Line data Source code
1 : // Copyright 2019 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 : "io"
11 : "sort"
12 :
13 : "github.com/cockroachdb/errors"
14 : "github.com/cockroachdb/pebble/internal/base"
15 : "github.com/cockroachdb/pebble/internal/keyspan"
16 : "github.com/cockroachdb/pebble/internal/manifest"
17 : )
18 :
19 : // This file implements DB.CheckLevels() which checks that every entry in the
20 : // DB is consistent with respect to the level invariant: any point (or the
21 : // infinite number of points in a range tombstone) has a seqnum such that a
22 : // point with the same UserKey at a lower level has a lower seqnum. This is an
23 : // expensive check since it involves iterating over all the entries in the DB,
24 : // hence only intended for tests or tools.
25 : //
26 : // If we ignore range tombstones, the consistency checking of points can be
27 : // done with a simplified version of mergingIter. simpleMergingIter is that
28 : // simplified version of mergingIter that only needs to step through points
29 : // (analogous to only doing Next()). It can also easily accommodate
30 : // consistency checking of points relative to range tombstones.
31 : // simpleMergingIter does not do any seek optimizations present in mergingIter
32 : // (it minimally needs to seek the range delete iterators to position them at
33 : // or past the current point) since it does not want to miss points for
34 : // purposes of consistency checking.
35 : //
36 : // Mutual consistency of range tombstones is non-trivial to check. One needs
37 : // to detect inversions of the form [a, c)#8 at higher level and [b, c)#10 at
38 : // a lower level. The start key of the former is not contained in the latter
39 : // and we can't use the exclusive end key, c, for a containment check since it
40 : // is the sentinel key. We observe that if these tombstones were fragmented
41 : // wrt each other we would have [a, b)#8 and [b, c)#8 at the higher level and
42 : // [b, c)#10 at the lower level and then it is is trivial to compare the two
43 : // [b, c) tombstones. Note that this fragmentation needs to take into account
44 : // that tombstones in a file may be untruncated and need to act within the
45 : // bounds of the file. This checking is performed by checkRangeTombstones()
46 : // and its helper functions.
47 :
48 : // The per-level structure used by simpleMergingIter.
49 : type simpleMergingIterLevel struct {
50 : iter internalIterator
51 : rangeDelIter keyspan.FragmentIterator
52 :
53 : iterKV *base.InternalKV
54 : tombstone *keyspan.Span
55 : }
56 :
57 2 : func (ml *simpleMergingIterLevel) setRangeDelIter(iter keyspan.FragmentIterator) {
58 2 : ml.tombstone = nil
59 2 : if ml.rangeDelIter != nil {
60 2 : ml.rangeDelIter.Close()
61 2 : }
62 2 : ml.rangeDelIter = iter
63 : }
64 :
65 : type simpleMergingIter struct {
66 : levels []simpleMergingIterLevel
67 : snapshot base.SeqNum
68 : heap simpleMergingIterHeap
69 : // The last point's key and level. For validation.
70 : lastKey InternalKey
71 : lastLevel int
72 : lastIterMsg string
73 : // A non-nil valueMerger means MERGE record processing is ongoing.
74 : valueMerger base.ValueMerger
75 : // The first error will cause step() to return false.
76 : err error
77 : numPoints int64
78 : merge Merge
79 : formatKey base.FormatKey
80 : }
81 :
82 : func (m *simpleMergingIter) init(
83 : merge Merge,
84 : cmp Compare,
85 : snapshot base.SeqNum,
86 : formatKey base.FormatKey,
87 : levels ...simpleMergingIterLevel,
88 2 : ) {
89 2 : m.levels = levels
90 2 : m.formatKey = formatKey
91 2 : m.merge = merge
92 2 : m.snapshot = snapshot
93 2 : m.lastLevel = -1
94 2 : m.heap.cmp = cmp
95 2 : m.heap.items = make([]simpleMergingIterItem, 0, len(levels))
96 2 : for i := range m.levels {
97 2 : l := &m.levels[i]
98 2 : l.iterKV = l.iter.First()
99 2 : if l.iterKV != nil {
100 2 : item := simpleMergingIterItem{
101 2 : index: i,
102 2 : value: l.iterKV.V,
103 2 : }
104 2 : item.key = l.iterKV.K.Clone()
105 2 : m.heap.items = append(m.heap.items, item)
106 2 : }
107 : }
108 2 : m.heap.init()
109 2 :
110 2 : if m.heap.len() == 0 {
111 2 : return
112 2 : }
113 2 : m.positionRangeDels()
114 : }
115 :
116 : // Positions all the rangedel iterators at or past the current top of the
117 : // heap, using SeekGE().
