download.go 10.7 KB
Newer Older
1
2
3
4
package server

import (
	"context"
Michael Yang's avatar
Michael Yang committed
5
	"encoding/json"
6
7
8
	"errors"
	"fmt"
	"io"
9
	"log/slog"
Jeffrey Morgan's avatar
Jeffrey Morgan committed
10
	"math"
11
	"math/rand/v2"
12
	"net/http"
Michael Yang's avatar
Michael Yang committed
13
	"net/url"
14
	"os"
Michael Yang's avatar
Michael Yang committed
15
	"path/filepath"
16
	"strconv"
Michael Yang's avatar
Michael Yang committed
17
	"strings"
18
19
	"sync"
	"sync/atomic"
20
	"syscall"
21
	"time"
22

Michael Yang's avatar
Michael Yang committed
23
	"golang.org/x/sync/errgroup"
24

25
26
	"github.com/ollama/ollama/api"
	"github.com/ollama/ollama/format"
27
28
)

29
30
31
32
33
const maxRetries = 6

var errMaxRetriesExceeded = errors.New("max retries exceeded")
var errPartStalled = errors.New("part stalled")

34
var blobDownloadManager sync.Map
35

36
37
38
type blobDownload struct {
	Name   string
	Digest string
39

40
41
	Total     int64
	Completed atomic.Int64
42

43
	Parts []*blobDownloadPart
44

45
	context.CancelFunc
Michael Yang's avatar
Michael Yang committed
46

47
	done       chan struct{}
Michael Yang's avatar
Michael Yang committed
48
	err        error
Michael Yang's avatar
Michael Yang committed
49
	references atomic.Int32
50
}
51

52
type blobDownloadPart struct {
53
54
55
56
57
58
59
	N         int
	Offset    int64
	Size      int64
	Completed atomic.Int64

	lastUpdatedMu sync.Mutex
	lastUpdated   time.Time
Michael Yang's avatar
Michael Yang committed
60
61
62
63

	*blobDownload `json:"-"`
}

64
65
const (
	numDownloadParts          = 64
Michael Yang's avatar
Michael Yang committed
66
67
	minDownloadPartSize int64 = 100 * format.MegaByte
	maxDownloadPartSize int64 = 1000 * format.MegaByte
68
69
)

Michael Yang's avatar
Michael Yang committed
70
71
72
73
func (p *blobDownloadPart) Name() string {
	return strings.Join([]string{
		p.blobDownload.Name, "partial", strconv.Itoa(p.N),
	}, "-")
74
}
75

76
func (p *blobDownloadPart) StartsAt() int64 {
77
	return p.Offset + p.Completed.Load()
78
79
80
81
82
83
}

func (p *blobDownloadPart) StopsAt() int64 {
	return p.Offset + p.Size
}

84
85
86
func (p *blobDownloadPart) Write(b []byte) (n int, err error) {
	n = len(b)
	p.blobDownload.Completed.Add(int64(n))
87
	p.lastUpdatedMu.Lock()
88
	p.lastUpdated = time.Now()
89
	p.lastUpdatedMu.Unlock()
90
91
92
	return n, nil
}

Michael Yang's avatar
Michael Yang committed
93
func (b *blobDownload) Prepare(ctx context.Context, requestURL *url.URL, opts *registryOptions) error {
94
	partFilePaths, err := filepath.Glob(b.Name + "-partial-*")
Michael Yang's avatar
Michael Yang committed
95
	if err != nil {
96
		return err
97
98
	}

99
100
	b.done = make(chan struct{})

Michael Yang's avatar
Michael Yang committed
101
	for _, partFilePath := range partFilePaths {
102
		part, err := b.readPart(partFilePath)
Bruce MacDonald's avatar
Bruce MacDonald committed
103
104
		if err != nil {
			return err
105
106
		}

107
		b.Total += part.Size
108
		b.Completed.Add(part.Completed.Load())
109
		b.Parts = append(b.Parts, part)
Michael Yang's avatar
Michael Yang committed
110
	}
111

112
	if len(b.Parts) == 0 {
Michael Yang's avatar
Michael Yang committed
113
		resp, err := makeRequestWithRetry(ctx, http.MethodHead, requestURL, nil, nil, opts)
114
		if err != nil {
Michael Yang's avatar
Michael Yang committed
115
116
117
118
			return err
		}
		defer resp.Body.Close()

119
		b.Total, _ = strconv.ParseInt(resp.Header.Get("Content-Length"), 10, 64)
Michael Yang's avatar
Michael Yang committed
120

