storj/cmd/satellite/repair_segment.go
paul cannon 1b8bd6c082 satellite/repair: unify repair logic
The repair checker and repair worker both need to determine which pieces
are healthy, which are retrievable, and which should be replaced, but
they have been doing it in different ways in different code, which has
been the cause of bugs. The same term could have very similar but subtly
different meanings between the two, causing much confusion.

With this change, the piece- and node-classification logic is
consolidated into one place within the satellite/repair package, so that
both subsystems can use it. This ought to make decision-making code more
concise and more readable.

The consolidated classification logic has been expanded to create more
sets, so that the decision-making code does not need to do as much
precalculation. It should now be clearer in comments and code that a
piece can belong to multiple sets arbitrarily (except where the
definition of the sets makes this logically impossible), and what the
precise meaning of each set is. These sets include Missing, Suspended,
Clumped, OutOfPlacement, InExcludedCountry, ForcingRepair,
UnhealthyRetrievable, Unhealthy, Retrievable, and Healthy.

Some other side effects of this change:

* CreatePutRepairOrderLimits no longer needs to special-case excluded
  countries; it can just create as many order limits as requested (by
  way of len(newNodes)).
* The repair checker will now queue a segment for repair when there are
  any pieces out of placement. The code calls this "forcing a repair".
* The checker.ReliabilityCache is now accessed by way of a GetNodes()
  function similar to the one on the overlay. The classification methods
  like MissingPieces(), OutOfPlacementPieces(), and
  PiecesNodesLastNetsInOrder() are removed in favor of the
  classification logic in satellite/repair/classification.go. This
  means the reliability cache no longer needs access to the placement
  rules or excluded countries list.

