796
+
}
797
+
```
798
+
799
+
## Quota Reconciliation
800
+
801
+
### PDS as Source of Truth
802
+
803
+
**Key insight:** Manifest records in PDS are publicly readable (no OAuth needed for reads).
804
+
805
+
Each manifest contains:
806
+
- Repository name
807
+
- Digest
808
+
- Layers array with digest + size
809
+
- Hold endpoint
810
+
811
+
The hold service can query the PDS to calculate the user's true quota:
812
+
813
+
```
814
+
1. List all io.atcr.manifest records for user
815
+
2. Filter manifests where holdEndpoint == this hold service
816
+
3. Extract unique layers (deduplicate by digest)
817
+
4. Sum layer sizes = true quota usage
818
+
5. Compare to quota file
819
+
6. Fix discrepancies
820
+
```
821
+
822
+
### Implementation
823
+
824
+
```go
825
+
// cmd/hold/quota/reconcile.go
826
+
827
+
type Reconciler struct {
828
+
quotaManager *Manager
829
+
atprotoResolver *atproto.Resolver
830
+
holdURL string
831
+
}
832
+
833
+
// ReconcileUser recalculates quota from PDS manifests
834
+
func (r *Reconciler) ReconcileUser(ctx context.Context, did string) error {
835
+
log.Printf("Reconciling quota for %s", did)
836
+
837
+
// Step 1: Resolve user's PDS endpoint
838
+
identity, err := r.atprotoResolver.ResolveIdentity(ctx, did)
839
+
if err != nil {
840
+
return fmt.Errorf("failed to resolve DID: %w", err)
841
+
}
842
+
843
+
// Step 2: Create unauthenticated ATProto client
844
+
// (manifest records are public - no OAuth needed)
845
+
client := atproto.NewClient(identity.PDSEndpoint, did, "")
846
+
847
+
// Step 3: List all manifest records for this user
848
+
manifests, err := client.ListRecords(ctx, atproto.ManifestCollection, 1000)
849
+
if err != nil {
850
+
return fmt.Errorf("failed to list manifests: %w", err)
851
+
}
852
+
853
+
// Step 4: Filter manifests stored in THIS hold service
854
+
// and extract unique layers
855
+
uniqueLayers := make(map[string]int64) // digest -> size
856
+
857
+
for _, record := range manifests {
858
+
var manifest atproto.ManifestRecord
859
+
if err := json.Unmarshal(record.Value, &manifest); err != nil {
860
+
log.Printf("Warning: failed to parse manifest: %v", err)
861
+
continue
862
+
}
863
+
864
+
// Only count manifests stored in this hold
865
+
if manifest.HoldEndpoint != r.holdURL {
866
+
continue
867
+
}
868
+
869
+
// Add config blob
870
+
if manifest.Config.Digest != "" {
871
+
uniqueLayers[manifest.Config.Digest] = manifest.Config.Size
872
+
}
873
+
874
+
// Add layer blobs
875
+
for _, layer := range manifest.Layers {
876
+
uniqueLayers[layer.Digest] = layer.Size
877
+
}
878
+
}
879
+
880
+
// Step 5: Calculate true quota usage
881
+
trueUsage := int64(0)
882
+
for _, size := range uniqueLayers {
883
+
trueUsage += size
884
+
}
885
+
886
+
log.Printf("User %s true usage from PDS: %d bytes (%d unique layers)",
887
+
did, trueUsage, len(uniqueLayers))
888
+
889
+
// Step 6: Compare with current quota file
890
+
quota, err := r.quotaManager.GetQuota(did)
891
+
if err != nil {
892
+
log.Printf("No existing quota for %s, creating new", did)
893
+
quota = &Quota{
894
+
DID: did,
895
+
Limit: r.quotaManager.DefaultLimit,
896
+
ClaimedLayers: make(map[string]int64),
897
+
}
898
+
}
899
+
900
+
// Step 7: Fix discrepancies
901
+
if quota.Used != trueUsage || len(quota.ClaimedLayers) != len(uniqueLayers) {
902
+
log.Printf("Quota mismatch for %s: recorded=%d, actual=%d (diff=%d)",
903
+
did, quota.Used, trueUsage, trueUsage - quota.Used)
904
+
905
+
// Update quota to match PDS truth
906
+
quota.Used = trueUsage
907
+
quota.ClaimedLayers = uniqueLayers
908
+
quota.LastUpdated = time.Now()
909
+
910
+
if err := r.quotaManager.SaveQuota(quota); err != nil {
911
+
return fmt.Errorf("failed to save reconciled quota: %w", err)
912
+
}
913
+
914
+
log.Printf("Reconciled quota for %s: %d bytes", did, trueUsage)
915
+
} else {
916
+
log.Printf("Quota for %s is accurate", did)
917
+
}
918
+
919
+
return nil
920
+
}
921
+
922
+
// ReconcileAll reconciles all users (run periodically)
923
+
func (r *Reconciler) ReconcileAll(ctx context.Context) error {
924
+
// Get list of all users with quota files
925
+
users, err := r.quotaManager.ListUsers()
926
+
if err != nil {
927
+
return err
928
+
}
929
+
930
+
log.Printf("Starting reconciliation for %d users", len(users))
931
+
932
+
for _, did := range users {
933
+
if err := r.ReconcileUser(ctx, did); err != nil {
934
+
log.Printf("Failed to reconcile %s: %v", did, err)
935
+
// Continue with other users
936
+
}
937
+
}
938
+
939
+
log.Println("Reconciliation complete")
940
+
return nil
941
+
}
942
+
```
943
+
944
+
### Reconciliation Cron
945
+
946
+
```go
947
+
// cmd/hold/main.go
948
+
949
+
func main() {
950
+
// ... setup ...
951
+
952
+
// Start reconciliation cron
953
+
if os.Getenv("QUOTA_RECONCILE_ENABLED") == "true" {
954
+
reconcileInterval := 24 * time.Hour // Daily
955
+
956
+
go func() {
957
+
ticker := time.NewTicker(reconcileInterval)
958
+
defer ticker.Stop()
959
+
960
+
for range ticker.C {
961
+
if err := reconciler.ReconcileAll(context.Background()); err != nil {
962
+
log.Printf("Reconciliation error: %v", err)
963
+
}
964
+
}
965
+
}()
966
+
967
+
log.Printf("Quota reconciliation cron started: runs every %v", reconcileInterval)
968
+
}
969
+
970
+
// ... start server ...