Michael Yang's avatar
Michael Yang committed
121
		size := b.Total / numDownloadParts
122
123
124
125
126
127
		switch {
		case size < minDownloadPartSize:
			size = minDownloadPartSize
		case size > maxDownloadPartSize:
			size = maxDownloadPartSize
		}
Michael Yang's avatar
Michael Yang committed
128

129
		var offset int64
130
131
132
133
134
		for offset < b.Total {
			if offset+size > b.Total {
				size = b.Total - offset
			}

Michael Yang's avatar
Michael Yang committed
135
			if err := b.newPart(offset, size); err != nil {
136
				return err
Michael Yang's avatar
Michael Yang committed
137
138
139
			}

			offset += size
140
141
142
		}
	}

143
	slog.Info(fmt.Sprintf("downloading %s in %d %s part(s)", b.Digest[7:19], len(b.Parts), format.HumanBytes(b.Parts[0].Size)))
144
145
146
	return nil
}

Michael Yang's avatar
Michael Yang committed
147
func (b *blobDownload) Run(ctx context.Context, requestURL *url.URL, opts *registryOptions) {
148
	defer close(b.done)
149
150
151
	b.err = b.run(ctx, requestURL, opts)
}

152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
func newBackoff(maxBackoff time.Duration) func(ctx context.Context) error {
	var n int
	return func(ctx context.Context) error {
		if ctx.Err() != nil {
			return ctx.Err()
		}

		n++

		// n^2 backoff timer is a little smoother than the
		// common choice of 2^n.
		d := min(time.Duration(n*n)*10*time.Millisecond, maxBackoff)
		// Randomize the delay between 0.5-1.5 x msec, in order
		// to prevent accidental "thundering herd" problems.
		d = time.Duration(float64(d) * (rand.Float64() + 0.5))
		t := time.NewTimer(d)
		defer t.Stop()
		select {
		case <-ctx.Done():
			return ctx.Err()
		case <-t.C:
			return nil
		}
	}
}

178
func (b *blobDownload) run(ctx context.Context, requestURL *url.URL, opts *registryOptions) error {
179
180
181
	defer blobDownloadManager.Delete(b.Digest)
	ctx, b.CancelFunc = context.WithCancel(ctx)

Michael Yang's avatar
Michael Yang committed
182
	file, err := os.OpenFile(b.Name+"-partial", os.O_CREATE|os.O_RDWR, 0o644)
183
	if err != nil {
184
		return err
Michael Yang's avatar
Michael Yang committed
185
	}
186
	defer file.Close()
187

Michael Yang's avatar
Michael Yang committed
188
	_ = file.Truncate(b.Total)
Michael Yang's avatar
Michael Yang committed
189

190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
	directURL, err := func() (*url.URL, error) {
		ctx, cancel := context.WithTimeout(ctx, 30*time.Second)
		defer cancel()

		backoff := newBackoff(10 * time.Second)
		for {
			// shallow clone opts to be used in the closure
			// without affecting the outer opts.
			newOpts := new(registryOptions)
			*newOpts = *opts

			newOpts.CheckRedirect = func(req *http.Request, via []*http.Request) error {
				if len(via) > 10 {
					return errors.New("maxium redirects exceeded (10) for directURL")
				}

				// if the hostname is the same, allow the redirect
				if req.URL.Hostname() == requestURL.Hostname() {
					return nil
				}

				// stop at the first redirect that is not
				// the same hostname as the original
				// request.
				return http.ErrUseLastResponse
			}

			resp, err := makeRequestWithRetry(ctx, http.MethodGet, requestURL, nil, nil, newOpts)
			if err != nil {
				slog.Warn("failed to get direct URL; backing off and retrying", "err", err)
				if err := backoff(ctx); err != nil {
					return nil, err
				}
				continue
			}
			defer resp.Body.Close()
			if resp.StatusCode != http.StatusTemporaryRedirect {
				return nil, fmt.Errorf("unexpected status code %d", resp.StatusCode)
			}
			return resp.Location()
		}
	}()
	if err != nil {
		return err
	}

236
237
	g, inner := errgroup.WithContext(ctx)
	g.SetLimit(numDownloadParts)
238
239
	for i := range b.Parts {
		part := b.Parts[i]
240
		if part.Completed.Load() == part.Size {
Michael Yang's avatar
Michael Yang committed
241
242
			continue
		}
243