118 2 : func (m *simpleMergingIter) positionRangeDels() {
119 2 : item := &m.heap.items[0]
120 2 : for i := range m.levels {
121 2 : l := &m.levels[i]
122 2 : if l.rangeDelIter == nil {
123 2 : continue
124 : }
125 2 : t, err := l.rangeDelIter.SeekGE(item.key.UserKey)
126 2 : m.err = firstError(m.err, err)
127 2 : l.tombstone = t
128 : }
129 : }
130 :
131 : // Returns true if not yet done.
132 2 : func (m *simpleMergingIter) step() bool {
133 2 : if m.heap.len() == 0 || m.err != nil {
134 2 : return false
135 2 : }
136 2 : item := &m.heap.items[0]
137 2 : l := &m.levels[item.index]
138 2 : // Sentinels are not relevant for this point checking.
139 2 : if !item.key.IsExclusiveSentinel() && item.key.Visible(m.snapshot, base.SeqNumMax) {
140 2 : // This is a visible point key.
141 2 : if !m.handleVisiblePoint(item, l) {
142 0 : return false
143 0 : }
144 : }
145 :
146 : // The iterator for the current level may be closed in the following call to
147 : // Next(). We save its debug string for potential use after it is closed -
148 : // either in this current step() invocation or on the next invocation.
149 2 : m.lastIterMsg = l.iter.String()
150 2 :
151 2 : // Step to the next point.
152 2 : l.iterKV = l.iter.Next()
153 2 : if l.iterKV == nil {
154 2 : m.err = errors.CombineErrors(l.iter.Error(), l.iter.Close())
155 2 : l.iter = nil
156 2 : m.heap.pop()
157 2 : } else {
158 2 : // Check point keys in an sstable are ordered. Although not required, we check
159 2 : // for memtables as well. A subtle check here is that successive sstables of
160 2 : // L1 and higher levels are ordered. This happens when levelIter moves to the
161 2 : // next sstable in the level, in which case item.key is previous sstable's
162 2 : // last point key.
163 2 : if !l.iterKV.K.IsExclusiveSentinel() && base.InternalCompare(m.heap.cmp, item.key, l.iterKV.K) >= 0 {
164 0 : m.err = errors.Errorf("out of order keys %s >= %s in %s",
165 0 : item.key.Pretty(m.formatKey), l.iterKV.K.Pretty(m.formatKey), l.iter)
166 0 : return false
167 0 : }
168 2 : item.key = base.InternalKey{
169 2 : Trailer: l.iterKV.K.Trailer,
170 2 : UserKey: append(item.key.UserKey[:0], l.iterKV.K.UserKey...),
171 2 : }
172 2 : item.value = l.iterKV.V
173 2 : if m.heap.len() > 1 {
174 2 : m.heap.fix(0)
175 2 : }
176 : }
177 2 : if m.err != nil {
178 0 : return false
179 0 : }
180 2 : if m.heap.len() == 0 {
181 2 : // If m.valueMerger != nil, the last record was a MERGE record.
182 2 : if m.valueMerger != nil {
183 2 : var closer io.Closer
184 2 : var err error
185 2 : _, closer, err = m.valueMerger.Finish(true /* includesBase */)
186 2 : if closer != nil {
187 0 : err = errors.CombineErrors(err, closer.Close())
188 0 : }
189 2 : if err != nil {
190 0 : m.err = errors.CombineErrors(m.err,
191 0 : errors.Wrapf(err, "merge processing error on key %s in %s",
192 0 : item.key.Pretty(m.formatKey), m.lastIterMsg))
193 0 : }
194 2 : m.valueMerger = nil
195 : }
196 2 : return false
197 : }
198 2 : m.positionRangeDels()
199 2 : return true
200 : }
201 :
202 : // handleVisiblePoint returns true if validation succeeded and level checking
203 : // can continue.