971
+
}
972
+
```
973
+
974
+
### Why PDS as Source of Truth Works
975
+
976
+
1. **Manifests are canonical** - If manifest exists in PDS, user owns those layers
977
+
2. **Public reads** - No OAuth needed, just resolve DID → PDS endpoint
978
+
3. **ATProto durability** - PDS is user's authoritative data store
979
+
4. **AppView is cache** - AppView database might lag or have inconsistencies
980
+
5. **Reconciliation fixes drift** - Periodic sync from PDS ensures accuracy
981
+
982
+
**Example reconciliation scenarios:**
983
+
984
+
- **Orphaned quota entries:** User deleted manifest from PDS, but hold quota still has it
985
+
→ Reconciliation removes from claimed_layers
986
+
987
+
- **Missing quota entries:** User pushed manifest, but quota update failed
988
+
→ Reconciliation adds to claimed_layers
989
+
990
+
- **Race condition duplicates:** Two concurrent pushes double-counted a layer
991
+
→ Reconciliation fixes to actual usage
992
+
993
+
## Configuration
994
+
995
+
### Hold Service Environment Variables
996
+
997
+
```bash
998
+
# .env.hold
999
+
1000
+
# ============================================================================
1001
+
# Quota Configuration
1002
+
# ============================================================================
1003
+
1004
+
# Enable quota enforcement
1005
+
QUOTA_ENABLED=true
1006
+
1007
+
# Default quota limit per user (bytes)
1008
+
# 10GB = 10737418240
1009
+
# 50GB = 53687091200
1010
+
# 100GB = 107374182400
1011
+
QUOTA_DEFAULT_LIMIT=10737418240
1012
+
1013
+
# Storage backend for quota data
1014
+
# Options: s3, sqlite
1015
+
QUOTA_STORAGE_BACKEND=s3
1016
+
1017
+
# For S3-based storage:
1018
+
# Quota files stored in same bucket as blobs
1019
+
QUOTA_STORAGE_PREFIX=/atcr/quota/
1020
+
1021
+
# For SQLite-based storage:
1022
+
QUOTA_DB_PATH=/var/lib/atcr/hold-quota.db
1023
+
1024
+
# ============================================================================
1025
+
# Garbage Collection
1026
+
# ============================================================================
1027
+
1028
+
# Enable periodic garbage collection
1029
+
GC_ENABLED=true
1030
+
1031
+
# GC interval (default: 24h)
1032
+
GC_INTERVAL=24h
1033
+
1034
+
# AppView URL for GC reference checking
1035
+
APPVIEW_URL=https://atcr.io
1036
+
1037
+
# ============================================================================
1038
+
# Quota Reconciliation
1039
+
# ============================================================================
1040
+
1041
+
# Enable quota reconciliation from PDS
1042
+
QUOTA_RECONCILE_ENABLED=true
1043
+
1044
+
# Reconciliation interval (default: 24h)
1045
+
QUOTA_RECONCILE_INTERVAL=24h
1046
+
1047
+
# ============================================================================
1048
+
# Hold Service Identity (Required)
1049
+
# ============================================================================
1050
+
1051
+
# Public URL of this hold service
1052
+
HOLD_PUBLIC_URL=https://hold1.example.com
1053
+
1054
+
# Owner DID (for auto-registration)
1055
+
HOLD_OWNER=did:plc:xyz123
1056
+
```
1057
+
1058
+
### AppView Configuration
1059
+
1060
+
```bash
1061
+
# .env.appview
1062
+
1063
+
# Internal API endpoint for hold services
1064
+
# Used for GC reference checking
1065
+
ATCR_INTERNAL_API_ENABLED=true
1066
+
1067
+
# Optional: authentication token for internal APIs
1068
+
ATCR_INTERNAL_API_TOKEN=secret123
1069
+
```
1070
+
1071
+
## Trade-offs & Design Decisions
1072
+
1073
+
### 1. Claimed Storage vs Physical Storage
1074
+
1075
+
**Decision:** Track claimed storage (logical accounting)
1076
+
1077
+
**Why:**
1078
+
- Predictable for users: "you pay for what you upload"
1079
+
- No complex cross-user dependencies
1080
+
- Delete always gives you quota back
1081
+
- Matches Harbor's proven model
1082
+
1083
+
**Trade-off:**
1084
+
- Total claimed can exceed physical storage
1085
+
- Users might complain "I uploaded 10GB but S3 only has 6GB"
1086
+
1087
+
**Mitigation:**
1088
+
- Show deduplication savings metric
1089
+
- Educate users: "You claimed 10GB, but deduplication saved 4GB"
1090
+
1091
+
### 2. S3 vs SQLite for Quota Storage
1092
+
1093
+
**Decision:** Support both, recommend based on use case
1094
+
1095
+
**S3 Pros:**
1096
+
- No database to manage
1097
+
- Quota data lives with blobs
1098
+
- Better for ephemeral BYOS
1099
+
1100
+
**SQLite Pros:**
1101
+
- Faster (no network)
1102
+
- ACID transactions (no race conditions)
1103
+
- Better for high-traffic shared holds
1104
+
1105
+
**Trade-off:**
1106
+
- S3: eventual consistency, race conditions
1107
+
- SQLite: stateful service, scaling challenges
1108
+
1109
+
**Mitigation:**
1110
+
- Reconciliation fixes S3 inconsistencies
1111
+
- SQLite can use shared DB for multi-instance
1112
+
1113
+
### 3. Optimistic Quota Update
1114
