// range is a keyword in Go, add Keys suffix.
func (s *store) rangeKeys(key, end []byte, limit, rangeRev int64, countOnly bool) (kvs []mvccpb.KeyValue, count int, curRev int64, err error) {
curRev = int64(s.currentRev.main)
if s.currentRev.sub > 0 {
curRev += 1
}
if rangeRev > curRev {
return nil, -1, s.currentRev.main, ErrFutureRev
}
var rev int64
if rangeRev <= 0 {
rev = curRev
} else {
rev = rangeRev
}
if rev < s.compactMainRev {
return nil, -1, 0, ErrCompacted
}
_, revpairs := s.kvindex.Range(key, end, int64(rev))
if len(revpairs) == 0 {
return nil, 0, curRev, nil
}
if countOnly {
return nil, len(revpairs), curRev, nil
}
for _, revpair := range revpairs {
start, end := revBytesRange(revpair)
_, vs := s.tx.UnsafeRange(keyBucketName, start, end, 0)
if len(vs) != 1 {
plog.Fatalf("range cannot find rev (%d,%d)", revpair.main, revpair.sub)
}
var kv mvccpb.KeyValue
if err := kv.Unmarshal(vs[0]); err != nil {
plog.Fatalf("cannot unmarshal event: %v", err)
}
kvs = append(kvs, kv)
if limit > 0 && len(kvs) >= int(limit) {
break
}
}
return kvs, len(revpairs), curRev, nil
}
// kvsToEvents gets all events for the watchers from all key-value pairs
func kvsToEvents(wg *watcherGroup, revs, vals [][]byte) (evs []mvccpb.Event) {
for i, v := range vals {
var kv mvccpb.KeyValue
if err := kv.Unmarshal(v); err != nil {
plog.Panicf("cannot unmarshal event: %v", err)
}
if !wg.contains(string(kv.Key)) {
continue
}
ty := mvccpb.PUT
if isTombstone(revs[i]) {
ty = mvccpb.DELETE
// patch in mod revision so watchers won't skip
kv.ModRevision = bytesToRev(revs[i]).main
}
evs = append(evs, mvccpb.Event{Kv: &kv, Type: ty})
}
return evs
}
func (s *store) delete(key []byte, rev revision) {
mainrev := s.currentRev.main + 1
ibytes := newRevBytes()
revToBytes(revision{main: mainrev, sub: s.currentRev.sub}, ibytes)
ibytes = appendMarkTombstone(ibytes)
kv := mvccpb.KeyValue{
Key: key,
}
d, err := kv.Marshal()
if err != nil {
plog.Fatalf("cannot marshal event: %v", err)
}
s.tx.UnsafeSeqPut(keyBucketName, ibytes, d)
err = s.kvindex.Tombstone(key, revision{main: mainrev, sub: s.currentRev.sub})
if err != nil {
plog.Fatalf("cannot tombstone an existing key (%s): %v", string(key), err)
}
s.changes = append(s.changes, kv)
s.currentRev.sub += 1
ibytes = newRevBytes()
revToBytes(rev, ibytes)
_, vs := s.tx.UnsafeRange(keyBucketName, ibytes, nil, 0)
kv.Reset()
if err = kv.Unmarshal(vs[0]); err != nil {
plog.Fatalf("cannot unmarshal value: %v", err)
}
if lease.LeaseID(kv.Lease) != lease.NoLease {
err = s.le.Detach(lease.LeaseID(kv.Lease), []lease.LeaseItem{{Key: string(kv.Key)}})
if err != nil {
plog.Fatalf("cannot detach %v", err)
}
}
}
func (s *store) put(key, value []byte, leaseID lease.LeaseID) {
rev := s.currentRev.main + 1
c := rev
oldLease := lease.NoLease
// if the key exists before, use its previous created and
// get its previous leaseID
grev, created, ver, err := s.kvindex.Get(key, rev)
if err == nil {
c = created.main
ibytes := newRevBytes()
revToBytes(grev, ibytes)
_, vs := s.tx.UnsafeRange(keyBucketName, ibytes, nil, 0)
var kv mvccpb.KeyValue
if err = kv.Unmarshal(vs[0]); err != nil {
plog.Fatalf("cannot unmarshal value: %v", err)
}
oldLease = lease.LeaseID(kv.Lease)
}
ibytes := newRevBytes()
revToBytes(revision{main: rev, sub: s.currentRev.sub}, ibytes)
