Line data Source code
1 : // Copyright 2020 The LevelDB-Go and Pebble Authors. All rights reserved. Use
2 : // of this source code is governed by a BSD-style license that can be found in
3 : // the LICENSE file.
4 :
5 : package pebble
6 :
7 : import (
8 : "context"
9 : "fmt"
10 : "sync/atomic"
11 : "time"
12 :
13 : "github.com/cockroachdb/errors"
14 : "github.com/cockroachdb/pebble/internal/base"
15 : "github.com/cockroachdb/pebble/internal/invariants"
16 : "github.com/cockroachdb/pebble/internal/keyspan"
17 : "github.com/cockroachdb/pebble/internal/keyspan/keyspanimpl"
18 : "github.com/cockroachdb/pebble/internal/manifest"
19 : )
20 :
21 : // flushable defines the interface for immutable memtables.
22 : type flushable interface {
23 : newIter(o *IterOptions) internalIterator
24 : newFlushIter(o *IterOptions) internalIterator
25 : newRangeDelIter(o *IterOptions) keyspan.FragmentIterator
26 : newRangeKeyIter(o *IterOptions) keyspan.FragmentIterator
27 : containsRangeKeys() bool
28 : // inuseBytes returns the number of inuse bytes by the flushable.
29 : inuseBytes() uint64
30 : // totalBytes returns the total number of bytes allocated by the flushable.
31 : totalBytes() uint64
32 : // readyForFlush returns true when the flushable is ready for flushing. See
33 : // memTable.readyForFlush for one implementation which needs to check whether
34 : // there are any outstanding write references.
35 : readyForFlush() bool
36 : // computePossibleOverlaps determines whether the flushable's keys overlap
37 : // with the bounds of any of the provided bounded items. If an item overlaps
38 : // or might overlap but it's not possible to determine overlap cheaply,
39 : // computePossibleOverlaps invokes the provided function with the object
40 : // that might overlap. computePossibleOverlaps must not perform any I/O and
41 : // implementations should invoke the provided function for items that would
42 : // require I/O to determine overlap.
43 : computePossibleOverlaps(overlaps func(bounded) shouldContinue, bounded ...bounded)
44 : }
45 :
46 : type shouldContinue bool
47 :
48 : const (
49 : continueIteration shouldContinue = true
50 : stopIteration = false
51 : )
52 :
53 : type bounded interface {
54 : UserKeyBounds() base.UserKeyBounds
55 : }
56 :
57 : var _ bounded = (*fileMetadata)(nil)
58 : var _ bounded = KeyRange{}
59 :
60 2 : func sliceAsBounded[B bounded](s []B) []bounded {
61 2 : ret := make([]bounded, len(s))
62 2 : for i := 0; i < len(s); i++ {
63 2 : ret[i] = s[i]
64 2 : }
65 2 : return ret
66 : }
67 :
68 : // flushableEntry wraps a flushable and adds additional metadata and
69 : // functionality that is common to all flushables.
70 : type flushableEntry struct {
71 : flushable
72 : // Channel which is closed when the flushable has been flushed.
73 : flushed chan struct{}
74 : // flushForced indicates whether a flush was forced on this memtable (either
75 : // manual, or due to ingestion). Protected by DB.mu.
76 : flushForced bool
77 : // delayedFlushForcedAt indicates whether a timer has been set to force a
78 : // flush on this memtable at some point in the future. Protected by DB.mu.
79 : // Holds the timestamp of when the flush will be issued.
80 : delayedFlushForcedAt time.Time
81 : // logNum corresponds to the WAL that contains the records present in the
82 : // receiver.
83 : logNum base.DiskFileNum
84 : // logSize is the size in bytes of the associated WAL. Protected by DB.mu.
85 : logSize uint64
86 : // The current logSeqNum at the time the memtable was created. This is
87 : // guaranteed to be less than or equal to any seqnum stored in the memtable.
88 : logSeqNum base.SeqNum
89 : // readerRefs tracks the read references on the flushable. The two sources of
90 : // reader references are DB.mu.mem.queue and readState.memtables. The memory
91 : // reserved by the flushable in the cache is released when the reader refs
92 : // drop to zero. If the flushable is referencing sstables, then the file
93 : // refount is also decreased once the reader refs drops to 0. If the
94 : // flushable is a memTable, when the reader refs drops to zero, the writer
95 : // refs will already be zero because the memtable will have been flushed and
96 : // that only occurs once the writer refs drops to zero.