+
1115
+
**Decision:** Update quota BEFORE upload completes
1116
+
1117
+
**Why:**
1118
+
- Prevent race conditions (two users uploading simultaneously)
1119
+
- Can reject before presigned URL generated
1120
+
- Simpler flow
1121
+
1122
+
**Trade-off:**
1123
+
- If upload fails, quota already incremented (user "paid" for nothing)
1124
+
1125
+
**Mitigation:**
1126
+
- Reconciliation from PDS fixes orphaned quota entries
1127
+
- Acceptable for MVP (upload failures are rare)
1128
+
1129
+
### 4. AppView as Intermediary
1130
+
1131
+
**Decision:** AppView notifies hold service on deletes
1132
+
1133
+
**Why:**
1134
+
- AppView already has manifest/layer database
1135
+
- Can efficiently check if layer still referenced
1136
+
- Hold service doesn't need to query PDS on every delete
1137
+
1138
+
**Trade-off:**
1139
+
- AppView → Hold dependency
1140
+
- Network hop on delete
1141
+
1142
+
**Mitigation:**
1143
+
- If notification fails, reconciliation fixes quota
1144
+
- Eventually consistent is acceptable
1145
+
1146
+
### 5. PDS as Source of Truth
1147
+
1148
+
**Decision:** Use PDS manifests for reconciliation
1149
+
1150
+
**Why:**
1151
+
- Manifests in PDS are canonical user data
1152
+
- Public reads (no OAuth for reconciliation)
1153
+
- AppView database might lag or be inconsistent
1154
+
1155
+
**Trade-off:**
1156
+
- Reconciliation requires PDS queries (slower)
1157
+
- Limited to 1000 manifests per query
1158
+
1159
+
**Mitigation:**
1160
+
- Run reconciliation daily (not real-time)
1161
+
- Paginate if user has >1000 manifests
1162
+
1163
+
## Future Enhancements
1164
+
1165
+
### 1. Quota API Endpoints
1166
+
1167
+
```
1168
+
GET /quota/usage - Get current user's quota
1169
+
GET /quota/breakdown - Get storage by repository
1170
+
POST /quota/limit - Update user's quota limit (admin)
1171
+
GET /quota/stats - Get hold-wide statistics
1172
+
```
1173
+
1174
+
### 2. Quota Alerts
1175
+
1176
+
Notify users when approaching limit:
1177
+
- Email/webhook at 80%, 90%, 95%
1178
+
- Reject uploads at 100% (currently implemented)
1179
+
- Grace period: allow 105% temporarily
1180
+
1181
+
### 3. Tiered Quotas
1182
+
1183
+
Different limits based on user tier:
1184
+
- Free: 10GB
1185
+
- Pro: 100GB
1186
+
- Enterprise: unlimited
1187
+
1188
+
### 4. Quota Purchasing
1189
+
1190
+
Allow users to buy additional storage:
1191
+
- Stripe integration
1192
+
- $0.10/GB/month pricing
1193
+
- Dynamic limit updates
1194
+
1195
+
### 5. Cross-Hold Deduplication
1196
+
1197
+
If multiple holds share same S3 bucket:
1198
+
- Track blob ownership globally
1199
+
- Split costs proportionally
1200
+
- More complex, but maximizes deduplication
1201
+
1202
+
### 6. Manifest-Based Quota (Alternative Model)
1203
+
1204
+
Instead of tracking layers, track manifests:
1205
+
- Simpler: just count manifest sizes
1206
+
- No deduplication benefits for users
1207
+
- Might be acceptable for some use cases
1208
+
1209
+
### 7. Redis-Based Quota (High Performance)
1210
+
1211
+
For high-traffic registries:
1212
+
- Use Redis instead of S3/SQLite
1213
+
- Sub-millisecond quota checks
1214
+
- Harbor-proven approach
1215
+
1216
+
### 8. Quota Visualizations
1217
+
1218
+
Web UI showing:
1219
+
- Storage usage over time
1220
+
- Top consumers by repository
1221
+
- Deduplication savings graph
1222
+
- Layer size distribution
1223
+
1224
+
## Appendix: SQL Queries
1225
+
1226
+
### Check if User Still References Layer
1227
+
1228
+
```sql
1229
+
-- After deleting manifest, check if user has other manifests using this layer
1230
+
SELECT COUNT(*)
1231
+
FROM layers l
1232
+
JOIN manifests m ON l.manifest_id = m.id
1233
+
WHERE m.did = ? -- User's DID
1234
+
AND l.digest = ? -- Layer digest to check
1235
+
AND m.id != ? -- Exclude the manifest being deleted
1236
+
```
1237
+
1238
+
### Get All Unique Layers for User
1239
+
1240
+
```sql
1241
+
-- Calculate true quota usage for a user
1242
+
SELECT DISTINCT l.digest, l.size
1243
+
FROM layers l
1244
+
JOIN manifests m ON l.manifest_id = m.id
1245
+
WHERE m.did = ?
1246
+
AND m.hold_endpoint = ?
1247
+
```
1248
+
1249
+
### Get Referenced Blobs for Hold
1250
+
1251
+
```sql
1252
+
-- For GC: get all blobs still referenced by any user of this hold
1253
+
SELECT DISTINCT l.digest
1254
+
FROM layers l
1255
+
JOIN manifests m ON l.manifest_id = m.id
1256
+
WHERE m.hold_endpoint = ?
1257
+
```
1258
+
1259
+
### Get Storage Stats by Repository
1260
+
1261
+
```sql
1262
+
-- User's storage broken down by repository
1263
+
SELECT
1264
+
m.repository,
1265
+
COUNT(DISTINCT m.id) as manifest_count,
1266
+
COUNT(DISTINCT l.digest) as unique_layers,
1267
+
SUM(l.size) as total_size
1268
+
FROM manifests m
1269
+
JOIN layers l ON l.manifest_id = m.id
1270
+
WHERE m.did = ?
1271
+
AND m.hold_endpoint = ?