204 : func (m *simpleMergingIter) handleVisiblePoint(
205 : item *simpleMergingIterItem, l *simpleMergingIterLevel,
206 2 : ) (ok bool) {
207 2 : m.numPoints++
208 2 : keyChanged := m.heap.cmp(item.key.UserKey, m.lastKey.UserKey) != 0
209 2 : if !keyChanged {
210 2 : // At the same user key. We will see them in decreasing seqnum
211 2 : // order so the lastLevel must not be lower.
212 2 : if m.lastLevel > item.index {
213 0 : m.err = errors.Errorf("found InternalKey %s in %s and InternalKey %s in %s",
214 0 : item.key.Pretty(m.formatKey), l.iter, m.lastKey.Pretty(m.formatKey),
215 0 : m.lastIterMsg)
216 0 : return false
217 0 : }
218 2 : m.lastLevel = item.index
219 2 : } else {
220 2 : // The user key has changed.
221 2 : m.lastKey.Trailer = item.key.Trailer
222 2 : m.lastKey.UserKey = append(m.lastKey.UserKey[:0], item.key.UserKey...)
223 2 : m.lastLevel = item.index
224 2 : }
225 : // Ongoing series of MERGE records ends with a MERGE record.
226 2 : if keyChanged && m.valueMerger != nil {
227 2 : var closer io.Closer
228 2 : _, closer, m.err = m.valueMerger.Finish(true /* includesBase */)
229 2 : if m.err == nil && closer != nil {
230 0 : m.err = closer.Close()
231 0 : }
232 2 : m.valueMerger = nil
233 : }
234 2 : itemValue, _, err := item.value.Value(nil)
235 2 : if err != nil {
236 0 : m.err = err
237 0 : return false
238 0 : }
239 2 : if m.valueMerger != nil {
240 2 : // Ongoing series of MERGE records.
241 2 : switch item.key.Kind() {
242 2 : case InternalKeyKindSingleDelete, InternalKeyKindDelete, InternalKeyKindDeleteSized:
243 2 : var closer io.Closer
244 2 : _, closer, m.err = m.valueMerger.Finish(true /* includesBase */)
245 2 : if m.err == nil && closer != nil {
246 0 : m.err = closer.Close()
247 0 : }
248 2 : m.valueMerger = nil
249 2 : case InternalKeyKindSet, InternalKeyKindSetWithDelete:
250 2 : m.err = m.valueMerger.MergeOlder(itemValue)
251 2 : if m.err == nil {
252 2 : var closer io.Closer
253 2 : _, closer, m.err = m.valueMerger.Finish(true /* includesBase */)
254 2 : if m.err == nil && closer != nil {
255 0 : m.err = closer.Close()
256 0 : }
257 : }
258 2 : m.valueMerger = nil
259 2 : case InternalKeyKindMerge:
260 2 : m.err = m.valueMerger.MergeOlder(itemValue)
261 0 : default:
262 0 : m.err = errors.Errorf("pebble: invalid internal key kind %s in %s",
263 0 : item.key.Pretty(m.formatKey),
264 0 : l.iter)
265 0 : return false
266 : }
267 2 : } else if item.key.Kind() == InternalKeyKindMerge && m.err == nil {
268 2 : // New series of MERGE records.
269 2 : m.valueMerger, m.err = m.merge(item.key.UserKey, itemValue)
270 2 : }
271 2 : if m.err != nil {
272 0 : m.err = errors.Wrapf(m.err, "merge processing error on key %s in %s",
273 0 : item.key.Pretty(m.formatKey), l.iter)
274 0 : return false
275 0 : }
276 : // Is this point covered by a tombstone at a lower level? Note that all these
277 : // iterators must be positioned at a key > item.key.