97 : readerRefs atomic.Int32
98 : // Closure to invoke to release memory accounting.
99 : releaseMemAccounting func()
100 : // unrefFiles, if not nil, should be invoked to decrease the ref count of
101 : // files which are backing the flushable.
102 : unrefFiles func() []*fileBacking
103 : // deleteFnLocked should be called if the caller is holding DB.mu.
104 : deleteFnLocked func(obsolete []*fileBacking)
105 : // deleteFn should be called if the caller is not holding DB.mu.
106 : deleteFn func(obsolete []*fileBacking)
107 : }
108 :
109 2 : func (e *flushableEntry) readerRef() {
110 2 : switch v := e.readerRefs.Add(1); {
111 0 : case v <= 1:
112 0 : panic(fmt.Sprintf("pebble: inconsistent reference count: %d", v))
113 : }
114 : }
115 :
116 : // db.mu must not be held when this is called.
117 2 : func (e *flushableEntry) readerUnref(deleteFiles bool) {
118 2 : e.readerUnrefHelper(deleteFiles, e.deleteFn)
119 2 : }
120 :
121 : // db.mu must be held when this is called.
122 2 : func (e *flushableEntry) readerUnrefLocked(deleteFiles bool) {
123 2 : e.readerUnrefHelper(deleteFiles, e.deleteFnLocked)
124 2 : }
125 :
126 : func (e *flushableEntry) readerUnrefHelper(
127 : deleteFiles bool, deleteFn func(obsolete []*fileBacking),
128 2 : ) {
129 2 : switch v := e.readerRefs.Add(-1); {
130 0 : case v < 0:
131 0 : panic(fmt.Sprintf("pebble: inconsistent reference count: %d", v))
132 2 : case v == 0:
133 2 : if e.releaseMemAccounting == nil {
134 0 : panic("pebble: memtable reservation already released")
135 : }
136 2 : e.releaseMemAccounting()
137 2 : e.releaseMemAccounting = nil
138 2 : if e.unrefFiles != nil {
139 2 : obsolete := e.unrefFiles()
140 2 : e.unrefFiles = nil
141 2 : if deleteFiles {
142 2 : deleteFn(obsolete)
143 2 : }
144 : }
145 : }
146 : }
147 :
148 : type flushableList []*flushableEntry
149 :
150 : // ingestedFlushable is the implementation of the flushable interface for the
151 : // ingesting sstables which are added to the flushable list.
152 : type ingestedFlushable struct {
153 : // files are non-overlapping and ordered (according to their bounds).
154 : files []physicalMeta
155 : comparer *Comparer
156 : newIters tableNewIters
157 : newRangeKeyIters keyspanimpl.TableNewSpanIter
158 :
159 : // Since the level slice is immutable, we construct and set it once. It
160 : // should be safe to read from slice in future reads.
161 : slice manifest.LevelSlice
162 : // hasRangeKeys is set on ingestedFlushable construction.
163 : hasRangeKeys bool
164 : // exciseSpan is populated if an excise operation should be performed during
165 : // flush.
166 : exciseSpan KeyRange
167 : exciseSeqNum base.SeqNum
168 : }
169 :
170 : func newIngestedFlushable(
171 : files []*fileMetadata,
172 : comparer *Comparer,
173 : newIters tableNewIters,
174 : newRangeKeyIters keyspanimpl.TableNewSpanIter,
175 : exciseSpan KeyRange,
176 : seqNum base.SeqNum,
177 2 : ) *ingestedFlushable {
178 2 : if invariants.Enabled {
179 2 : for i := 1; i < len(files); i++ {
180 2 : prev := files[i-1].UserKeyBounds()
181 2 : this := files[i].UserKeyBounds()
182 2 : if prev.End.IsUpperBoundFor(comparer.Compare, this.Start) {
183 0 : panic(errors.AssertionFailedf("ingested flushable files overlap: %s %s", prev, this))
184 : }
185 : }
186 : }
187 2 : var physicalFiles []physicalMeta
188 2 : var hasRangeKeys bool
189 2 : for _, f := range files {
190 2 : if f.HasRangeKeys {
191 2 : hasRangeKeys = true
192 2 : }
193 2 : physicalFiles = append(physicalFiles, f.PhysicalMeta())
194 : }
195 :
196 2 : ret := &ingestedFlushable{
197 2 : files: physicalFiles,
198 2 : comparer: comparer,
199 2 : newIters: newIters,
200 2 : newRangeKeyIters: newRangeKeyIters,
201 2 : // slice is immutable and can be set once and used many times.