1272
+
GROUP BY m.repository
1273
+
ORDER BY total_size DESC
1274
+
```
1275
+
1276
+
## References
1277
+
1278
+
- **Harbor Quotas:** https://goharbor.io/docs/1.10/administration/configure-project-quotas/
1279
+
- **Harbor Source:** https://github.com/goharbor/harbor
1280
+
- **ATProto Spec:** https://atproto.com/specs/record
1281
+
- **OCI Distribution Spec:** https://github.com/opencontainers/distribution-spec
1282
+
- **S3 API Reference:** https://docs.aws.amazon.com/AmazonS3/latest/API/
1283
+
- **Distribution GC:** https://github.com/distribution/distribution/blob/main/registry/storage/garbagecollect.go
1284
+
1285
+
---
1286
+
1287
+
**Document Version:** 1.0
1288
+
**Last Updated:** 2025-10-09
1289
+
**Author:** Generated from implementation research and Harbor analysis
-460
docs/SPEC.md
-460
docs/SPEC.md
···
1
-
ATProto Container Registry (atcr.io) Implementation Plan
2
-
3
-
Project Structure
4
-
5
-
/home/data/atcr.io/
6
-
├── cmd/
7
-
│ └── registry/
8
-
│ └── main.go # Entrypoint that imports distribution
9
-
├── pkg/
10
-
│ ├── atproto/
11
-
│ │ ├── client.go # ATProto client wrapper (using indigo)
12
-
│ │ ├── manifest_store.go # Implements distribution.ManifestService
13
-
│ │ ├── resolver.go # DID/handle resolution (alice → did:plc:...)
14
-
│ │ └── lexicon.go # ATProto record schemas for manifests
15
-
│ ├── storage/
16
-
│ │ ├── s3_blob_store.go # Wraps distribution's S3 driver for blobs
17
-
│ │ └── routing_repository.go # Routes manifests→ATProto, blobs→S3
18
-
│ ├── middleware/
19
-
│ │ ├── repository.go # Repository middleware registration
20
-
│ │ └── registry.go # Registry middleware for name resolution
21
-
│ └── server/
22
-
│ └── handler.go # HTTP wrapper for custom name resolution
23
-
├── config/
24
-
│ └── config.yml # Registry configuration
25
-
├── go.mod
26
-
├── go.sum
27
-
├── Dockerfile
28
-
├── README.md
29
-
└── CLAUDE.md # Updated with architecture docs
30
-
31
-
32
-
Implementation Steps
33
-
34
-
Phase 1: Project Setup
35
-
36
-
1. Initialize Go module with github.com/distribution/distribution/v3 and github.com/bluesky-social/indigo
37
-
2. Create basic project structure
38
-
3. Set up cmd/appview/main.go that imports distribution and registers middleware
39
-
40
-
Phase 2: Core ATProto Integration
41
-
42
-
4. Implement DID/handle resolver (pkg/atproto/resolver.go)
43
-
- Resolve handles to DIDs (alice.bsky.social → did:plc:xyz)
44
-
- Discover PDS endpoints from DID documents
45
-
5. Create ATProto client wrapper (pkg/atproto/client.go)
46
-
- Wrap indigo SDK for manifest storage
47
-
- Handle authentication with PDS
48
-
6. Design ATProto lexicon for manifest records (pkg/atproto/lexicon.go)
49
-
- Define schema for storing OCI manifests as ATProto records
50
-
51
-
Phase 3: Storage Layer
52
-
53
-
7. Implement ATProto manifest store (pkg/atproto/manifest_store.go)
54
-
- Implements distribution.ManifestService
55
-
- Stores/retrieves manifests from PDS
56
-
8. Implement S3 blob store wrapper (pkg/storage/s3_blob_store.go)
57
-
- Wraps distribution's built-in S3 driver
58
-
9. Create routing repository (pkg/storage/routing_repository.go)
59
-
- Returns ATProto store for Manifests()
60
-
- Returns S3 store for Blobs()
61
-
62
-
Phase 4: Middleware Layer
63
-
64
-
10. Implement repository middleware (pkg/middleware/repository.go)
65
-
- Registers routing repository
66
-
- Configurable via YAML
67
-
11. Implement registry/namespace middleware (pkg/middleware/registry.go)
68
-
- Intercepts Repository(name) calls
69
-
- Performs name resolution before repository creation
70
-
71
-
Phase 5: HTTP Layer (if needed)
72
-
73
-
12. Create custom HTTP handler (pkg/server/handler.go)
74
-
- Wraps distribution's HTTP handlers
75
-
- Performs early name resolution: atcr.io/alice/myimage → resolve alice
76
-
- Delegates to distribution handlers
77
-
78
-
Phase 6: Configuration & Deployment
79
-
80
-
13. Create registry configuration (config/config.yml)
81
-
14. Create Dockerfile for building atcr-appview binary
82
-
16. Write README.md with usage instructions
83
-
84
-
Phase 7: Documentation
85
-
86
-
17. Update CLAUDE.md with:
87
-
- Architecture overview (ATProto for manifests, S3 for blobs)
88
-
- Build/run/test commands
89
-
- How name resolution works
90
-
- Middleware registration flow
91
-
- Key design decisions
92
-
93
-
Key Technical Decisions
94
-
95
-
ATProto Storage Design:
96
-
- Manifests stored as ATProto records in user's PDS
97
-
- Each image manifest is a record in a io.atcr.manifest collection
98
-
- Record key = image digest (sha256:...)
99
-
100
-
Name Resolution:
101
-
- atcr.io/alice/myimage → resolve alice to DID → discover PDS
102
-
- Support both handles (alice.bsky.social) and DIDs (did:plc:xyz)
103
-
104
-
Blob Storage:
105
-
- All layers/blobs in S3 (content-addressable by digest)
106
-
- Manifests reference S3-stored blobs by digest
107
-
- S3 provides cheap, durable blob storage
108
-
109
-
Middleware Strategy:
110
-
- Repository middleware for storage routing
111
-
- Registry middleware (or HTTP wrapper) for name resolution
112
-
- No fork of distribution core needed
113
-
114
-
Perfect. To match Docker Hub/ghcr.io/gcr.io, here's what we need:
115
-
116
-
● Implementation Plan (Drop-in replacement for Docker Hub/ghcr.io)
117
-
118
-
Flow 1: Standard Token Auth (Like Docker Hub) - PRIMARY
119
-
120
-
# User experience
121
-
docker login atcr.io -u alice.bsky.social -p <atproto-app-password>
122
-
docker push atcr.io/alice/myapp:latest
123
-
124
-
# Behind the scenes
125
-
1. docker login stores credentials locally
126
-
2. docker push → Registry returns 401 with WWW-Authenticate: Bearer realm="https://atcr.io/auth/token"...