278 2 : for level := item.index + 1; level < len(m.levels); level++ {
279 2 : lvl := &m.levels[level]
280 2 : if lvl.rangeDelIter == nil || lvl.tombstone.Empty() {
281 2 : continue
282 : }
283 2 : if lvl.tombstone.Contains(m.heap.cmp, item.key.UserKey) && lvl.tombstone.CoversAt(m.snapshot, item.key.SeqNum()) {
284 0 : m.err = errors.Errorf("tombstone %s in %s deletes key %s in %s",
285 0 : lvl.tombstone.Pretty(m.formatKey), lvl.iter, item.key.Pretty(m.formatKey),
286 0 : l.iter)
287 0 : return false
288 0 : }
289 : }
290 2 : return true
291 : }
292 :
293 : // Checking that range tombstones are mutually consistent is performed by
294 : // checkRangeTombstones(). See the overview comment at the top of the file.
295 : //
296 : // We do this check as follows:
297 : // - Collect the tombstones for each level, put them into one pool of tombstones
298 : // along with their level information (addTombstonesFromIter()).
299 : // - Collect the start and end user keys from all these tombstones
300 : // (collectAllUserKey()) and use them to fragment all the tombstones
301 : // (fragmentUsingUserKey()).
302 : // - Sort tombstones by start key and decreasing seqnum
303 : // (tombstonesByStartKeyAndSeqnum) - all tombstones that have the same start
304 : // key will have the same end key because they have been fragmented.
305 : // - Iterate and check (iterateAndCheckTombstones()).
306 : //
307 : // Note that this simple approach requires holding all the tombstones across all
308 : // levels in-memory. A more sophisticated incremental approach could be devised,
309 : // if necessary.
310 :
311 : // A tombstone and the corresponding level it was found in.
312 : type tombstoneWithLevel struct {
313 : keyspan.Span
314 : level int
315 : // The level in LSM. A -1 means it's a memtable.
316 : lsmLevel int
317 : fileNum FileNum
318 : }
319 :
320 : // For sorting tombstoneWithLevels in increasing order of start UserKey and
321 : // for the same start UserKey in decreasing order of seqnum.
322 : type tombstonesByStartKeyAndSeqnum struct {
323 : cmp Compare
324 : buf []tombstoneWithLevel
325 : }
326 :
327 2 : func (v *tombstonesByStartKeyAndSeqnum) Len() int { return len(v.buf) }
328 2 : func (v *tombstonesByStartKeyAndSeqnum) Less(i, j int) bool {
329 2 : less := v.cmp(v.buf[i].Start, v.buf[j].Start)
330 2 : if less == 0 {
331 2 : return v.buf[i].LargestSeqNum() > v.buf[j].LargestSeqNum()
332 2 : }
333 2 : return less < 0
334 : }
335 2 : func (v *tombstonesByStartKeyAndSeqnum) Swap(i, j int) {
336 2 : v.buf[i], v.buf[j] = v.buf[j], v.buf[i]
337 2 : }
338 :
339 : func iterateAndCheckTombstones(
340 : cmp Compare, formatKey base.FormatKey, tombstones []tombstoneWithLevel,
341 2 : ) error {
342 2 : sortBuf := tombstonesByStartKeyAndSeqnum{
343 2 : cmp: cmp,
344 2 : buf: tombstones,
345 2 : }
346 2 : sort.Sort(&sortBuf)
347 2 :
348 2 : // For a sequence of tombstones that share the same start UserKey, we will
349 2 : // encounter them in non-increasing seqnum order and so should encounter them
350 2 : // in non-decreasing level order.
351 2 : lastTombstone := tombstoneWithLevel{}
352 2 : for _, t := range tombstones {
353 2 : if cmp(lastTombstone.Start, t.Start) == 0 && lastTombstone.level > t.level {
354 0 : return errors.Errorf("encountered tombstone %s in %s"+
355 0 : " that has a lower seqnum than the same tombstone in %s",
356 0 : t.Span.Pretty(formatKey), levelOrMemtable(t.lsmLevel, t.fileNum),
357 0 : levelOrMemtable(lastTombstone.lsmLevel, lastTombstone.fileNum))
358 0 : }
359 2 : lastTombstone = t
360 : }
361 2 : return nil
362 : }
363 :
364 : type checkConfig struct {
365 : logger Logger
366 : comparer *Comparer
367 : readState *readState
368 : newIters tableNewIters
369 : seqNum base.SeqNum
370 : stats *CheckLevelsStats
371 : merge Merge
372 : formatKey base.FormatKey
373 : }
374 :
375 : // cmp is shorthand for comparer.Compare.
