erigon-pulse/turbo/backup/backup.go
2023-09-22 14:36:54 +07:00

284 lines
6.6 KiB
Go

package backup
import (
"context"
"encoding/binary"
"encoding/hex"
"fmt"
"runtime"
"sync"
"sync/atomic"
"time"
"github.com/c2h5oh/datasize"
"github.com/erigontech/mdbx-go/mdbx"
common2 "github.com/ledgerwatch/erigon-lib/common"
"github.com/ledgerwatch/erigon-lib/common/dbg"
"github.com/ledgerwatch/erigon-lib/kv"
mdbx2 "github.com/ledgerwatch/erigon-lib/kv/mdbx"
"github.com/ledgerwatch/log/v3"
"golang.org/x/exp/maps"
"golang.org/x/sync/errgroup"
"golang.org/x/sync/semaphore"
)
func OpenPair(from, to string, label kv.Label, targetPageSize datasize.ByteSize, logger log.Logger) (kv.RoDB, kv.RwDB) {
const ThreadsHardLimit = 9_000
src := mdbx2.NewMDBX(logger).Path(from).
Label(label).
RoTxsLimiter(semaphore.NewWeighted(ThreadsHardLimit)).
WithTableCfg(func(_ kv.TableCfg) kv.TableCfg { return kv.TablesCfgByLabel(label) }).
Flags(func(flags uint) uint { return flags | mdbx.Accede }).
MustOpen()
if targetPageSize <= 0 {
targetPageSize = datasize.ByteSize(src.PageSize())
}
info, err := src.(*mdbx2.MdbxKV).Env().Info(nil)
if err != nil {
panic(err)
}
dst := mdbx2.NewMDBX(logger).Path(to).
Label(label).
PageSize(targetPageSize.Bytes()).
MapSize(datasize.ByteSize(info.Geo.Upper)).
GrowthStep(8 * datasize.GB).
Flags(func(flags uint) uint { return flags | mdbx.WriteMap }).
WithTableCfg(func(_ kv.TableCfg) kv.TableCfg { return kv.TablesCfgByLabel(label) }).
MustOpen()
return src, dst
}
func Kv2kv(ctx context.Context, src kv.RoDB, dst kv.RwDB, tables []string, readAheadThreads int, logger log.Logger) error {
srcTx, err1 := src.BeginRo(ctx)
if err1 != nil {
return err1
}
defer srcTx.Rollback()
commitEvery := time.NewTicker(5 * time.Minute)
defer commitEvery.Stop()
logEvery := time.NewTicker(20 * time.Second)
defer logEvery.Stop()
tablesMap := src.AllTables()
if len(tables) > 0 {
tablesMapCopy := maps.Clone(tablesMap)
tablesMap = kv.TableCfg{}
for _, name := range tables {
tablesMap[name] = tablesMapCopy[name]
}
}
for name, b := range tablesMap {
if b.IsDeprecated {
continue
}
if err := backupTable(ctx, src, srcTx, dst, name, readAheadThreads, logEvery, logger); err != nil {
return err
}
}
logger.Info("done")
return nil
}
func backupTable(ctx context.Context, src kv.RoDB, srcTx kv.Tx, dst kv.RwDB, table string, readAheadThreads int, logEvery *time.Ticker, logger log.Logger) error {
var total uint64
wg := sync.WaitGroup{}
defer wg.Wait()
warmupCtx, warmupCancel := context.WithCancel(ctx)
defer warmupCancel()
wg.Add(1)
go func() {
defer wg.Done()
WarmupTable(warmupCtx, src, table, log.LvlTrace, readAheadThreads)
}()
srcC, err := srcTx.Cursor(table)
if err != nil {
return err
}
total, _ = srcC.Count()
if err := dst.Update(ctx, func(tx kv.RwTx) error {
return tx.ClearBucket(table)
}); err != nil {
return err
}
dstTx, err := dst.BeginRw(ctx)
if err != nil {
return err
}
defer dstTx.Rollback()
c, err := dstTx.RwCursor(table)
if err != nil {
return err
}
casted, isDupsort := c.(kv.RwCursorDupSort)
i := uint64(0)
for k, v, err := srcC.First(); k != nil; k, v, err = srcC.Next() {
if err != nil {
return err
}