127
-
3. Docker auto-calls /auth/token with Basic auth (alice.bsky.social:app-password)
128
-
4. Auth service validates against ATProto createSession
129
-
5. Returns JWT token with scope for alice/myapp
130
-
6. Docker uses JWT for manifest/blob uploads
131
-
7. Registry validates JWT signature and scope
132
-
133
-
Components:
134
-
- /auth/token endpoint (standalone service or embedded)
135
-
- ATProto session validator (username/password → validate via PDS)
136
-
- JWT issuer/signer
137
-
- JWT validator middleware for registry
138
-
139
-
Flow 2: Credential Helper (Like gcr.io) - ADVANCED
140
-
141
-
# User experience
142
-
docker-credential-atcr configure
143
-
# Opens browser for ATProto OAuth
144
-
docker push atcr.io/alice/myapp:latest
145
-
# No manual login needed
146
-
147
-
# Behind the scenes
148
-
1. Helper does OAuth flow → gets ATProto access token
149
-
2. Caches token securely
150
-
3. When Docker needs credentials, calls helper via stdin/stdout
151
-
4. Helper exchanges ATProto token for registry JWT at /auth/exchange
152
-
5. Returns JWT to Docker
153
-
6. Docker uses JWT for requests
154
-
155
-
Components:
156
-
- cmd/credential-helper/main.go - Standalone binary
157
-
- ATProto OAuth client
158
-
- Token exchange endpoint (/auth/exchange)
159
-
- Secure token cache
160
-
161
-
Architecture:
162
-
163
-
pkg/auth/
164
-
├── token/
165
-
│ ├── service.go # HTTP handler for /auth/token
166
-
│ ├── claims.go # JWT claims structure
167
-
│ ├── issuer.go # Signs JWTs
168
-
│ └── validator.go # Validates JWTs (middleware for registry)
169
-
├── atproto/
170
-
│ ├── session.go # Validates username/password via ATProto
171
-
│ └── oauth.go # OAuth flow implementation
172
-
├── exchange/
173
-
│ └── handler.go # /auth/exchange endpoint (OAuth → JWT)
174
-
└── scope.go # Parses/validates Docker scopes
175
-
176
-
cmd/
177
-
├── registry/main.go # Registry server (existing)
178
-
├── auth/main.go # Standalone auth service (optional)
179
-
└── credential-helper/
180
-
└── main.go # docker-credential-atcr binary
181
-
182
-
Config:
183
-
184
-
auth:
185
-
token:
186
-
realm: https://atcr.io/auth/token # Where Docker gets tokens
187
-
service: atcr.io
188
-
issuer: atcr.io
189
-
rootcertbundle: /etc/atcr/token-signing.crt
190
-
privatekey: /etc/atcr/token-signing.pem
191
-
expiration: 300
192
-
193
-
atproto:
194
-
# Used by auth service to validate credentials
195
-
pds_endpoint: https://bsky.social
196
-
client_id: atcr-appview
197
-
oauth_redirect: http://localhost:8888/callback
198
-
199
-
ATProto OAuth Implementation Plan
200
-
201
-
Architecture
202
-
203
-
Dependencies:
204
-
- authelia.com/client/oauth2 - OAuth + PAR support
205
-
- github.com/AxisCommunications/go-dpop - DPoP proof generation (handles JWK automatically)
206
-
- github.com/golang-jwt/jwt/v5 - JWT library (transitive via go-dpop)
207
-
- Our existing pkg/atproto/resolver.go - ATProto identity resolution
208
-
209
-
Implementation Components
210
-
211
-
1. OAuth Client (pkg/auth/oauth/client.go) - ~100 lines
212
-
213
-
type Client struct {
214
-
config *oauth2.Config
215
-
dpopKey *ecdsa.PrivateKey
216
-
resolver *atproto.Resolver
217
-
clientID string // URL to our metadata document
218
-
redirectURI string
219
-
dpopNonce string // Server-provided nonce
220
-
}
221
-
222
-
func NewClient(clientID, redirectURI string) (*Client, error)
223
-
func (c *Client) AuthorizeURL(handle string, scopes []string) (string, error)
224
-
func (c *Client) Exchange(code string) (*Token, error)
225
-
func (c *Client) addDPoPHeader(req *http.Request, method, url string) error
226
-
227
-
Flow:
228
-
1. Generate ECDSA P-256 key for DPoP
229
-
2. Discover authorization server from handle/DID
230
-
3. Use authelia's PushedAuth() for PAR with DPoP header
231
-
4. Exchange code for token with DPoP proof
232
-
233
-
2. Authorization Server Discovery (pkg/auth/oauth/discovery.go) - ~30 lines
234
-
235
-
type AuthServerMetadata struct {
236
-
Issuer string `json:"issuer"`
237
-
AuthorizationEndpoint string `json:"authorization_endpoint"`
238
-
TokenEndpoint string `json:"token_endpoint"`
239
-
PushedAuthorizationRequestEndpoint string `json:"pushed_authorization_request_endpoint"`
240
-
DPoPSigningAlgValuesSupported []string `json:"dpop_signing_alg_values_supported"`
241
-
}
242
-
243
-
func DiscoverAuthServer(pdsEndpoint string) (*AuthServerMetadata, error)
244
-
245
-
Implementation:
246
-
- GET {pds}/.well-known/oauth-authorization-server
247
-
- Parse JSON metadata
248
-
- Validate required endpoints exist
249
-
250
-
3. Client Metadata Server (pkg/auth/oauth/metadata.go) - ~40 lines
251
-
252
-
type ClientMetadata struct {
253
-
ClientID string `json:"client_id"`
254
-
RedirectURIs []string `json:"redirect_uris"`
255
-
GrantTypes []string `json:"grant_types"`
256
-
ResponseTypes []string `json:"response_types"`
257
-
Scope string `json:"scope"`
258
-
DPoPBoundAccessTokens bool `json:"dpop_bound_access_tokens"`
259
-
}
260
-
261
-
func ServeMetadata(clientID string, redirectURIs []string) http.Handler
262
-
263
-
Serves: https://atcr.io/oauth/client-metadata.json
264
-
265
-
4. Token Storage (pkg/auth/oauth/storage.go) - ~50 lines
266
-
267
-
type TokenStore struct {
268
-
AccessToken string
269
-
RefreshToken string
270
-
DPoPKey *ecdsa.PrivateKey // Persist for refresh
271
-
ExpiresAt time.Time
272
-
}
273
-
274
-
func (s *TokenStore) Save(path string) error
275
-
func LoadTokenStore(path string) (*TokenStore, error)
276
-
277
-
Storage location: ~/.atcr/oauth-tokens.json
278
-
279
-
5. Credential Helper (cmd/credential-helper/main.go) - ~80 lines
280
-
281
-
// Docker credential helper protocol
282
-
// Reads JSON from stdin, writes to stdout
283
-
284
-
func main() {
285
-
if len(os.Args) < 2 {
286
-
os.Exit(1)
287
-
}
288
-
289
-
switch os.Args[1] {
290
-
case "get":
291
-
handleGet() // Return credentials for registry
