2
2
mirror of https://github.com/octoleo/restic.git synced 2024-12-23 11:28:54 +00:00

Merge pull request #2066 from restic/fix-list-error

rebuild-index/prune: Correctly handle error listing files
This commit is contained in:
Alexander Neumann 2018-10-28 14:26:40 +01:00
commit 9f7ce7ce5a
7 changed files with 210 additions and 296 deletions

View File

@ -5,14 +5,13 @@ import (
"context"
"fmt"
"os"
"sync"
"github.com/restic/restic/internal/debug"
"github.com/restic/restic/internal/list"
"github.com/restic/restic/internal/errors"
"github.com/restic/restic/internal/pack"
"github.com/restic/restic/internal/restic"
"github.com/restic/restic/internal/worker"
"github.com/restic/restic/internal/errors"
"golang.org/x/sync/errgroup"
)
// Pack contains information about the contents of a pack.
@ -35,38 +34,118 @@ func newIndex() *Index {
}
}
const listPackWorkers = 10
// Lister lists files and their contents
type Lister interface {
// List runs fn for all files of type t in the repo.
List(ctx context.Context, t restic.FileType, fn func(restic.ID, int64) error) error
// ListPack returns the list of blobs saved in the pack id and the length
// of the file as stored in the backend.
ListPack(ctx context.Context, id restic.ID, size int64) ([]restic.Blob, int64, error)
}
// New creates a new index for repo from scratch. InvalidFiles contains all IDs
// of files that cannot be listed successfully.
func New(ctx context.Context, repo restic.Repository, ignorePacks restic.IDSet, p *restic.Progress) (idx *Index, invalidFiles restic.IDs, err error) {
func New(ctx context.Context, repo Lister, ignorePacks restic.IDSet, p *restic.Progress) (idx *Index, invalidFiles restic.IDs, err error) {
p.Start()
defer p.Done()
ch := make(chan worker.Job)
go list.AllPacks(ctx, repo, ignorePacks, ch)
type Job struct {
PackID restic.ID
Size int64
}
type Result struct {
Error error
PackID restic.ID
Size int64
Entries []restic.Blob
}
inputCh := make(chan Job)
outputCh := make(chan Result)
wg, ctx := errgroup.WithContext(ctx)
// list the files in the repo, send to inputCh
wg.Go(func() error {
defer close(inputCh)
return repo.List(ctx, restic.DataFile, func(id restic.ID, size int64) error {
if ignorePacks.Has(id) {
return nil
}
job := Job{
PackID: id,
Size: size,
}
select {
case inputCh <- job:
case <-ctx.Done():
}
return nil
})
})
// run the workers listing the files, read from inputCh, send to outputCh
var workers sync.WaitGroup
for i := 0; i < listPackWorkers; i++ {
workers.Add(1)
go func() {
defer workers.Done()
for job := range inputCh {
res := Result{PackID: job.PackID}
res.Entries, res.Size, res.Error = repo.ListPack(ctx, job.PackID, job.Size)
select {
case outputCh <- res:
case <-ctx.Done():