376 2 : func (c *checkConfig) cmp(a, b []byte) int { return c.comparer.Compare(a, b) }
377 :
378 2 : func checkRangeTombstones(c *checkConfig) error {
379 2 : var level int
380 2 : var tombstones []tombstoneWithLevel
381 2 : var err error
382 2 :
383 2 : memtables := c.readState.memtables
384 2 : for i := len(memtables) - 1; i >= 0; i-- {
385 2 : iter := memtables[i].newRangeDelIter(nil)
386 2 : if iter == nil {
387 2 : continue
388 : }
389 2 : tombstones, err = addTombstonesFromIter(
390 2 : iter, level, -1, 0, tombstones, c.seqNum, c.cmp, c.formatKey,
391 2 : )
392 2 : iter.Close()
393 2 : if err != nil {
394 0 : return err
395 0 : }
396 2 : level++
397 : }
398 :
399 2 : current := c.readState.current
400 2 : addTombstonesFromLevel := func(files manifest.LevelIterator, lsmLevel int) error {
401 2 : for f := files.First(); f != nil; f = files.Next() {
402 2 : lf := files.Take()
403 2 : iters, err := c.newIters(
404 2 : context.Background(), lf.FileMetadata, &IterOptions{layer: manifest.Level(lsmLevel)},
405 2 : internalIterOpts{}, iterRangeDeletions)
406 2 : if err != nil {
407 0 : return err
408 0 : }
409 2 : tombstones, err = addTombstonesFromIter(iters.RangeDeletion(), level, lsmLevel, f.FileNum,
410 2 : tombstones, c.seqNum, c.cmp, c.formatKey)
411 2 : iters.CloseAll()
412 2 :
413 2 : if err != nil {
414 0 : return err
415 0 : }
416 : }
417 2 : return nil
418 : }
419 : // Now the levels with untruncated tombsones.
420 2 : for i := len(current.L0SublevelFiles) - 1; i >= 0; i-- {
421 2 : if current.L0SublevelFiles[i].Empty() {
422 0 : continue
423 : }
424 2 : err := addTombstonesFromLevel(current.L0SublevelFiles[i].Iter(), 0)
425 2 : if err != nil {
426 0 : return err
427 0 : }
428 2 : level++
429 : }
430 2 : for i := 1; i < len(current.Levels); i++ {
431 2 : if err := addTombstonesFromLevel(current.Levels[i].Iter(), i); err != nil {
432 0 : return err
433 0 : }
434 2 : level++
435 : }
436 2 : if c.stats != nil {
437 1 : c.stats.NumTombstones = len(tombstones)
438 1 : }
439 : // We now have truncated tombstones.
440 : // Fragment them all.
441 2 : userKeys := collectAllUserKeys(c.cmp, tombstones)
442 2 : tombstones = fragmentUsingUserKeys(c.cmp, tombstones, userKeys)
443 2 : return iterateAndCheckTombstones(c.cmp, c.formatKey, tombstones)
444 : }
445 :
446 0 : func levelOrMemtable(lsmLevel int, fileNum FileNum) string {
447 0 : if lsmLevel == -1 {
448 0 : return "memtable"
449 0 : }
450 0 : return fmt.Sprintf("L%d: fileNum=%s", lsmLevel, fileNum)
451 : }
452 :
453 : func addTombstonesFromIter(
454 : iter keyspan.FragmentIterator,
455 : level int,
456 : lsmLevel int,
457 : fileNum FileNum,
458 : tombstones []tombstoneWithLevel,
459 : seqNum base.SeqNum,
460 : cmp Compare,
461 : formatKey base.FormatKey,
462 2 : ) (_ []tombstoneWithLevel, err error) {
463 2 : var prevTombstone keyspan.Span
464 2 : tomb, err := iter.First()
465 2 : for ; tomb != nil; tomb, err = iter.Next() {
466 2 : t := tomb.Visible(seqNum)
467 2 : if t.Empty() {
468 2 : continue
469 : }
470 2 : t = t.Clone()
471 2 : // This is mainly a test for rangeDelV2 formatted blocks which are expected to
472 2 : // be ordered and fragmented on disk. But we anyways check for memtables,
473 2 : // rangeDelV1 as well.