ver = ver + 1
kv := mvccpb.KeyValue{
Key: key,
Value: value,
CreateRevision: c,
ModRevision: rev,
Version: ver,
Lease: int64(leaseID),
}
d, err := kv.Marshal()
if err != nil {
plog.Fatalf("cannot marshal event: %v", err)
}
s.tx.UnsafeSeqPut(keyBucketName, ibytes, d)
s.kvindex.Put(key, revision{main: rev, sub: s.currentRev.sub})
s.changes = append(s.changes, kv)
s.currentRev.sub += 1
if oldLease != lease.NoLease {
if s.le == nil {
panic("no lessor to detach lease")
}
err = s.le.Detach(oldLease, []lease.LeaseItem{{Key: string(key)}})
if err != nil {
panic("unexpected error from lease detach")
}
}
if leaseID != lease.NoLease {
if s.le == nil {
panic("no lessor to attach lease")
}
err = s.le.Attach(leaseID, []lease.LeaseItem{{Key: string(key)}})
if err != nil {
panic("unexpected error from lease Attach")
}
}
}
func (s *store) restore() error {
min, max := newRevBytes(), newRevBytes()
revToBytes(revision{main: 1}, min)
revToBytes(revision{main: math.MaxInt64, sub: math.MaxInt64}, max)
// restore index
tx := s.b.BatchTx()
tx.Lock()
_, finishedCompactBytes := tx.UnsafeRange(metaBucketName, finishedCompactKeyName, nil, 0)
if len(finishedCompactBytes) != 0 {
s.compactMainRev = bytesToRev(finishedCompactBytes[0]).main
plog.Printf("restore compact to %d", s.compactMainRev)
}
// TODO: limit N to reduce max memory usage
keys, vals := tx.UnsafeRange(keyBucketName, min, max, 0)
for i, key := range keys {
var kv mvccpb.KeyValue
if err := kv.Unmarshal(vals[i]); err != nil {
plog.Fatalf("cannot unmarshal event: %v", err)
}
rev := bytesToRev(key[:revBytesLen])
// restore index
switch {
case isTombstone(key):
s.kvindex.Tombstone(kv.Key, rev)
if lease.LeaseID(kv.Lease) != lease.NoLease {
err := s.le.Detach(lease.LeaseID(kv.Lease), []lease.LeaseItem{{Key: string(kv.Key)}})
if err != nil && err != lease.ErrLeaseNotFound {
plog.Fatalf("unexpected Detach error %v", err)
}
}
default:
s.kvindex.Restore(kv.Key, revision{kv.CreateRevision, 0}, rev, kv.Version)
if lease.LeaseID(kv.Lease) != lease.NoLease {
if s.le == nil {
panic("no lessor to attach lease")
}
err := s.le.Attach(lease.LeaseID(kv.Lease), []lease.LeaseItem{{Key: string(kv.Key)}})
// We are walking through the kv history here. It is possible that we attached a key to
// the lease and the lease was revoked later.
// Thus attaching an old version of key to a none existing lease is possible here, and
// we should just ignore the error.
if err != nil && err != lease.ErrLeaseNotFound {
panic("unexpected Attach error")
}
}
}
// update revision
s.currentRev = rev
}
_, scheduledCompactBytes := tx.UnsafeRange(metaBucketName, scheduledCompactKeyName, nil, 0)
scheduledCompact := int64(0)
if len(scheduledCompactBytes) != 0 {
scheduledCompact = bytesToRev(scheduledCompactBytes[0]).main
if scheduledCompact <= s.compactMainRev {
scheduledCompact = 0
}
}
tx.Unlock()
if scheduledCompact != 0 {
s.Compact(scheduledCompact)
plog.Printf("resume scheduled compaction at %d", scheduledCompact)
}
return nil
}
func (s *store) restore() error {
min, max := newRevBytes(), newRevBytes()
revToBytes(revision{main: 1}, min)
revToBytes(revision{main: math.MaxInt64, sub: math.MaxInt64}, max)
keyToLease := make(map[string]lease.LeaseID)
// use an unordered map to hold the temp index data to speed up
// the initial key index recovery.