202 2 : slice: manifest.NewLevelSliceKeySorted(comparer.Compare, files),
203 2 : hasRangeKeys: hasRangeKeys,
204 2 : exciseSpan: exciseSpan,
205 2 : exciseSeqNum: seqNum,
206 2 : }
207 2 :
208 2 : return ret
209 : }
210 :
211 : // TODO(sumeer): ingestedFlushable iters also need to plumb context for
212 : // tracing.
213 :
214 : // newIter is part of the flushable interface.
215 2 : func (s *ingestedFlushable) newIter(o *IterOptions) internalIterator {
216 2 : var opts IterOptions
217 2 : if o != nil {
218 2 : opts = *o
219 2 : }
220 2 : return newLevelIter(
221 2 : context.Background(), opts, s.comparer, s.newIters, s.slice.Iter(), manifest.FlushableIngestsLayer(),
222 2 : internalIterOpts{},
223 2 : )
224 : }
225 :
226 : // newFlushIter is part of the flushable interface.
227 0 : func (s *ingestedFlushable) newFlushIter(*IterOptions) internalIterator {
228 0 : // newFlushIter is only used for writing memtables to disk as sstables.
229 0 : // Since ingested sstables are already present on disk, they don't need to
230 0 : // make use of a flush iter.
231 0 : panic("pebble: not implemented")
232 : }
233 :
234 : func (s *ingestedFlushable) constructRangeDelIter(
235 : ctx context.Context, file *manifest.FileMetadata, _ keyspan.SpanIterOptions,
236 2 : ) (keyspan.FragmentIterator, error) {
237 2 : iters, err := s.newIters(ctx, file, nil, internalIterOpts{}, iterRangeDeletions)
238 2 : if err != nil {
239 0 : return nil, err
240 0 : }
241 2 : return iters.RangeDeletion(), nil
242 : }
243 :
244 : // newRangeDelIter is part of the flushable interface.
245 : // TODO(bananabrick): Using a level iter instead of a keyspan level iter to
246 : // surface range deletes is more efficient.
247 : //
248 : // TODO(sumeer): *IterOptions are being ignored, so the index block load for
249 : // the point iterator in constructRangeDeIter is not tracked.
250 2 : func (s *ingestedFlushable) newRangeDelIter(_ *IterOptions) keyspan.FragmentIterator {
251 2 : liter := keyspanimpl.NewLevelIter(
252 2 : context.TODO(),
253 2 : keyspan.SpanIterOptions{}, s.comparer.Compare,
254 2 : s.constructRangeDelIter, s.slice.Iter(), manifest.FlushableIngestsLayer(),
255 2 : manifest.KeyTypePoint,
256 2 : )
257 2 : if !s.exciseSpan.Valid() {
258 2 : return liter
259 2 : }
260 : // We have an excise span to weave into the rangedel iterators.
261 : //
262 : // TODO(bilal): should this be pooled?
263 2 : miter := &keyspanimpl.MergingIter{}
264 2 : rdel := keyspan.Span{
265 2 : Start: s.exciseSpan.Start,
266 2 : End: s.exciseSpan.End,
267 2 : Keys: []keyspan.Key{{Trailer: base.MakeTrailer(s.exciseSeqNum, base.InternalKeyKindRangeDelete)}},
268 2 : }
269 2 : rdelIter := keyspan.NewIter(s.comparer.Compare, []keyspan.Span{rdel})
270 2 : miter.Init(s.comparer, keyspan.NoopTransform, new(keyspanimpl.MergingBuffers), liter, rdelIter)
271 2 : return miter
272 : }
273 :
274 : // newRangeKeyIter is part of the flushable interface.
275 2 : func (s *ingestedFlushable) newRangeKeyIter(o *IterOptions) keyspan.FragmentIterator {
276 2 : var rkeydelIter keyspan.FragmentIterator
277 2 : if s.exciseSpan.Valid() {
278 2 : // We have an excise span to weave into the rangekey iterators.