474 2 : if cmp(prevTombstone.End, t.Start) > 0 {
475 0 : return nil, errors.Errorf("unordered or unfragmented range delete tombstones %s, %s in %s",
476 0 : prevTombstone.Pretty(formatKey), t.Pretty(formatKey), levelOrMemtable(lsmLevel, fileNum))
477 0 : }
478 2 : prevTombstone = t
479 2 :
480 2 : if !t.Empty() {
481 2 : tombstones = append(tombstones, tombstoneWithLevel{
482 2 : Span: t,
483 2 : level: level,
484 2 : lsmLevel: lsmLevel,
485 2 : fileNum: fileNum,
486 2 : })
487 2 : }
488 : }
489 2 : if err != nil {
490 0 : return nil, err
491 0 : }
492 2 : return tombstones, nil
493 : }
494 :
495 : type userKeysSort struct {
496 : cmp Compare
497 : buf [][]byte
498 : }
499 :
500 2 : func (v *userKeysSort) Len() int { return len(v.buf) }
501 2 : func (v *userKeysSort) Less(i, j int) bool {
502 2 : return v.cmp(v.buf[i], v.buf[j]) < 0
503 2 : }
504 2 : func (v *userKeysSort) Swap(i, j int) {
505 2 : v.buf[i], v.buf[j] = v.buf[j], v.buf[i]
506 2 : }
507 2 : func collectAllUserKeys(cmp Compare, tombstones []tombstoneWithLevel) [][]byte {
508 2 : keys := make([][]byte, 0, len(tombstones)*2)
509 2 : for _, t := range tombstones {
510 2 : keys = append(keys, t.Start)
511 2 : keys = append(keys, t.End)
512 2 : }
513 2 : sorter := userKeysSort{
514 2 : cmp: cmp,
515 2 : buf: keys,
516 2 : }
517 2 : sort.Sort(&sorter)
518 2 : var last, curr int
519 2 : for last, curr = -1, 0; curr < len(keys); curr++ {
520 2 : if last < 0 || cmp(keys[last], keys[curr]) != 0 {
521 2 : last++
522 2 : keys[last] = keys[curr]
523 2 : }
524 : }
525 2 : keys = keys[:last+1]
526 2 : return keys
527 : }
528 :
529 : func fragmentUsingUserKeys(
530 : cmp Compare, tombstones []tombstoneWithLevel, userKeys [][]byte,
531 2 : ) []tombstoneWithLevel {
532 2 : var buf []tombstoneWithLevel
533 2 : for _, t := range tombstones {
534 2 : // Find the first position with tombstone start < user key
535 2 : i := sort.Search(len(userKeys), func(i int) bool {
536 2 : return cmp(t.Start, userKeys[i]) < 0
537 2 : })
538 2 : for ; i < len(userKeys); i++ {
539 2 : if cmp(userKeys[i], t.End) >= 0 {
540 2 : break
541 : }
542 2 : tPartial := t
543 2 : tPartial.End = userKeys[i]
544 2 : buf = append(buf, tPartial)
545 2 : t.Start = userKeys[i]
546 : }
547 2 : buf = append(buf, t)
548 : }
549 2 : return buf
550 : }
551 :
552 : // CheckLevelsStats provides basic stats on points and tombstones encountered.
553 : type CheckLevelsStats struct {
554 : NumPoints int64
555 : NumTombstones int
556 : }
557 :
558 : // CheckLevels checks:
559 : // - Every entry in the DB is consistent with the level invariant. See the
560 : // comment at the top of the file.
561 : // - Point keys in sstables are ordered.
562 : // - Range delete tombstones in sstables are ordered and fragmented.
563 : // - Successful processing of all MERGE records.