244
		g.Go(func() error {
Michael Yang's avatar
Michael Yang committed
245
			var err error
Michael Yang's avatar
Michael Yang committed
246
			for try := 0; try < maxRetries; try++ {
247
				w := io.NewOffsetWriter(file, part.StartsAt())
248
				err = b.downloadChunk(inner, directURL, w, part)
249
				switch {
250
251
				case errors.Is(err, context.Canceled), errors.Is(err, syscall.ENOSPC):
					// return immediately if the context is canceled or the device is out of space
252
					return err
253
254
255
				case errors.Is(err, errPartStalled):
					try--
					continue
256
				case err != nil:
Jeffrey Morgan's avatar
Jeffrey Morgan committed
257
					sleep := time.Second * time.Duration(math.Pow(2, float64(try)))
258
					slog.Info(fmt.Sprintf("%s part %d attempt %d failed: %v, retrying in %s", b.Digest[7:19], part.N, try, err, sleep))
Michael Yang's avatar
Michael Yang committed
259
					time.Sleep(sleep)
Michael Yang's avatar
Michael Yang committed
260
					continue
261
262
				default:
					return nil
Michael Yang's avatar
Michael Yang committed
263
264
265
				}
			}

Michael Yang's avatar
Michael Yang committed
266
			return fmt.Errorf("%w: %w", errMaxRetriesExceeded, err)
Michael Yang's avatar
Michael Yang committed
267
		})
268
269
	}

Michael Yang's avatar
Michael Yang committed
270
	if err := g.Wait(); err != nil {
271
		return err
272
273
	}

274
275
	// explicitly close the file so we can rename it
	if err := file.Close(); err != nil {
276
		return err
Michael Yang's avatar
Michael Yang committed
277
278
	}

279
	for i := range b.Parts {
280
		if err := os.Remove(file.Name() + "-" + strconv.Itoa(i)); err != nil {
281
			return err
Michael Yang's avatar
Michael Yang committed
282
		}
283
284
	}

Michael Yang's avatar
Michael Yang committed
285
	if err := os.Rename(file.Name(), b.Name); err != nil {
286
		return err
Michael Yang's avatar
Michael Yang committed
287
288
	}

289
	return nil
290
291
}

292
func (b *blobDownload) downloadChunk(ctx context.Context, requestURL *url.URL, w io.Writer, part *blobDownloadPart) error {
293
294
	g, ctx := errgroup.WithContext(ctx)
	g.Go(func() error {
295
296
297
298
299
300
		req, err := http.NewRequestWithContext(ctx, http.MethodGet, requestURL.String(), nil)
		if err != nil {
			return err
		}
		req.Header.Set("Range", fmt.Sprintf("bytes=%d-%d", part.StartsAt(), part.StopsAt()-1))
		resp, err := http.DefaultClient.Do(req)
301
302
303
304
		if err != nil {
			return err
		}
		defer resp.Body.Close()
305

306
		n, err := io.CopyN(w, io.TeeReader(resp.Body, part), part.Size-part.Completed.Load())
307
308
309
310
311
		if err != nil && !errors.Is(err, context.Canceled) && !errors.Is(err, io.ErrUnexpectedEOF) {
			// rollback progress
			b.Completed.Add(-n)
			return err
		}
312

313
		part.Completed.Add(n)
314
315
316
317
318
		if err := b.writePart(part.Name(), part); err != nil {
			return err
		}

		// return nil or context.Canceled or UnexpectedEOF (resumable)
Michael Yang's avatar
Michael Yang committed
319
		return err
320
321
322
323
324
325
326
	})

	g.Go(func() error {
		ticker := time.NewTicker(time.Second)
		for {
			select {
			case <-ticker.C:
327
				if part.Completed.Load() >= part.Size {
328
329
330
					return nil
				}

331
332
333
334
335
				part.lastUpdatedMu.Lock()
				lastUpdated := part.lastUpdated
				part.lastUpdatedMu.Unlock()

				if !lastUpdated.IsZero() && time.Since(lastUpdated) > 5*time.Second {
336
337
					const msg = "%s part %d stalled; retrying. If this persists, press ctrl-c to exit, then 'ollama pull' to find a faster connection."
					slog.Info(fmt.Sprintf(msg, b.Digest[7:19], part.N))
338
					// reset last updated
339
					part.lastUpdatedMu.Lock()
340
					part.lastUpdated = time.Time{}
341
					part.lastUpdatedMu.Unlock()
342
343
344
345
346
347
348
					return errPartStalled
				}
			case <-ctx.Done():
				return ctx.Err()
			}
		}
	})
Michael Yang's avatar
Michael Yang committed
349

