/
storage.go
465 lines (426 loc) · 13.1 KB
/
storage.go
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
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
178
179
180
181
182
183
184
185
186
187
188
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
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
package fileset
import (
"context"
"math"
"strings"
"time"
units "github.com/docker/go-units"
"golang.org/x/sync/semaphore"
"github.com/pachyderm/pachyderm/v2/src/internal/dbutil"
"github.com/pachyderm/pachyderm/v2/src/internal/errors"
"github.com/pachyderm/pachyderm/v2/src/internal/errutil"
"github.com/pachyderm/pachyderm/v2/src/internal/pachsql"
"github.com/pachyderm/pachyderm/v2/src/internal/storage/chunk"
"github.com/pachyderm/pachyderm/v2/src/internal/storage/fileset/index"
"github.com/pachyderm/pachyderm/v2/src/internal/storage/renew"
"github.com/pachyderm/pachyderm/v2/src/internal/storage/track"
)
const (
// DefaultMemoryThreshold is the default for the memory threshold that must
// be met before a file set part is serialized (excluding close).
DefaultMemoryThreshold = units.GB
// DefaultCompactionLevelFactor is the default factor that level sizes increase by in a compacted fileset.
DefaultCompactionLevelFactor = 10
DefaultPrefetchLimit = 10
DefaultBatchThreshold = units.MB
// DefaultIndexCacheSize is the default size of the index cache.
DefaultIndexCacheSize = 100
// TrackerPrefix is used for creating tracker objects for filesets
TrackerPrefix = "fileset/"
// DefaultFileDatum is the default file datum.
DefaultFileDatum = "default"
)
var (
// ErrNoFileSetFound is returned by the methods on Storage when a fileset does not exist
ErrNoFileSetFound = errors.Errorf("no fileset found")
)
// Storage is an abstraction for interfacing with file sets.
// A storage instance:
// - Provides methods for writing file sets, opening file sets for reading, and managing file sets.
// - Manages tracker state to keep internal file sets alive while writing file sets.
// - Manages an internal index cache that supports logarithmic lookup in the multilevel indexes.
// - Provides methods for processing file set compaction tasks.
// - Provides a method for creating a garbage collector.
type Storage struct {
tracker track.Tracker
store MetadataStore
chunks *chunk.Storage
idxCache *index.Cache
memThreshold int64
shardConfig *index.ShardConfig
compactionConfig *CompactionConfig
filesetSem *semaphore.Weighted
prefetchLimit int
}
type CompactionConfig struct {
LevelFactor int64
}
// NewStorage creates a new Storage.
func NewStorage(mds MetadataStore, tr track.Tracker, chunks *chunk.Storage, opts ...StorageOption) *Storage {
s := &Storage{
store: mds,
tracker: tr,
chunks: chunks,
idxCache: index.NewCache(chunks, DefaultIndexCacheSize),
memThreshold: DefaultMemoryThreshold,
shardConfig: &index.ShardConfig{
NumFiles: index.DefaultShardNumThreshold,
SizeBytes: index.DefaultShardSizeThreshold,
},
compactionConfig: &CompactionConfig{
LevelFactor: DefaultCompactionLevelFactor,
},
filesetSem: semaphore.NewWeighted(math.MaxInt64),
prefetchLimit: DefaultPrefetchLimit,
}
for _, opt := range opts {
opt(s)
}
if s.compactionConfig.LevelFactor < 1 {
panic("level factor cannot be < 1")
}
return s
}
func (s *Storage) ShardConfig() *index.ShardConfig {
return s.shardConfig
}
// NewUnorderedWriter creates a new unordered file set writer.
func (s *Storage) NewUnorderedWriter(ctx context.Context, opts ...UnorderedWriterOption) (*UnorderedWriter, error) {
return newUnorderedWriter(ctx, s, s.memThreshold, s.shardConfig.NumFiles/2, opts...)
}
// NewWriter creates a new file set writer.
func (s *Storage) NewWriter(ctx context.Context, opts ...WriterOption) *Writer {
return s.newWriter(ctx, opts...)
}
func (s *Storage) newWriter(ctx context.Context, opts ...WriterOption) *Writer {
return newWriter(ctx, s, opts...)
}
func (s *Storage) newReader(id ID) *Reader {
return newReader(s.store, s.chunks, s.idxCache, id)
}
// Open opens a file set for reading.
func (s *Storage) Open(ctx context.Context, ids []ID) (FileSet, error) {
var err error
ids, err = s.FlattenAll(ctx, ids)
if err != nil {
return nil, err
}
var fss []FileSet
for _, id := range ids {
fss = append(fss, s.newReader(id))
}
if len(fss) == 0 {
return emptyFileSet{}, nil
}
if len(fss) == 1 {
return fss[0], nil
}
return newMergeReader(s.chunks, fss), nil
}
// Compose produces a composite fileset from the filesets under ids.