564 2 : func (d *DB) CheckLevels(stats *CheckLevelsStats) error {
565 2 : // Grab and reference the current readState.
566 2 : readState := d.loadReadState()
567 2 : defer readState.unref()
568 2 :
569 2 : // Determine the seqnum to read at after grabbing the read state (current and
570 2 : // memtables) above.
571 2 : seqNum := d.mu.versions.visibleSeqNum.Load()
572 2 :
573 2 : checkConfig := &checkConfig{
574 2 : logger: d.opts.Logger,
575 2 : comparer: d.opts.Comparer,
576 2 : readState: readState,
577 2 : newIters: d.newIters,
578 2 : seqNum: seqNum,
579 2 : stats: stats,
580 2 : merge: d.merge,
581 2 : formatKey: d.opts.Comparer.FormatKey,
582 2 : }
583 2 : return checkLevelsInternal(checkConfig)
584 2 : }
585 :
586 2 : func checkLevelsInternal(c *checkConfig) (err error) {
587 2 : // Phase 1: Use a simpleMergingIter to step through all the points and ensure
588 2 : // that points with the same user key at different levels are not inverted
589 2 : // wrt sequence numbers and the same holds for tombstones that cover points.
590 2 : // To do this, one needs to construct a simpleMergingIter which is similar to
591 2 : // how one constructs a mergingIter.
592 2 :
593 2 : // Add mem tables from newest to oldest.
594 2 : var mlevels []simpleMergingIterLevel
595 2 : defer func() {
596 2 : for i := range mlevels {
597 2 : l := &mlevels[i]
598 2 : if l.iter != nil {
599 2 : err = firstError(err, l.iter.Close())
600 2 : l.iter = nil
601 2 : }
602 2 : if l.rangeDelIter != nil {
603 2 : l.rangeDelIter.Close()
604 2 : l.rangeDelIter = nil
605 2 : }
606 : }
607 : }()
608 :
609 2 : memtables := c.readState.memtables
610 2 : for i := len(memtables) - 1; i >= 0; i-- {
611 2 : mem := memtables[i]
612 2 : mlevels = append(mlevels, simpleMergingIterLevel{
613 2 : iter: mem.newIter(nil),
614 2 : rangeDelIter: mem.newRangeDelIter(nil),
615 2 : })
616 2 : }
617 :
618 2 : current := c.readState.current
619 2 : // Determine the final size for mlevels so that there are no more
620 2 : // reallocations. levelIter will hold a pointer to elements in mlevels.
621 2 : start := len(mlevels)
622 2 : for sublevel := len(current.L0SublevelFiles) - 1; sublevel >= 0; sublevel-- {
623 2 : if current.L0SublevelFiles[sublevel].Empty() {
624 0 : continue
625 : }
626 2 : mlevels = append(mlevels, simpleMergingIterLevel{})
627 : }
628 2 : for level := 1; level < len(current.Levels); level++ {
629 2 : if current.Levels[level].Empty() {
630 2 : continue
631 : }
632 2 : mlevels = append(mlevels, simpleMergingIterLevel{})
633 : }
634 2 : mlevelAlloc := mlevels[start:]
635 2 : // Add L0 files by sublevel.
636 2 : for sublevel := len(current.L0SublevelFiles) - 1; sublevel >= 0; sublevel-- {
637 2 : if current.L0SublevelFiles[sublevel].Empty() {
638 0 : continue
639 : }
640 2 : manifestIter := current.L0SublevelFiles[sublevel].Iter()
641 2 : iterOpts := IterOptions{logger: c.logger}
642 2 : li := &levelIter{}
643 2 : li.init(context.Background(), iterOpts, c.comparer, c.newIters, manifestIter,
644 2 : manifest.L0Sublevel(sublevel), internalIterOpts{})
645 2 : li.initRangeDel(&mlevelAlloc[0])
646 2 : mlevelAlloc[0].iter = li
647 2 : mlevelAlloc = mlevelAlloc[1:]
648 : }
649 2 : for level := 1; level < len(current.Levels); level++ {
650 2 : if current.Levels[level].Empty() {
651 2 : continue
652 : }
653 :
654 2 : iterOpts := IterOptions{logger: c.logger}
655 2 : li := &levelIter{}
656 2 : li.init(context.Background(), iterOpts, c.comparer, c.newIters,
657 2 : current.Levels[level].Iter(), manifest.Level(level), internalIterOpts{})
658 2 : li.initRangeDel(&mlevelAlloc[0])
659 2 : mlevelAlloc[0].iter = li
660 2 : mlevelAlloc = mlevelAlloc[1:]
661 : }
662 :
663 2 : mergingIter := &simpleMergingIter{}
664 2 : mergingIter.init(c.merge, c.cmp, c.seqNum, c.formatKey, mlevels...)