// we will convert this unordered map into the tree index later.
unordered := make(map[string]*keyIndex, 100000)
// restore index
tx := s.b.BatchTx()
tx.Lock()
_, finishedCompactBytes := tx.UnsafeRange(metaBucketName, finishedCompactKeyName, nil, 0)
if len(finishedCompactBytes) != 0 {
s.compactMainRev = bytesToRev(finishedCompactBytes[0]).main
plog.Printf("restore compact to %d", s.compactMainRev)
}
// TODO: limit N to reduce max memory usage
keys, vals := tx.UnsafeRange(keyBucketName, min, max, 0)
for i, key := range keys {
var kv mvccpb.KeyValue
if err := kv.Unmarshal(vals[i]); err != nil {
plog.Fatalf("cannot unmarshal event: %v", err)
}
rev := bytesToRev(key[:revBytesLen])
// restore index
switch {
case isTombstone(key):
if ki, ok := unordered[string(kv.Key)]; ok {
ki.tombstone(rev.main, rev.sub)
}
delete(keyToLease, string(kv.Key))
default:
ki, ok := unordered[string(kv.Key)]
if ok {
ki.put(rev.main, rev.sub)
} else {
ki = &keyIndex{key: kv.Key}
ki.restore(revision{kv.CreateRevision, 0}, rev, kv.Version)
unordered[string(kv.Key)] = ki
}
if lid := lease.LeaseID(kv.Lease); lid != lease.NoLease {
keyToLease[string(kv.Key)] = lid
} else {
delete(keyToLease, string(kv.Key))
}
}
// update revision
s.currentRev = rev
}
// restore the tree index from the unordered index.
for _, v := range unordered {
s.kvindex.Insert(v)
}
// keys in the range [compacted revision -N, compaction] might all be deleted due to compaction.
// the correct revision should be set to compaction revision in the case, not the largest revision
// we have seen.
if s.currentRev.main < s.compactMainRev {
s.currentRev.main = s.compactMainRev
}
for key, lid := range keyToLease {
if s.le == nil {
panic("no lessor to attach lease")
}
err := s.le.Attach(lid, []lease.LeaseItem{{Key: key}})
if err != nil {
plog.Errorf("unexpected Attach error: %v", err)
}
}
_, scheduledCompactBytes := tx.UnsafeRange(metaBucketName, scheduledCompactKeyName, nil, 0)
scheduledCompact := int64(0)
if len(scheduledCompactBytes) != 0 {
scheduledCompact = bytesToRev(scheduledCompactBytes[0]).main
if scheduledCompact <= s.compactMainRev {
scheduledCompact = 0
}
}
tx.Unlock()
if scheduledCompact != 0 {
s.Compact(scheduledCompact)
plog.Printf("resume scheduled compaction at %d", scheduledCompact)
}
return nil
}
func (s *store) restore() error {
min, max := newRevBytes(), newRevBytes()
revToBytes(revision{main: 1}, min)
revToBytes(revision{main: math.MaxInt64, sub: math.MaxInt64}, max)
keyToLease := make(map[string]lease.LeaseID)
// restore index
tx := s.b.BatchTx()
tx.Lock()
_, finishedCompactBytes := tx.UnsafeRange(metaBucketName, finishedCompactKeyName, nil, 0)
if len(finishedCompactBytes) != 0 {
s.compactMainRev = bytesToRev(finishedCompactBytes[0]).main
plog.Printf("restore compact to %d", s.compactMainRev)
}
// TODO: limit N to reduce max memory usage
keys, vals := tx.UnsafeRange(keyBucketName, min, max, 0)
for i, key := range keys {
var kv mvccpb.KeyValue
if err := kv.Unmarshal(vals[i]); err != nil {
plog.Fatalf("cannot unmarshal event: %v", err)
}
rev := bytesToRev(key[:revBytesLen])
// restore index
switch {
case isTombstone(key):
s.kvindex.Tombstone(kv.Key, rev)
delete(keyToLease, string(kv.Key))
default:
s.kvindex.Restore(kv.Key, revision{kv.CreateRevision, 0}, rev, kv.Version)
if lid := lease.LeaseID(kv.Lease); lid != lease.NoLease {
keyToLease[string(kv.Key)] = lid
} else {
delete(keyToLease, string(kv.Key))
}
}
// update revision
s.currentRev = rev
}
for key, lid := range keyToLease {
if s.le == nil {
panic("no lessor to attach lease")
}
err := s.le.Attach(lid, []lease.LeaseItem{{Key: key}})
if err != nil {
plog.Errorf("unexpected Attach error: %v", err)
}
}
_, scheduledCompactBytes := tx.UnsafeRange(metaBucketName, scheduledCompactKeyName, nil, 0)
scheduledCompact := int64(0)
if len(scheduledCompactBytes) != 0 {
scheduledCompact = bytesToRev(scheduledCompactBytes[0]).main
if scheduledCompact <= s.compactMainRev {
scheduledCompact = 0
}
}
tx.Unlock()
if scheduledCompact != 0 {
s.Compact(scheduledCompact)
plog.Printf("resume scheduled compaction at %d", scheduledCompact)
}
return nil
}