292
-
case "store":
293
-
handleStore() // Store credentials
294
-
case "erase":
295
-
handleErase() // Remove credentials
296
-
}
297
-
}
298
-
299
-
func handleGet() {
300
-
var request struct {
301
-
ServerURL string `json:"ServerURL"`
302
-
}
303
-
json.NewDecoder(os.Stdin).Decode(&request)
304
-
305
-
// Load token from storage
306
-
// Exchange for registry JWT if needed
307
-
// Output: {"Username": "oauth2", "Secret": "<jwt>"}
308
-
}
309
-
310
-
6. OAuth Flow (cmd/credential-helper/oauth.go) - ~60 lines
311
-
312
-
func RunOAuthFlow(handle string) (*TokenStore, error) {
313
-
// 1. Start local HTTP server on :8888
314
-
// 2. Open browser to authorization URL
315
-
// 3. Wait for callback with code
316
-
// 4. Exchange code for token
317
-
// 5. Save token store
318
-
// 6. Return token
319
-
}
320
-
321
-
func startCallbackServer() (chan string, *http.Server)
322
-
323
-
Complete Flow Example
324
-
325
-
User runs:
326
-
docker-credential-atcr configure
327
-
328
-
What happens:
329
-
330
-
1. Generate DPoP key (client.go)
331
-
dpopKey, _ := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
332
-
333
-
2. Resolve handle → DID → PDS (using our resolver)
334
-
did, pds, _ := resolver.ResolveIdentity(ctx, "alice.bsky.social")
335
-
336
-
3. Discover auth server (discovery.go)
337
-
metadata, _ := DiscoverAuthServer(pds)
338
-
// Returns: PAR endpoint, token endpoint, etc.
339
-
340
-
4. Create PAR request with DPoP (client.go + go-dpop)
341
-
// Generate DPoP proof for PAR endpoint
342
-
claims := &dpop.ProofTokenClaims{
343
-
Method: dpop.POST,
344
-
URL: metadata.PushedAuthorizationRequestEndpoint,
345
-
RegisteredClaims: &jwt.RegisteredClaims{
346
-
IssuedAt: jwt.NewNumericDate(time.Now()),
347
-
},
348
-
}
349
-
dpopProof, _ := dpop.Create(jwt.SigningMethodES256, claims, dpopKey)
350
-
351
-
// Use authelia for PAR
352
-
config := &oauth2.Config{
353
-
ClientID: "https://atcr.io/oauth/client-metadata.json",
354
-
Endpoint: oauth2.Endpoint{
355
-
AuthURL: metadata.AuthorizationEndpoint,
356
-
TokenURL: metadata.TokenEndpoint,
357
-
},
358
-
}
359
-
360
-
// Create custom HTTP client that adds DPoP header
361
-
client := &http.Client{
362
-
Transport: &dpopTransport{
363
-
base: http.DefaultTransport,
364
-
dpopKey: dpopKey,
365
-
},
366
-
}
367
-
ctx := context.WithValue(context.Background(), oauth2.HTTPClient, client)
368
-
369
-
// PAR request (authelia handles this)
370
-
authURL, parResp, _ := config.PushedAuth(ctx, state,
371
-
oauth2.SetAuthURLParam("code_challenge", pkceChallenge),
372
-
oauth2.SetAuthURLParam("code_challenge_method", "S256"),
373
-
)
374
-
375
-
5. Open browser, get code (oauth.go)
376
-
exec.Command("open", authURL).Run()
377
-
// User authorizes
378
-
// Callback: http://localhost:8888?code=xyz&state=abc
379
-
380
-
6. Exchange code for token with DPoP (client.go + go-dpop)
381
-
// Generate DPoP proof for token endpoint
382
-
claims := &dpop.ProofTokenClaims{
383
-
Method: dpop.POST,
384
-
URL: metadata.TokenEndpoint,
385
-
RegisteredClaims: &jwt.RegisteredClaims{
386
-
IssuedAt: jwt.NewNumericDate(time.Now()),
387
-
},
388
-
}
389
-
dpopProof, _ := dpop.Create(jwt.SigningMethodES256, claims, dpopKey)
390
-
391
-
// Exchange (with DPoP header added by our transport)
392
-
token, _ := config.Exchange(ctx, code,
393
-
oauth2.SetAuthURLParam("code_verifier", pkceVerifier),
394
-
)
395
-
396
-
7. Save token + DPoP key (storage.go)
397
-
store := &TokenStore{
398
-
AccessToken: token.AccessToken,
399
-
RefreshToken: token.RefreshToken,
400
-
DPoPKey: dpopKey,
401
-
ExpiresAt: token.Expiry,
402
-
}
403
-
store.Save("~/.atcr/oauth-tokens.json")
404
-
405
-
Later, when docker push happens:
406
-
docker push atcr.io/alice/myapp:latest
407
-
408
-
1. Docker calls credential helper: docker-credential-atcr get
409
-
2. Helper loads stored token
410
-
3. Helper calls /auth/exchange with OAuth token → gets registry JWT
411
-
4. Returns JWT to Docker
412
-
5. Docker uses JWT for push
413
-
414
-
Directory Structure
415
-
416
-
pkg/auth/oauth/
417
-
├── client.go # OAuth client with DPoP integration
418
-
├── discovery.go # Authorization server discovery
419
-
├── metadata.go # Client metadata server
420
-
├── storage.go # Token persistence
421
-
└── transport.go # HTTP transport that adds DPoP headers
422
-
423
-
cmd/credential-helper/
424
-
├── main.go # Docker credential helper protocol
425
-
├── oauth.go # OAuth flow (browser, callback)
426
-
└── config.go # Configuration
427
-
428
-
go.mod additions:
429
-
authelia.com/client/oauth2 v0.25.0
430
-
github.com/AxisCommunications/go-dpop v1.1.2
431
-
432
-
Unified Model
433
-
434
-
Every hold service requires HOLD_OWNER:
435
-
- Owner's PDS has the io.atcr.hold record
436
-
- Owner's PDS has all io.atcr.hold.crew records
437
-
- Authorization is always governed by PDS records
438
-
439
-
For "public" hold (like Tangled's public knot):
440
-
- Owner creates hold with public: true
441
-
- Anyone can push/pull without being crew
442
-
- Owner can add crew records for special privileges/tracking if desired
443
-
444
-
Config has emergency override:
445
-
auth:
446
-
# Emergency freeze: ignore public setting, restrict to crew only
447
-
# Use this to stop abuse without changing PDS records
448
-
freeze: false
449
-
450
-
Authorization logic:
451
-
1. Check freeze in config → if true, skip to crew check
452
-
2. Query owner's PDS for io.atcr.hold record
453
-
3. If public: true → allow all operations (unless frozen)
454
-
4. If public: false OR frozen → query io.atcr.hold.crew records, check membership
455
-
456
-
Remove from config:
457
-
- allow_all (replaced by public: true in PDS)
458
-
- allowed_dids (replaced by crew records in PDS)
459
-
460
-
This way the hold owner at atcr.io can run a public hold at hold1.atcr.io that anyone can use, but can freeze it instantly if needed without touching PDS records.