665 2 : for cont := mergingIter.step(); cont; cont = mergingIter.step() {
666 2 : }
667 2 : if err := mergingIter.err; err != nil {
668 0 : return err
669 0 : }
670 2 : if c.stats != nil {
671 1 : c.stats.NumPoints = mergingIter.numPoints
672 1 : }
673 :
674 : // Phase 2: Check that the tombstones are mutually consistent.
675 2 : return checkRangeTombstones(c)
676 : }
677 :
678 : type simpleMergingIterItem struct {
679 : index int
680 : key InternalKey
681 : value base.LazyValue
682 : }
683 :
684 : type simpleMergingIterHeap struct {
685 : cmp Compare
686 : reverse bool
687 : items []simpleMergingIterItem
688 : }
689 :
690 2 : func (h *simpleMergingIterHeap) len() int {
691 2 : return len(h.items)
692 2 : }
693 :
694 2 : func (h *simpleMergingIterHeap) less(i, j int) bool {
695 2 : ikey, jkey := h.items[i].key, h.items[j].key
696 2 : if c := h.cmp(ikey.UserKey, jkey.UserKey); c != 0 {
697 2 : if h.reverse {
698 0 : return c > 0
699 0 : }
700 2 : return c < 0
701 : }
702 2 : if h.reverse {
703 0 : return ikey.Trailer < jkey.Trailer
704 0 : }
705 2 : return ikey.Trailer > jkey.Trailer
706 : }
707 :
708 2 : func (h *simpleMergingIterHeap) swap(i, j int) {
709 2 : h.items[i], h.items[j] = h.items[j], h.items[i]
710 2 : }
711 :
712 : // init, fix, up and down are copied from the go stdlib.
713 2 : func (h *simpleMergingIterHeap) init() {
714 2 : // heapify
715 2 : n := h.len()
716 2 : for i := n/2 - 1; i >= 0; i-- {
717 2 : h.down(i, n)
718 2 : }
719 : }
720 :
721 2 : func (h *simpleMergingIterHeap) fix(i int) {
722 2 : if !h.down(i, h.len()) {
723 2 : h.up(i)
724 2 : }
725 : }
726 :
727 2 : func (h *simpleMergingIterHeap) pop() *simpleMergingIterItem {
728 2 : n := h.len() - 1
729 2 : h.swap(0, n)
730 2 : h.down(0, n)
731 2 : item := &h.items[n]
732 2 : h.items = h.items[:n]
733 2 : return item
734 2 : }
735 :
736 2 : func (h *simpleMergingIterHeap) up(j int) {
737 2 : for {
738 2 : i := (j - 1) / 2 // parent
739 2 : if i == j || !h.less(j, i) {
740 2 : break
741 : }
742 0 : h.swap(i, j)
743 0 : j = i
744 : }
745 : }
746 :
747 2 : func (h *simpleMergingIterHeap) down(i0, n int) bool {
748 2 : i := i0
749 2 : for {
750 2 : j1 := 2*i + 1
751 2 : if j1 >= n || j1 < 0 { // j1 < 0 after int overflow
752 2 : break
753 : }
754 2 : j := j1 // left child
755 2 : if j2 := j1 + 1; j2 < n && h.less(j2, j1) {
756 2 : j = j2 // = 2*i + 2 // right child
757 2 : }
758 2 : if !h.less(j, i) {
759 2 : break
760 : }
761 2 : h.swap(i, j)
762 2 : i = j
763 : }
764 2 : return i > i0
765 : }
|