if isDupsort {
if err = casted.AppendDup(k, v); err != nil {
return err
}
} else {
if err = c.Append(k, v); err != nil {
return err
}
}
i++
if i%100_000 == 0 {
select {
case <-ctx.Done():
return ctx.Err()
case <-logEvery.C:
var m runtime.MemStats
dbg.ReadMemStats(&m)
logger.Info("Progress", "table", table, "progress", fmt.Sprintf("%.1fm/%.1fm", float64(i)/1_000_000, float64(total)/1_000_000), "key", hex.EncodeToString(k),
"alloc", common2.ByteCount(m.Alloc), "sys", common2.ByteCount(m.Sys))
default:
}
}
}
// migrate bucket sequences to native mdbx implementation
//currentID, err := srcTx.Sequence(name, 0)
//if err != nil {
// return err
//}
//_, err = dstTx.Sequence(name, currentID)
//if err != nil {
// return err
//}
if err2 := dstTx.Commit(); err2 != nil {
return err2
}
return nil
}
const ReadAheadThreads = 1024
func WarmupTable(ctx context.Context, db kv.RoDB, bucket string, lvl log.Lvl, readAheadThreads int) {
var ThreadsLimit = readAheadThreads
var total uint64
db.View(ctx, func(tx kv.Tx) error {
c, _ := tx.Cursor(bucket)
total, _ = c.Count()
return nil
})
if total < 10_000 {
return
}
progress := atomic.Int64{}
logEvery := time.NewTicker(20 * time.Second)
defer logEvery.Stop()
g, ctx := errgroup.WithContext(ctx)
g.SetLimit(ThreadsLimit)
for i := 0; i < 256; i++ {
for j := 0; j < 256; j++ {
i := i
j := j
g.Go(func() error {
return db.View(ctx, func(tx kv.Tx) error {
it, err := tx.Prefix(bucket, []byte{byte(i), byte(j)})
if err != nil {
return err
}
for it.HasNext() {
k, v, err := it.Next()
if err != nil {
return err
}
if len(k) > 0 {
_, _ = k[0], k[len(k)-1]
}
if len(v) > 0 {
_, _ = v[0], v[len(v)-1]
}
progress.Add(1)
select {
case <-ctx.Done():
return ctx.Err()
case <-logEvery.C:
log.Log(lvl, fmt.Sprintf("Progress: %s %.2f%%", bucket, 100*float64(progress.Load())/float64(total)))
default:
}
}
return nil
})
})
}
}
for i := 0; i < 1_000; i++ {
i := i
g.Go(func() error {
return db.View(ctx, func(tx kv.Tx) error {
seek := make([]byte, 8)
binary.BigEndian.PutUint64(seek, uint64(i*100_000))
it, err := tx.Prefix(bucket, seek)
if err != nil {
return err
}
for it.HasNext() {
k, v, err := it.Next()
if err != nil {
return err
}
if len(k) > 0 {
_, _ = k[0], k[len(k)-1]
}
if len(v) > 0 {
_, _ = v[0], v[len(v)-1]
}
select {
case <-ctx.Done():
return ctx.Err()
case <-logEvery.C:
log.Log(lvl, fmt.Sprintf("Progress: %s %.2f%%", bucket, 100*float64(progress.Load())/float64(total)))
default:
}
}
return nil
})
})
}
_ = g.Wait()
}
func ClearTables(ctx context.Context, db kv.RoDB, tx kv.RwTx, tables ...string) error {
for _, tbl := range tables {
if err := ClearTable(ctx, db, tx, tbl); err != nil {
return err
}
}
return nil
}
func ClearTable(ctx context.Context, db kv.RoDB, tx kv.RwTx, table string) error {
ctx, cancel := context.WithCancel(ctx)
clean := warmup(ctx, db, table)
defer func() {
cancel()
clean()
}()
log.Info("Clear", "table", table)
return tx.ClearBucket(table)
}
func warmup(ctx context.Context, db kv.RoDB, bucket string) func() {
wg := sync.WaitGroup{}
wg.Add(1)
go func() {
defer wg.Done()
WarmupTable(ctx, db, bucket, log.LvlInfo, ReadAheadThreads)
}()
return func() { wg.Wait() }
}