-334
docs/TESTING.md
-334
docs/TESTING.md
···
1
-
# Local Testing Guide
2
-
3
-
## Quick Start
4
-
5
-
```bash
6
-
./test-local.sh
7
-
```
8
-
9
-
This automated script will:
10
-
1. Create storage directories
11
-
2. Build all binaries
12
-
3. Start both services
13
-
4. Show test commands
14
-
15
-
## Manual Testing Steps
16
-
17
-
### 1. Setup Directories
18
-
19
-
```bash
20
-
sudo mkdir -p /var/lib/atcr/{blobs,hold,auth}
21
-
sudo chown -R $USER:$USER /var/lib/atcr
22
-
```
23
-
24
-
### 2. Build Binaries
25
-
26
-
```bash
27
-
go build -o atcr-appview ./cmd/appview
28
-
go build -o atcr-hold ./cmd/hold
29
-
go build -o docker-credential-atcr ./cmd/credential-helper
30
-
```
31
-
32
-
### 3. Configure Environment
33
-
34
-
Create a `.env` file in the project root:
35
-
36
-
```bash
37
-
cp .env.example .env
38
-
```
39
-
40
-
Edit `.env` with your credentials:
41
-
42
-
```env
43
-
# Your ATProto handle
44
-
ATPROTO_HANDLE=your-handle.bsky.social
45
-
46
-
# Hold service public URL (hostname becomes the hold name)
47
-
HOLD_PUBLIC_URL=http://127.0.0.1:8080
48
-
49
-
# Enable OAuth registration on startup
50
-
HOLD_AUTO_REGISTER=true
51
-
```
52
-
53
-
**Notes:**
54
-
- Use your Bluesky handle (e.g., `alice.bsky.social`)
55
-
- For localhost, use `127.0.0.1` instead of `localhost` for OAuth
56
-
- The hostname from the URL becomes the hold name (e.g., `127.0.0.1` or `hold1.atcr.io`)
57
-
58
-
**Load environment:**
59
-
```bash
60
-
export $(cat .env | xargs)
61
-
```
62
-
63
-
### 4. Start Services
64
-
65
-
**Terminal 1 - AppView:**
66
-
```bash
67
-
./atcr-appview serve config/config.yml
68
-
```
69
-
70
-
**Terminal 2 - Hold:**
71
-
```bash
72
-
./atcr-hold config/hold.yml
73
-
```
74
-
75
-
### 5. Start Services and OAuth Registration
76
-
77
-
**Terminal 1 - AppView:**
78
-
```bash
79
-
./atcr-appview serve config/config.yml
80
-
```
81
-
82
-
**Terminal 2 - Hold (OAuth registration):**
83
-
```bash
84
-
./atcr-hold config/hold.yml
85
-
```
86
-
87
-
The hold service will start an OAuth flow. You'll see output like:
88
-
89
-
```
90
-
================================================================================
91
-
OAUTH AUTHORIZATION REQUIRED
92
-
================================================================================
93
-
94
-
Please visit this URL to authorize the hold service:
95
-
96
-
https://bsky.social/oauth/authorize?...
97
-
98
-
Waiting for authorization...
99
-
================================================================================
100
-
```
101
-
102
-
**Steps:**
103
-
1. Copy the OAuth URL from the logs
104
-
2. Open it in your browser
105
-
3. Sign in to Bluesky and authorize
106
-
4. The callback will complete automatically
107
-
5. Hold service registers in your PDS
108
-
109
-
After successful OAuth, you'll see:
110
-
```
111
-
✓ Created hold record: at://did:plc:.../io.atcr.hold/127.0.0.1
112
-
✓ Created crew record: at://did:plc:.../io.atcr.hold.crew/127.0.0.1-did:plc:...
113
-
================================================================================
114
-
REGISTRATION COMPLETE
115
-
================================================================================
116
-
Hold service is now registered and ready to use!