return
}
}
}()
}
// wait until all the workers are done, then close outputCh
wg.Go(func() error {
workers.Wait()
close(outputCh)
return nil
})
idx = newIndex()
for job := range ch {
for res := range outputCh {
p.Report(restic.Stat{Blobs: 1})
j := job.Result.(list.Result)
if job.Error != nil {
cause := errors.Cause(job.Error)
if res.Error != nil {
cause := errors.Cause(res.Error)
if _, ok := cause.(pack.InvalidFileError); ok {
invalidFiles = append(invalidFiles, j.PackID())
invalidFiles = append(invalidFiles, res.PackID)
continue
}
fmt.Fprintf(os.Stderr, "pack file cannot be listed %v: %v\n", j.PackID(), job.Error)
fmt.Fprintf(os.Stderr, "pack file cannot be listed %v: %v\n", res.PackID, res.Error)
continue
}
debug.Log("pack %v contains %d blobs", j.PackID(), len(j.Entries()))
debug.Log("pack %v contains %d blobs", res.PackID, len(res.Entries))
err := idx.AddPack(j.PackID(), j.Size(), j.Entries())
err := idx.AddPack(res.PackID, res.Size, res.Entries)
if err != nil {
return nil, nil, err
}
select {
case <-ctx.Done(): // an error occurred
default:
}
}
err = wg.Wait()
if err != nil {
return nil, nil, err
}
return idx, invalidFiles, nil
@ -89,7 +168,13 @@ type indexJSON struct {
Packs []packJSON `json:"packs"`
}
func loadIndexJSON(ctx context.Context, repo restic.Repository, id restic.ID) (*indexJSON, error) {
// ListLoader allows listing files and their content, in addition to loading and unmarshaling JSON files.
type ListLoader interface {
Lister
LoadJSONUnpacked(context.Context, restic.FileType, restic.ID, interface{}) error
}
func loadIndexJSON(ctx context.Context, repo ListLoader, id restic.ID) (*indexJSON, error) {
debug.Log("process index %v\n", id)
var idx indexJSON
@ -102,7 +187,7 @@ func loadIndexJSON(ctx context.Context, repo restic.Repository, id restic.ID) (*
}
// Load creates an index by loading all index files from the repo.
func Load(ctx context.Context, repo restic.Repository, p *restic.Progress) (*Index, error) {
func Load(ctx context.Context, repo ListLoader, p *restic.Progress) (*Index, error) {
debug.Log("loading indexes")
p.Start()
@ -259,8 +344,13 @@ func (idx *Index) FindBlob(h restic.BlobHandle) (result []Location, err error) {
const maxEntries = 3000
// Saver saves structures as JSON.
type Saver interface {
SaveJSONUnpacked(ctx context.Context, t restic.FileType, item interface{}) (restic.ID, error)
}
// Save writes the complete index to the repo.
func (idx *Index) Save(ctx context.Context, repo restic.Repository, supersedes restic.IDs) (restic.IDs, error) {
func (idx *Index) Save(ctx context.Context, repo Saver, supersedes restic.IDs) (restic.IDs, error) {
debug.Log("pack files: %d\n", len(idx.Packs))
var indexIDs []restic.ID