279 2 : rkeydel := keyspan.Span{
280 2 : Start: s.exciseSpan.Start,
281 2 : End: s.exciseSpan.End,
282 2 : Keys: []keyspan.Key{{Trailer: base.MakeTrailer(s.exciseSeqNum, base.InternalKeyKindRangeKeyDelete)}},
283 2 : }
284 2 : rkeydelIter = keyspan.NewIter(s.comparer.Compare, []keyspan.Span{rkeydel})
285 2 : }
286 :
287 2 : if !s.hasRangeKeys {
288 2 : if rkeydelIter == nil {
289 2 : // NB: we have to return the nil literal as opposed to the nil
290 2 : // value of rkeydelIter, otherwise callers of this function will
291 2 : // have the return value fail == nil checks.
292 2 : return nil
293 2 : }
294 1 : return rkeydelIter
295 : }
296 :
297 2 : liter := keyspanimpl.NewLevelIter(
298 2 : context.TODO(),
299 2 : keyspan.SpanIterOptions{}, s.comparer.Compare, s.newRangeKeyIters,
300 2 : s.slice.Iter(), manifest.FlushableIngestsLayer(), manifest.KeyTypeRange,
301 2 : )
302 2 : if rkeydelIter == nil {
303 2 : return liter
304 2 : }
305 : // TODO(bilal): should this be pooled?
306 2 : miter := &keyspanimpl.MergingIter{}
307 2 : miter.Init(s.comparer, keyspan.NoopTransform, new(keyspanimpl.MergingBuffers), liter, rkeydelIter)
308 2 : return miter
309 : }
310 :
311 : // containsRangeKeys is part of the flushable interface.
312 2 : func (s *ingestedFlushable) containsRangeKeys() bool {
313 2 : return s.hasRangeKeys || s.exciseSpan.Valid()
314 2 : }
315 :
316 : // inuseBytes is part of the flushable interface.
317 0 : func (s *ingestedFlushable) inuseBytes() uint64 {
318 0 : // inuseBytes is only used when memtables are flushed to disk as sstables.
319 0 : panic("pebble: not implemented")
320 : }
321 :
322 : // totalBytes is part of the flushable interface.
323 2 : func (s *ingestedFlushable) totalBytes() uint64 {
324 2 : // We don't allocate additional bytes for the ingestedFlushable.
325 2 : return 0
326 2 : }
327 :
328 : // readyForFlush is part of the flushable interface.
329 2 : func (s *ingestedFlushable) readyForFlush() bool {
330 2 : // ingestedFlushable should always be ready to flush. However, note that
331 2 : // memtables before the ingested sstables in the memtable queue must be
332 2 : // flushed before an ingestedFlushable can be flushed. This is because the
333 2 : // ingested sstables need an updated view of the Version to
334 2 : // determine where to place the files in the lsm.
335 2 : return true
336 2 : }
337 :
338 : // computePossibleOverlaps is part of the flushable interface.
339 : func (s *ingestedFlushable) computePossibleOverlaps(
340 : fn func(bounded) shouldContinue, bounded ...bounded,
341 2 : ) {
342 2 : for _, b := range bounded {
343 2 : if s.anyFileOverlaps(b.UserKeyBounds()) {
344 2 : // Some file overlaps in key boundaries. The file doesn't necessarily
345 2 : // contain any keys within the key range, but we would need to perform I/O
346 2 : // to know for sure. The flushable interface dictates that we're not
347 2 : // permitted to perform I/O here, so err towards assuming overlap.
348 2 : if !fn(b) {
349 2 : return
350 2 : }
351 : }
352 : }
353 : }
354 :
355 : // anyFileBoundsOverlap returns true if there is at least a file in s.files with
356 : // bounds that overlap the given bounds.
357 2 : func (s *ingestedFlushable) anyFileOverlaps(bounds base.UserKeyBounds) bool {
358 2 : // Note that s.files are non-overlapping and sorted.
359 2 : for _, f := range s.files {
360 2 : fileBounds := f.UserKeyBounds()
361 2 : if !fileBounds.End.IsUpperBoundFor(s.comparer.Compare, bounds.Start) {
362 2 : // The file ends before the bounds start. Go to the next file.
363 2 : continue
364 : }
365 2 : if !bounds.End.IsUpperBoundFor(s.comparer.Compare, fileBounds.Start) {
366 2 : // The file starts after the bounds end. There is no overlap, and
367 2 : // further files will not overlap either (the files are sorted).
368 2 : break
369 : }
370 : // There is overlap. Note that UserKeyBounds.Overlaps() performs exactly the
371 : // checks above.