Change-Id: I105109fb94ee126952f07d747c6e11131164fadb
2023-09-25 09:42:08 -05:00

405 lines
12 KiB
Go

// Copyright (C) 2022 Storj Labs, Inc.
// See LICENSE for copying information.
package main
import (
"bytes"
"context"
"encoding/csv"
"errors"
"fmt"
"io"
"net"
"os"
"strconv"
"time"
"github.com/spf13/cobra"
"github.com/vivint/infectious"
"github.com/zeebo/errs"
"go.uber.org/zap"
"golang.org/x/sync/errgroup"
"storj.io/common/errs2"
"storj.io/common/pb"
"storj.io/common/peertls/tlsopts"
"storj.io/common/rpc"
"storj.io/common/rpc/rpcstatus"
"storj.io/common/signing"
"storj.io/common/storj"
"storj.io/common/uuid"
"storj.io/private/process"
"storj.io/storj/private/revocation"
"storj.io/storj/satellite"
"storj.io/storj/satellite/metabase"
"storj.io/storj/satellite/orders"
"storj.io/storj/satellite/overlay"
"storj.io/storj/satellite/repair/repairer"
"storj.io/storj/satellite/satellitedb"
"storj.io/uplink/private/eestream"
)
type segment struct {
StreamID uuid.UUID
Position uint64
}
func cmdRepairSegment(cmd *cobra.Command, args []string) (err error) {
ctx, _ := process.Ctx(cmd)
log := zap.L()
segments, err := collectInputSegments(args)
if err != nil {
return err
}
identity, err := runCfg.Identity.Load()
if err != nil {
log.Error("Failed to load identity.", zap.Error(err))
return errs.New("Failed to load identity: %+v", err)
}
db, err := satellitedb.Open(ctx, log.Named("db"), runCfg.Database, satellitedb.Options{ApplicationName: "satellite-repair-segment"})
if err != nil {
return errs.New("Error starting master database: %+v", err)
}
defer func() {
err = errs.Combine(err, db.Close())
}()
metabaseDB, err := metabase.Open(ctx, log.Named("metabase"), runCfg.Metainfo.DatabaseURL,
runCfg.Config.Metainfo.Metabase("satellite-repair-segment"))
if err != nil {
return errs.New("Error creating metabase connection: %+v", err)
}
defer func() {
err = errs.Combine(err, metabaseDB.Close())
}()
revocationDB, err := revocation.OpenDBFromCfg(ctx, runCfg.Server.Config)
if err != nil {
return errs.New("Error creating revocation database: %+v", err)
}
defer func() {
err = errs.Combine(err, revocationDB.Close())
}()
config := runCfg
tlsOptions, err := tlsopts.NewOptions(identity, config.Server.Config, revocationDB)
if err != nil {
return err
}
dialer := rpc.NewDefaultDialer(tlsOptions)
placement, err := config.Placement.Parse()
if err != nil {
return err
}
overlayService, err := overlay.NewService(log.Named("overlay"), db.OverlayCache(), db.NodeEvents(), placement.CreateFilters, config.Console.ExternalAddress, config.Console.SatelliteName, config.Overlay)
if err != nil {
return err
}
orders, err := orders.NewService(
log.Named("orders"),
signing.SignerFromFullIdentity(identity),
overlayService,
orders.NewNoopDB(),
placement.CreateFilters,
config.Orders,
)
if err != nil {
return err
}
ecRepairer := repairer.NewECRepairer(
log.Named("ec-repair"),
dialer,
signing.SigneeFromPeerIdentity(identity.PeerIdentity()),
config.Repairer.DialTimeout,
config.Repairer.DownloadTimeout,
true) // force inmemory download of pieces
segmentRepairer := repairer.NewSegmentRepairer(
log.Named("segment-repair"),
metabaseDB,
orders,
overlayService,
nil, // TODO add noop version
ecRepairer,
placement.CreateFilters,
config.Checker.RepairOverrides,
config.Repairer,
)
// TODO reorganize to avoid using peer.
peer := &satellite.Repairer{}
peer.Overlay = overlayService
peer.Orders.Service = orders
peer.EcRepairer = ecRepairer
peer.SegmentRepairer = segmentRepairer
cancelCtx, cancel := context.WithCancel(ctx)
group := errgroup.Group{}
group.Go(func() error {
return peer.Overlay.UploadSelectionCache.Run(cancelCtx)
})
defer func() {
cancel()
err := group.Wait()
if err != nil {
log.Error("upload cache error", zap.Error(err))
}
}()
for _, segment := range segments {
segment, err := metabaseDB.GetSegmentByPosition(ctx, metabase.GetSegmentByPosition{
StreamID: segment.StreamID,
Position: metabase.SegmentPositionFromEncoded(segment.Position),
})
if err != nil {
if metabase.ErrSegmentNotFound.Has(err) {
printOutput(segment.StreamID, segment.Position.Encode(), "segment not found in metabase db", 0, 0)
} else {
log.Error("unknown error when getting segment metadata",
zap.Stringer("stream-id", segment.StreamID),
zap.Uint64("position", segment.Position.Encode()),
zap.Error(err))
printOutput(segment.StreamID, segment.Position.Encode(), "internal", 0, 0)
}
continue
}
repairSegment(ctx, log, peer, metabaseDB, segment)
}
return nil
}
func collectInputSegments(args []string) (segments []segment, err error) {
convert := func(streamIDString, positionString string) (segment, error) {
streamID, err := uuid.FromString(streamIDString)
if err != nil {
return segment{}, errs.New("invalid stream-id (should be in UUID form): %w", err)
}
streamPosition, err := strconv.ParseUint(positionString, 10, 64)
if err != nil {
return segment{}, errs.New("stream position must be a number: %w", err)
}
return segment{
StreamID: streamID,
Position: streamPosition,
}, nil
}
if len(args) == 1 {
csvFile, err := os.Open(args[0])
if err != nil {
return nil, err
}
defer func() {
err = errs.Combine(err, csvFile.Close())
}()
csvReader := csv.NewReader(csvFile)
allEntries, err := csvReader.ReadAll()
if err != nil {
return nil, err
}
if len(allEntries) > 1 {
// ignore first line with headers
for _, entry := range allEntries[1:] {
segment, err := convert(entry[0], entry[1])
if err != nil {
return nil, err
}
segments = append(segments, segment)
}
}
} else {
segment, err := convert(args[0], args[1])
if err != nil {
return nil, err
}
segments = append(segments, segment)
}
return segments, nil
}
// repairSegment will repair selected segment no matter if it's healthy or not.
//
// Logic for this method is:
// * download whole segment into memory, use all available pieces