View File

@ -2,10 +2,12 @@ package index
import (
"context"
"sync"
"testing"
"time"
"github.com/restic/restic/internal/checker"
"github.com/restic/restic/internal/errors"
"github.com/restic/restic/internal/repository"
"github.com/restic/restic/internal/restic"
"github.com/restic/restic/internal/test"
@ -48,7 +50,7 @@ func TestIndexNew(t *testing.T) {
repo, cleanup := createFilledRepo(t, 3, 0)
defer cleanup()
idx, _, err := New(context.TODO(), repo, restic.NewIDSet(), nil)
idx, invalid, err := New(context.TODO(), repo, restic.NewIDSet(), nil)
if err != nil {
t.Fatalf("New() returned error %v", err)
}
@ -57,9 +59,102 @@ func TestIndexNew(t *testing.T) {
t.Fatalf("New() returned nil index")
}
if len(invalid) > 0 {
t.Fatalf("New() returned invalid files: %v", invalid)
}
validateIndex(t, repo, idx)
}
type ErrorRepo struct {
restic.Repository
MaxListFiles int
MaxPacks int
MaxPacksMutex sync.Mutex
}
// List returns an error after repo.MaxListFiles files.
func (repo *ErrorRepo) List(ctx context.Context, t restic.FileType, fn func(restic.ID, int64) error) error {
if repo.MaxListFiles == 0 {
return errors.New("test error, max is zero")
}
max := repo.MaxListFiles
return repo.Repository.List(ctx, t, func(id restic.ID, size int64) error {
if max == 0 {
return errors.New("test error, max reached zero")
}
max--
return fn(id, size)
})
}
// ListPack returns an error after repo.MaxPacks files.
func (repo *ErrorRepo) ListPack(ctx context.Context, id restic.ID, size int64) ([]restic.Blob, int64, error) {
repo.MaxPacksMutex.Lock()
max := repo.MaxPacks
if max > 0 {
repo.MaxPacks--
}
repo.MaxPacksMutex.Unlock()
if max == 0 {
return nil, 0, errors.New("test list pack error")
}
return repo.Repository.ListPack(ctx, id, size)
}
func TestIndexNewListErrors(t *testing.T) {
repo, cleanup := createFilledRepo(t, 3, 0)
defer cleanup()
for _, max := range []int{0, 3, 5} {
errRepo := &ErrorRepo{
Repository: repo,
MaxListFiles: max,
}
idx, invalid, err := New(context.TODO(), errRepo, restic.NewIDSet(), nil)
if err == nil {
t.Errorf("expected error not found, got nil")
}
if idx != nil {
t.Errorf("expected nil index, got %v", idx)
}
if len(invalid) != 0 {
t.Errorf("expected empty invalid list, got %v", invalid)
}
}
}
func TestIndexNewPackErrors(t *testing.T) {
repo, cleanup := createFilledRepo(t, 3, 0)
defer cleanup()
for _, max := range []int{0, 3, 5} {
errRepo := &ErrorRepo{
Repository: repo,
MaxPacks: max,
}
idx, invalid, err := New(context.TODO(), errRepo, restic.NewIDSet(), nil)
if err == nil {
t.Errorf("expected error not found, got nil")
}
if idx != nil {
t.Errorf("expected nil index, got %v", idx)
}
if len(invalid) != 0 {
t.Errorf("expected empty invalid list, got %v", invalid)
}
}
}
func TestIndexLoad(t *testing.T) {
repo, cleanup := createFilledRepo(t, 3, 0)
defer cleanup()
@ -186,7 +281,7 @@ func BenchmarkIndexSave(b *testing.B) {
}
func TestIndexDuplicateBlobs(t *testing.T) {
repo, cleanup := createFilledRepo(t, 3, 0.01)
repo, cleanup := createFilledRepo(t, 3, 0.05)
defer cleanup()
idx, _, err := New(context.TODO(), repo, restic.NewIDSet(), nil)
@ -252,6 +347,7 @@ func TestIndexSave(t *testing.T) {
}
ctx, cancel := context.WithCancel(context.TODO())
defer cancel()
errCh := make(chan error)
go checker.Structure(ctx, errCh)

View File

@ -1,79 +0,0 @@
package list
import (
"context"
"github.com/restic/restic/internal/restic"
"github.com/restic/restic/internal/worker"
)
const listPackWorkers = 10
// Lister combines lists packs in a repo and blobs in a pack.
type Lister interface {
List(context.Context, restic.FileType, func(restic.ID, int64) error) error
ListPack(context.Context, restic.ID, int64) ([]restic.Blob, int64, error)
}
// Result is returned in the channel from LoadBlobsFromAllPacks.
type Result struct {
packID restic.ID
size int64
entries []restic.Blob
}
// PackID returns the pack ID of this result.
func (l Result) PackID() restic.ID {
return l.packID
}
// Size returns the size of the pack.
func (l Result) Size() int64 {
return l.size
}
// Entries returns a list of all blobs saved in the pack.
func (l Result) Entries() []restic.Blob {
return l.entries
}
// AllPacks sends the contents of all packs to ch.
func AllPacks(ctx context.Context, repo Lister, ignorePacks restic.IDSet, ch chan<- worker.Job) {
type fileInfo struct {
id restic.ID
size int64
}
f := func(ctx context.Context, job worker.Job) (interface{}, error) {
packInfo := job.Data.(fileInfo)
entries, size, err := repo.ListPack(ctx, packInfo.id, packInfo.size)
return Result{
packID: packInfo.id,
size: size,
entries: entries,
}, err
}
jobCh := make(chan worker.Job)
wp := worker.New(ctx, listPackWorkers, f, jobCh, ch)
go func() {
defer close(jobCh)
_ = repo.List(ctx, restic.DataFile, func(id restic.ID, size int64) error {
if ignorePacks.Has(id) {
return nil
}
select {
case jobCh <- worker.Job{Data: fileInfo{id: id, size: size}, Result: Result{packID: id}}:
case <-ctx.Done():
return ctx.Err()
}
return nil
})
}()
wp.Wait()
}