372 2 : return true
373 : }
374 2 : if s.exciseSpan.Valid() {
375 2 : uk := s.exciseSpan.UserKeyBounds()
376 2 : return uk.Overlaps(s.comparer.Compare, &bounds)
377 2 : }
378 2 : return false
379 : }
380 :
381 : // computePossibleOverlapsGenericImpl is an implementation of the flushable
382 : // interface's computePossibleOverlaps function for flushable implementations
383 : // with only in-memory state that do not have special requirements and should
384 : // read through the ordinary flushable iterators.
385 : //
386 : // This function must only be used with implementations that are infallible (eg,
387 : // memtable iterators) and will panic if an error is encountered.
388 : func computePossibleOverlapsGenericImpl[F flushable](
389 : f F, cmp Compare, fn func(bounded) shouldContinue, bounded []bounded,
390 2 : ) {
391 2 : iter := f.newIter(nil)
392 2 : rangeDelIter := f.newRangeDelIter(nil)
393 2 : rangeKeyIter := f.newRangeKeyIter(nil)
394 2 : for _, b := range bounded {
395 2 : overlap, err := determineOverlapAllIters(cmp, b.UserKeyBounds(), iter, rangeDelIter, rangeKeyIter)
396 2 : if invariants.Enabled && err != nil {
397 0 : panic(errors.AssertionFailedf("expected iterator to be infallible: %v", err))
398 : }
399 2 : if overlap {
400 2 : if !fn(b) {
401 2 : break
402 : }
403 : }
404 : }
405 :
406 2 : if iter != nil {
407 2 : if err := iter.Close(); err != nil {
408 0 : // This implementation must be used in circumstances where
409 0 : // reading through the iterator is infallible.
410 0 : panic(err)
411 : }
412 : }
413 2 : if rangeDelIter != nil {
414 2 : rangeDelIter.Close()
415 2 : }
416 2 : if rangeKeyIter != nil {
417 2 : rangeKeyIter.Close()
418 2 : }
419 : }
420 :
421 : // determineOverlapAllIters checks for overlap in a point iterator, range
422 : // deletion iterator and range key iterator.
423 : func determineOverlapAllIters(
424 : cmp base.Compare,
425 : bounds base.UserKeyBounds,
426 : pointIter base.InternalIterator,
427 : rangeDelIter, rangeKeyIter keyspan.FragmentIterator,
428 2 : ) (bool, error) {
429 2 : if pointIter != nil {
430 2 : if pointOverlap, err := determineOverlapPointIterator(cmp, bounds, pointIter); pointOverlap || err != nil {
431 2 : return pointOverlap, err
432 2 : }
433 : }
434 2 : if rangeDelIter != nil {
435 2 : if rangeDelOverlap, err := determineOverlapKeyspanIterator(cmp, bounds, rangeDelIter); rangeDelOverlap || err != nil {
436 2 : return rangeDelOverlap, err
437 2 : }
438 : }
439 2 : if rangeKeyIter != nil {
440 2 : return determineOverlapKeyspanIterator(cmp, bounds, rangeKeyIter)
441 2 : }
442 2 : return false, nil
443 : }
444 :
445 : func determineOverlapPointIterator(
446 : cmp base.Compare, bounds base.UserKeyBounds, iter internalIterator,
447 2 : ) (bool, error) {
448 2 : kv := iter.SeekGE(bounds.Start, base.SeekGEFlagsNone)
449 2 : if kv == nil {
450 2 : return false, iter.Error()
451 2 : }
452 2 : return bounds.End.IsUpperBoundForInternalKey(cmp, kv.K), nil
453 : }
454 :
455 : func determineOverlapKeyspanIterator(
456 : cmp base.Compare, bounds base.UserKeyBounds, iter keyspan.FragmentIterator,
457 2 : ) (bool, error) {
458 2 : // NB: The spans surfaced by the fragment iterator are non-overlapping.
459 2 : span, err := iter.SeekGE(bounds.Start)
460 2 : if err != nil {
461 0 : return false, err
462 0 : }
463 2 : for ; span != nil; span, err = iter.Next() {
464 2 : if !bounds.End.IsUpperBoundFor(cmp, span.Start) {
465 2 : // The span starts after our bounds.
466 2 : return false, nil
467 2 : }
468 2 : if !span.Empty() {
469 2 : return true, nil
470 2 : }
471 : }
472 2 : return false, err
473 : }
|