350
	return g.Wait()
351
352
}

Michael Yang's avatar
Michael Yang committed
353
354
355
356
357
358
359
360
361
362
func (b *blobDownload) newPart(offset, size int64) error {
	part := blobDownloadPart{blobDownload: b, Offset: offset, Size: size, N: len(b.Parts)}
	if err := b.writePart(part.Name(), &part); err != nil {
		return err
	}

	b.Parts = append(b.Parts, &part)
	return nil
}

363
364
365
366
367
368
369
370
371
372
373
func (b *blobDownload) readPart(partName string) (*blobDownloadPart, error) {
	var part blobDownloadPart
	partFile, err := os.Open(partName)
	if err != nil {
		return nil, err
	}
	defer partFile.Close()

	if err := json.NewDecoder(partFile).Decode(&part); err != nil {
		return nil, err
	}
374

Michael Yang's avatar
Michael Yang committed
375
	part.blobDownload = b
376
	return &part, nil
Michael Yang's avatar
Michael Yang committed
377
378
}

379
func (b *blobDownload) writePart(partName string, part *blobDownloadPart) error {
Michael Yang's avatar
Michael Yang committed
380
	partFile, err := os.OpenFile(partName, os.O_CREATE|os.O_RDWR|os.O_TRUNC, 0o644)
Michael Yang's avatar
Michael Yang committed
381
382
	if err != nil {
		return err
383
	}
Michael Yang's avatar
Michael Yang committed
384
	defer partFile.Close()
385

Michael Yang's avatar
Michael Yang committed
386
	return json.NewEncoder(partFile).Encode(part)
387
}
388

Michael Yang's avatar
Michael Yang committed
389
390
391
392
393
394
395
396
397
398
func (b *blobDownload) acquire() {
	b.references.Add(1)
}

func (b *blobDownload) release() {
	if b.references.Add(-1) == 0 {
		b.CancelFunc()
	}
}

399
func (b *blobDownload) Wait(ctx context.Context, fn func(api.ProgressResponse)) error {
Michael Yang's avatar
Michael Yang committed
400
401
	b.acquire()
	defer b.release()
402
403
404
405

	ticker := time.NewTicker(60 * time.Millisecond)
	for {
		select {
406
407
		case <-b.done:
			return b.err
408
		case <-ticker.C:
409
410
411
412
413
414
			fn(api.ProgressResponse{
				Status:    fmt.Sprintf("pulling %s", b.Digest[7:19]),
				Digest:    b.Digest,
				Total:     b.Total,
				Completed: b.Completed.Load(),
			})
415
416
417
418
419
420
421
422
423
		case <-ctx.Done():
			return ctx.Err()
		}
	}
}

type downloadOpts struct {
	mp      ModelPath
	digest  string
Michael Yang's avatar
Michael Yang committed
424
	regOpts *registryOptions
425
426
427
428
	fn      func(api.ProgressResponse)
}

// downloadBlob downloads a blob from the registry and stores it in the blobs directory
429
func downloadBlob(ctx context.Context, opts downloadOpts) (cacheHit bool, _ error) {
430
431
	fp, err := GetBlobsPath(opts.digest)
	if err != nil {
432
		return false, err
433
434
435
436
437
438
	}

	fi, err := os.Stat(fp)
	switch {
	case errors.Is(err, os.ErrNotExist):
	case err != nil:
439
		return false, err
440
441
	default:
		opts.fn(api.ProgressResponse{
Jeffrey Morgan's avatar
Jeffrey Morgan committed
442
			Status:    fmt.Sprintf("pulling %s", opts.digest[7:19]),
443
444
445
446
447
			Digest:    opts.digest,
			Total:     fi.Size(),
			Completed: fi.Size(),
		})

448
		return true, nil
449
450
	}

Michael Yang's avatar
names  
Michael Yang committed
451
452
	data, ok := blobDownloadManager.LoadOrStore(opts.digest, &blobDownload{Name: fp, Digest: opts.digest})
	download := data.(*blobDownload)
453
454
455
	if !ok {
		requestURL := opts.mp.BaseURL()
		requestURL = requestURL.JoinPath("v2", opts.mp.GetNamespaceRepository(), "blobs", opts.digest)
Michael Yang's avatar
names  
Michael Yang committed
456
		if err := download.Prepare(ctx, requestURL, opts.regOpts); err != nil {
457
			blobDownloadManager.Delete(opts.digest)
458
			return false, err
459
460
		}

Michael Yang's avatar
Michael Yang committed
461
		//nolint:contextcheck
Michael Yang's avatar
names  
Michael Yang committed
462
		go download.Run(context.Background(), requestURL, opts.regOpts)
463
464
	}

465
	return false, download.Wait(ctx, opts.fn)
466
}