// It does not perform a merge or check that the filesets at ids in any way
// other than ensuring that they exist.
func (s *Storage) Compose(ctx context.Context, ids []ID, ttl time.Duration) (*ID, error) {
var result *ID
if err := dbutil.WithTx(ctx, s.store.DB(), func(ctx context.Context, tx *pachsql.Tx) error {
var err error
result, err = s.ComposeTx(tx, ids, ttl)
return err
}); err != nil {
return nil, err
}
return result, nil
}
// ComposeTx produces a composite fileset from the filesets under ids.
// It does not perform a merge or check that the filesets at ids in any way
// other than ensuring that they exist.
func (s *Storage) ComposeTx(tx *pachsql.Tx, ids []ID, ttl time.Duration) (*ID, error) {
c := &Composite{
Layers: idsToHex(ids),
}
return s.newCompositeTx(tx, c, ttl)
}
// CloneTx creates a new fileset, identical to the fileset at id, but with the specified ttl.
// The ttl can be ignored by using track.NoTTL
func (s *Storage) CloneTx(tx *pachsql.Tx, id ID, ttl time.Duration) (*ID, error) {
md, err := s.store.GetTx(tx, id)
if err != nil {
return nil, errors.EnsureStack(err)
}
switch x := md.Value.(type) {
case *Metadata_Primitive:
return s.newPrimitiveTx(tx, x.Primitive, ttl)
case *Metadata_Composite:
return s.newCompositeTx(tx, x.Composite, ttl)
default:
return nil, errors.Errorf("cannot clone type %T", md.Value)
}
}
// Flatten iterates through IDs and replaces references to composite file sets
// with all their layers in place and executes the user provided callback
// against each primitive file set.
func (s *Storage) Flatten(ctx context.Context, ids []ID, cb func(id ID) error) error {
for _, id := range ids {
md, err := s.store.Get(ctx, id)
if err != nil {
return err
}
switch x := md.Value.(type) {
case *Metadata_Primitive:
if err := cb(id); err != nil {
if errors.Is(err, errutil.ErrBreak) {
return nil
}
return err
}
case *Metadata_Composite:
ids, err := x.Composite.PointsTo()
if err != nil {
return err
}
if err := s.Flatten(ctx, ids, cb); err != nil {
if errors.Is(err, errutil.ErrBreak) {
return nil
}
return err
}
default:
// TODO: should it be?
return errors.Errorf("Flatten is not defined for empty filesets")
}
}
return nil
}
// FlattenAll is like Flatten, but collects the primitives to return to the user.
func (s *Storage) FlattenAll(ctx context.Context, ids []ID) ([]ID, error) {
flattened := make([]ID, 0, len(ids))
if err := s.Flatten(ctx, ids, func(id ID) error {
flattened = append(flattened, id)
return nil
}); err != nil {
return nil, errors.EnsureStack(err)
}
return flattened, nil
}
func (s *Storage) flattenPrimitives(ctx context.Context, ids []ID) ([]*Primitive, error) {
ids, err := s.FlattenAll(ctx, ids)
if err != nil {
return nil, err
}
return s.getPrimitives(ctx, ids)
}
func (s *Storage) getPrimitives(ctx context.Context, ids []ID) ([]*Primitive, error) {
var prims []*Primitive
for _, id := range ids {
prim, err := s.getPrimitive(ctx, id)
if err != nil {
return nil, err
}
prims = append(prims, prim)
}
return prims, nil
}
// Concat is a special case of Merge, where the filesets each contain paths for distinct ranges.
// The path ranges must be non-overlapping and the ranges must be lexigraphically sorted.
// Concat always returns the ID of a primitive fileset.
func (s *Storage) Concat(ctx context.Context, ids []ID, ttl time.Duration) (*ID, error) {
var size int64
additive := index.NewWriter(ctx, s.chunks, "additive-index-writer")
deletive := index.NewWriter(ctx, s.chunks, "deletive-index-writer")
for _, id := range ids {
md, err := s.store.Get(ctx, id)
if err != nil {
return nil, errors.EnsureStack(err)
}
prim := md.GetPrimitive()
if prim == nil {
return nil, errors.Errorf("file set %v is not primitive", id)
}
if prim.Additive != nil {
if err := additive.WriteIndex(prim.Additive); err != nil {
return nil, err
}
}
if prim.Deletive != nil {
if err := deletive.WriteIndex(prim.Deletive); err != nil {
return nil, err
}
}
size += prim.SizeBytes
}
additiveIdx, err := additive.Close()
if err != nil {
return nil, err
}
deletiveIdx, err := deletive.Close()
if err != nil {
return nil, err
}
return s.newPrimitive(ctx, &Primitive{
Additive: additiveIdx,
Deletive: deletiveIdx,
SizeBytes: size,
}, ttl)
}
// Drop allows a fileset to be deleted if it is not otherwise referenced.
func (s *Storage) Drop(ctx context.Context, id ID) error {
_, err := s.SetTTL(ctx, id, track.ExpireNow)
return err
}
// SetTTL sets the time-to-live for the fileset at id
func (s *Storage) SetTTL(ctx context.Context, id ID, ttl time.Duration) (time.Time, error) {
oid := id.TrackerID()
res, err := s.tracker.SetTTL(ctx, oid, ttl)
return res, err
}
// SizeUpperBound returns an upper bound for the size of the data in the file set in bytes.