// * reupload segment into new nodes
// * replace segment.Pieces field with just new nodes.
func repairSegment(ctx context.Context, log *zap.Logger, peer *satellite.Repairer, metabaseDB *metabase.DB, segment metabase.Segment) {
log = log.With(zap.Stringer("stream-id", segment.StreamID), zap.Uint64("position", segment.Position.Encode()))
segmentData, failedDownloads, err := downloadSegment(ctx, log, peer, metabaseDB, segment)
if err != nil {
log.Error("download failed", zap.Error(err))
printOutput(segment.StreamID, segment.Position.Encode(), "download failed", len(segment.Pieces), failedDownloads)
return
}
if err := reuploadSegment(ctx, log, peer, metabaseDB, segment, segmentData); err != nil {
log.Error("upload failed", zap.Error(err))
printOutput(segment.StreamID, segment.Position.Encode(), "upload failed", len(segment.Pieces), failedDownloads)
return
}
printOutput(segment.StreamID, segment.Position.Encode(), "successful", len(segment.Pieces), failedDownloads)
}
func reuploadSegment(ctx context.Context, log *zap.Logger, peer *satellite.Repairer, metabaseDB *metabase.DB, segment metabase.Segment, segmentData []byte) error {
excludeNodeIDs := make([]storj.NodeID, 0, len(segment.Pieces))
for _, piece := range segment.Pieces {
excludeNodeIDs = append(excludeNodeIDs, piece.StorageNode)
}
redundancy, err := eestream.NewRedundancyStrategyFromStorj(segment.Redundancy)
if err != nil {
return err
}
request := overlay.FindStorageNodesRequest{
RequestedCount: redundancy.OptimalThreshold(),
ExcludedIDs: excludeNodeIDs,
Placement: segment.Placement,
}
newNodes, err := peer.Overlay.FindStorageNodesForUpload(ctx, request)
if err != nil {
return err
}
if len(newNodes) < redundancy.RepairThreshold() {
return errs.New("not enough new nodes were found for repair: min %v got %v", redundancy.RepairThreshold(), len(newNodes))
}
putLimits, putPrivateKey, err := peer.Orders.Service.CreatePutRepairOrderLimits(ctx, segment, make([]*pb.AddressedOrderLimit, len(newNodes)),
make(map[uint16]struct{}), newNodes)
if err != nil {
return errs.New("could not create PUT_REPAIR order limits: %w", err)
}
timeout := 5 * time.Minute
successfulNeeded := redundancy.OptimalThreshold()
successful, _, err := peer.EcRepairer.Repair(ctx, putLimits, putPrivateKey, redundancy, bytes.NewReader(segmentData),
timeout, successfulNeeded)
if err != nil {
return err
}
var repairedPieces metabase.Pieces
for i, node := range successful {
if node == nil {
continue
}
repairedPieces = append(repairedPieces, metabase.Piece{
Number: uint16(i),
StorageNode: node.Id,
})
}
if len(repairedPieces) < redundancy.RepairThreshold() {
return errs.New("not enough pieces were uploaded during repair: min %v got %v", redundancy.RepairThreshold(), len(repairedPieces))
}
// UpdateSegmentPieces is doing compare and swap
return metabaseDB.UpdateSegmentPieces(ctx, metabase.UpdateSegmentPieces{
StreamID: segment.StreamID,
Position: segment.Position,
OldPieces: segment.Pieces,
NewRedundancy: segment.Redundancy,
NewPieces: repairedPieces,
NewRepairedAt: time.Now(),
})
}
func downloadSegment(ctx context.Context, log *zap.Logger, peer *satellite.Repairer, metabaseDB *metabase.DB, segment metabase.Segment) ([]byte, int, error) {
// AdminFetchPieces downloads all pieces for specified segment and returns readers, readers data is kept on disk or inmemory
pieceInfos, err := peer.SegmentRepairer.AdminFetchPieces(ctx, &segment, "")
if err != nil {
return nil, 0, err
}
numberOfOtherFailures := 0
numberOfFileNotFound := 0
numberOfOffline := 0
pieceReaders := make(map[int]io.ReadCloser, len(pieceInfos))
for pieceNum, pieceInfo := range pieceInfos {
if pieceInfo.GetLimit == nil {
continue
}
log := log.With(zap.Int("piece num", pieceNum))
var dnsErr *net.DNSError
var opError *net.OpError
if err := pieceInfo.FetchError; err != nil {
if errs2.IsRPC(err, rpcstatus.NotFound) {
numberOfFileNotFound++
} else if errors.As(err, &dnsErr) || errors.As(err, &opError) {
numberOfOffline++
} else {
numberOfOtherFailures++
}
log.Error("unable to fetch piece", zap.Error(pieceInfo.FetchError))
continue
}
if pieceInfo.Reader == nil {
log.Error("piece reader is empty")
continue
}
pieceReaders[pieceNum] = pieceInfo.Reader
}
log.Info("download summary",
zap.Int("number of pieces", len(segment.Pieces)), zap.Int("pieces downloaded", len(pieceReaders)),
zap.Int("file not found", numberOfFileNotFound), zap.Int("offline nodes", numberOfOffline),
zap.Int("other errors", numberOfOtherFailures),
)
failedDownloads := numberOfFileNotFound + numberOfOtherFailures
redundancy, err := eestream.NewRedundancyStrategyFromStorj(segment.Redundancy)
if err != nil {
return nil, failedDownloads, errs.New("invalid redundancy strategy: %w", err)
}
if len(pieceReaders) < redundancy.RequiredCount() {
return nil, failedDownloads, errs.New("not enough pieces to reconstruct the segment, pieces: %d required: %d",
len(pieceReaders), redundancy.RequiredCount())
}
fec, err := infectious.NewFEC(redundancy.RequiredCount(), redundancy.TotalCount())
if err != nil {
return nil, failedDownloads, err
}
esScheme := eestream.NewUnsafeRSScheme(fec, redundancy.ErasureShareSize())
pieceSize := segment.PieceSize()
expectedSize := pieceSize * int64(redundancy.RequiredCount())
ctx, cancel := context.WithCancel(ctx)
defer cancel()
segmentReader := eestream.DecodeReaders2(ctx, cancel, pieceReaders, esScheme, expectedSize, 0, false)
data, err := io.ReadAll(segmentReader)
return data, failedDownloads, err
}
// printOutput prints result to standard output in a way to be able to combine
// single results into single csv file.
func printOutput(streamID uuid.UUID, position uint64, result string, numberOfPieces, failedDownloads int) {
fmt.Printf("%s,%d,%s,%d,%d\n", streamID, position, result, numberOfPieces, failedDownloads)
}