117
-
```
118
-
119
-
This creates two records in your PDS:
120
-
- `io.atcr.hold` - Defines the storage endpoint URL
121
-
- `io.atcr.hold.crew` - Grants you admin access
122
-
123
-
### 6. Test Docker Push/Pull
124
-
125
-
**Test 1: Basic Push**
126
-
```bash
127
-
# Tag an image
128
-
docker tag alpine:latest localhost:5000/alice/alpine:test
129
-
130
-
# Push to local registry
131
-
docker push localhost:5000/alice/alpine:test
132
-
```
133
-
134
-
**Test 2: Pull**
135
-
```bash
136
-
# Remove local image
137
-
docker rmi localhost:5000/alice/alpine:test
138
-
139
-
# Pull from registry
140
-
docker pull localhost:5000/alice/alpine:test
141
-
```
142
-
143
-
**Test 3: Verify Storage**
144
-
```bash
145
-
# Check manifests were stored in ATProto
146
-
# (Check your PDS for io.atcr.manifest records)
147
-
148
-
# Check blobs were stored locally
149
-
ls -lh /var/lib/atcr/blobs/docker/registry/v2/
150
-
```
151
-
152
-
## OAuth Testing (Optional)
153
-
154
-
### Setup Credential Helper
155
-
156
-
```bash
157
-
# Configure OAuth
158
-
./docker-credential-atcr configure
159
-
160
-
# Follow the browser flow to authorize
161
-
162
-
# Verify token was saved
163
-
ls -la ~/.atcr/oauth-token.json
164
-
```
165
-
166
-
### Configure Docker to Use Helper
167
-
168
-
Edit `~/.docker/config.json`:
169
-
```json
170
-
{
171
-
"credHelpers": {
172
-
"localhost:5000": "atcr"
173
-
}
174
-
}
175
-
```
176
-
177
-
### Test with OAuth
178
-
179
-
```bash
180
-
# Push should now use OAuth automatically
181
-
docker push localhost:5000/alice/myapp:latest
182
-
```
183
-
184
-
## Troubleshooting
185
-
186
-
### Registry won't start
187
-
188
-
**Error:** `failed to create storage driver`
189
-
```bash
190
-
# Check directory permissions
191
-
ls -ld /var/lib/atcr/blobs
192
-
# Should be owned by your user
193
-
194
-
# Fix permissions
195
-
sudo chown -R $USER:$USER /var/lib/atcr
196
-
```
197
-
198
-
**Error:** `address already in use`
199
-
```bash
200
-
# Check what's using port 5000
201
-
lsof -i :5000
202
-
203
-
# Kill existing process
204
-
kill $(lsof -t -i :5000)
205
-
```
206
-
207
-
### Hold service won't start
208
-
209
-
**Error:** `failed to create storage driver`
210
-
```bash
211
-
# Check hold directory
212
-
ls -ld /var/lib/atcr/hold
213
-
sudo chown -R $USER:$USER /var/lib/atcr/hold
214
-
```
215
-
216
-
**Error:** `address already in use`
217
-
```bash
218
-
# Check port 8080
219
-
lsof -i :8080
220
-
kill $(lsof -t -i :8080)
221
-
```
222
-
223
-
### Docker push fails
224
-
225
-
**Error:** `unauthorized: authentication required`
226
-
- Check `ATPROTO_DID` and `ATPROTO_ACCESS_TOKEN` are set
227
-
- Verify token is valid (not expired)
228
-
- Check registry logs for auth errors
229
-
230
-
**Error:** `denied: requested access to the resource is denied`
231
-
- Check the identity in the image name matches your DID
232
-
- Example: If your handle is `alice.bsky.social`, use:
233
-
```bash
234
-
docker push localhost:5000/alice/myapp:test
235
-
# NOT localhost:5000/bob/myapp:test
236
-
```
237
-
238
-
**Error:** `failed to resolve identity`
239
-
- Check internet connection (needs to resolve DIDs)
240
-
- Verify handle is correct
241
-
- Try using DID directly instead of handle
242
-
243
-
### OAuth issues
244
-
245
-
**Error:** `Failed to exchange token`
246
-
- Ensure registry is running and accessible
247
-
- Check `/auth/exchange` endpoint is responding
248
-
- Verify OAuth token hasn't expired
249
-
250
-
**Error:** `Token validation failed`
251
-
- Token might be expired
252
-
- Run `./docker-credential-atcr configure` again
253
-
- Check PDS is accessible
254
-
255
-
## Verifying the Flow
256
-
257
-
### Check Registry is Running
258
-
```bash
259
-
curl http://localhost:5000/v2/
260
-
# Should return: {}
261
-
```
262
-
263
-
### Check Hold is Running
264
-
```bash
265
-
curl http://localhost:8080/health
266
-
# Should return: {"status":"ok"}
267
-
```
268
-
269
-
### Check Auth Endpoint
270
-
```bash
271
-
curl -v http://localhost:5000/v2/
272
-
# Should return 401 with WWW-Authenticate header
273
-
```
274
-
275
-
### Inspect Stored Data
276
-
277
-
**Manifests (in ATProto):**
278
-
- Check your PDS web interface
279
-
- Look for `io.atcr.manifest` collection records
280
-
281
-
**Blobs (local filesystem):**
282
-
```bash
283
-
# List blobs
284
-
find /var/lib/atcr/blobs -type f
285
-
286
-
# Check blob content (should be binary)
287
-
ls -lh /var/lib/atcr/blobs/docker/registry/v2/blobs/sha256/
288
-
```
289
-
290
-
## Clean Up
291
-
292
-
### Stop Services
293
-
```bash
294
-
# If using test script
295
-
kill $(cat .atcr-pids)
296
-
297
-
# Or manually
298
-
pkill atcr-appview
299
-
pkill atcr-hold
300
-
```
301
-
302
-
### Remove Test Data
303
-
```bash
304
-
# Remove all stored data
305
-
sudo rm -rf /var/lib/atcr/*
306
-
307
-
# Remove OAuth tokens
308
-
rm -rf ~/.atcr/
309
-
```
310
-
311
-
### Reset Docker Config
312
-
```bash
313
-
# Remove credential helper config
314
-
# Edit ~/.docker/config.json and remove "credHelpers" section
315
-
```
316
-
317
-
## Next Steps
318
-
319
-
Once local testing works:
320
-
321
-
1. **Deploy to production:**
322
-
- Use S3/Storj for blob storage
323
-
- Deploy registry and hold to separate hosts
324
-
- Configure DNS for `atcr.io`
325
-
326
-
2. **Enable BYOS:**
327
-
- Users create `io.atcr.hold` records
328
-
- Deploy their own hold service
329
-
- AppView automatically routes to their storage
330
-
331
-
3. **Add monitoring:**
332
-
- Registry metrics
333
-
- Hold service metrics
334
-
- Storage usage tracking