// The upper bound is cheaper to compute than the actual size.
func (s *Storage) SizeUpperBound(ctx context.Context, id ID) (int64, error) {
prims, err := s.flattenPrimitives(ctx, []ID{id})
if err != nil {
return 0, err
}
var total int64
for _, prim := range prims {
total += prim.SizeBytes
}
return total, nil
}
// Size returns the size of the data in the file set in bytes.
func (s *Storage) Size(ctx context.Context, id ID) (int64, error) {
fs, err := s.Open(ctx, []ID{id})
if err != nil {
return 0, err
}
var total int64
if err := fs.Iterate(ctx, func(f File) error {
total += index.SizeBytes(f.Index())
return nil
}); err != nil {
return 0, err
}
return total, nil
}
// WithRenewer calls cb with a Renewer, and a context which will be canceled if the renewer is unable to renew a path.
func (s *Storage) WithRenewer(ctx context.Context, ttl time.Duration, cb func(context.Context, *Renewer) error) (retErr error) {
r := newRenewer(ctx, s, ttl)
defer func() {
if err := r.Close(); retErr == nil {
retErr = err
}
}()
return cb(r.Context(), r)
}
func (s *Storage) NewGC(d time.Duration) *track.GarbageCollector {
tmpDeleter := renew.NewTmpDeleter()
chunkDeleter := s.chunks.NewDeleter()
filesetDeleter := &deleter{
store: s.store,
}
mux := track.DeleterMux(func(id string) track.Deleter {
switch {
case strings.HasPrefix(id, renew.TmpTrackerPrefix):
return tmpDeleter
case strings.HasPrefix(id, chunk.TrackerPrefix):
return chunkDeleter
case strings.HasPrefix(id, TrackerPrefix):
return filesetDeleter
default:
return nil
}
})
return track.NewGarbageCollector(s.tracker, d, mux)
}
func (s *Storage) exists(ctx context.Context, id ID) (bool, error) {
exists, err := s.store.Exists(ctx, id)
return exists, err
}
func (s *Storage) newPrimitive(ctx context.Context, prim *Primitive, ttl time.Duration) (*ID, error) {
var result *ID
if err := dbutil.WithTx(ctx, s.store.DB(), func(ctx context.Context, tx *pachsql.Tx) error {
var err error
result, err = s.newPrimitiveTx(tx, prim, ttl)
return err
}); err != nil {
return nil, err
}
return result, nil
}
func (s *Storage) newPrimitiveTx(tx *pachsql.Tx, prim *Primitive, ttl time.Duration) (*ID, error) {
id := newID()
md := &Metadata{
Value: &Metadata_Primitive{
Primitive: prim,
},
}
var pointsTo []string
for _, chunkID := range prim.PointsTo() {
pointsTo = append(pointsTo, chunkID.TrackerID())
}
if err := s.store.SetTx(tx, id, md); err != nil {
return nil, err
}
if err := s.tracker.CreateTx(tx, id.TrackerID(), pointsTo, ttl); err != nil {
return nil, err
}
return &id, nil
}
func (s *Storage) newComposite(ctx context.Context, comp *Composite, ttl time.Duration) (*ID, error) {
var result *ID
if err := dbutil.WithTx(ctx, s.store.DB(), func(ctx context.Context, tx *pachsql.Tx) error {
var err error
result, err = s.newCompositeTx(tx, comp, ttl)
return err
}); err != nil {
return nil, err
}
return result, nil
}
func (s *Storage) newCompositeTx(tx *pachsql.Tx, comp *Composite, ttl time.Duration) (*ID, error) {
id := newID()
md := &Metadata{
Value: &Metadata_Composite{
Composite: comp,
},
}
ids, err := comp.PointsTo()
if err != nil {
return nil, err
}
var pointsTo []string
for _, id := range ids {
pointsTo = append(pointsTo, id.TrackerID())
}
if err := s.store.SetTx(tx, id, md); err != nil {
return nil, err
}
if err := s.tracker.CreateTx(tx, id.TrackerID(), pointsTo, ttl); err != nil {
return nil, err
}
return &id, nil
}
func (s *Storage) getPrimitive(ctx context.Context, id ID) (*Primitive, error) {
md, err := s.store.Get(ctx, id)
if err != nil {
return nil, err
}
prim := md.GetPrimitive()
if prim == nil {
return nil, errors.Errorf("fileset %v is not primitive", id)
}
return prim, nil
}
var _ track.Deleter = &deleter{}
type deleter struct {
store MetadataStore
}
func (d *deleter) DeleteTx(tx *pachsql.Tx, oid string) error {
if !strings.HasPrefix(oid, TrackerPrefix) {
return errors.Errorf("don't know how to delete %v", oid)
}
id, err := ParseID(oid[len(TrackerPrefix):])
if err != nil {
return err
}
return d.store.DeleteTx(tx, *id)
}