View File

@ -26,6 +26,7 @@ type fakeFileSystem struct {
duplication float32
buf []byte
chunker *chunker.Chunker
rand *rand.Rand
}
// saveFile reads from rd and saves the blobs in the repository. The list of
@ -87,7 +88,7 @@ func (fs *fakeFileSystem) treeIsKnown(tree *Tree) (bool, []byte, ID) {
}
func (fs *fakeFileSystem) blobIsKnown(id ID, t BlobType) bool {
if rand.Float32() < fs.duplication {
if fs.rand.Float32() < fs.duplication {
return false
}
@ -175,6 +176,7 @@ func TestCreateSnapshot(t testing.TB, repo Repository, at time.Time, depth int,
repo: repo,
knownBlobs: NewIDSet(),
duplication: duplication,
rand: rand.New(rand.NewSource(seed)),
}
treeID := fs.saveTree(context.TODO(), seed, depth)

View File

@ -1,2 +0,0 @@
// Package worker implements a worker pool.
package worker

View File

@ -1,101 +0,0 @@
package worker
import "context"
// Job is one unit of work. It is given to a Func, and the returned result and
// error are stored in Result and Error.
type Job struct {
Data interface{}
Result interface{}
Error error
}
// Func does the actual work within a Pool.
type Func func(ctx context.Context, job Job) (result interface{}, err error)
// Pool implements a worker pool.
type Pool struct {
f Func
jobCh <-chan Job
resCh chan<- Job
numWorkers int
workersExit chan struct{}
allWorkersDone chan struct{}
}
// New returns a new worker pool with n goroutines, each running the function
// f. The workers are started immediately.
func New(ctx context.Context, n int, f Func, jobChan <-chan Job, resultChan chan<- Job) *Pool {
p := &Pool{
f: f,
workersExit: make(chan struct{}),
allWorkersDone: make(chan struct{}),
numWorkers: n,
jobCh: jobChan,
resCh: resultChan,
}
for i := 0; i < n; i++ {
go p.runWorker(ctx, i)
}
go p.waitForExit()
return p
}
// waitForExit receives from p.workersExit until all worker functions have
// exited, then closes the result channel.
func (p *Pool) waitForExit() {
n := p.numWorkers
for n > 0 {
<-p.workersExit
n--
}
close(p.allWorkersDone)
close(p.resCh)
}
// runWorker runs a worker function.
func (p *Pool) runWorker(ctx context.Context, numWorker int) {
defer func() {
p.workersExit <- struct{}{}
}()
var (
// enable the input channel when starting up a new goroutine
inCh = p.jobCh
// but do not enable the output channel until we have a result
outCh chan<- Job
job Job
ok bool
)
for {
select {
case <-ctx.Done():
return
case job, ok = <-inCh:
if !ok {
return
}
job.Result, job.Error = p.f(ctx, job)
inCh = nil
outCh = p.resCh
case outCh <- job:
outCh = nil
inCh = p.jobCh
}
}
}
// Wait waits for all worker goroutines to terminate, afterwards the output
// channel is closed.
func (p *Pool) Wait() {
<-p.allWorkersDone
}

View File

@ -1,92 +0,0 @@
package worker_test
import (
"context"
"testing"
"github.com/restic/restic/internal/errors"
"github.com/restic/restic/internal/worker"
)
const concurrency = 10
var errTooLarge = errors.New("too large")
func square(ctx context.Context, job worker.Job) (interface{}, error) {
n := job.Data.(int)
if n > 2000 {
return nil, errTooLarge
}
return n * n, nil
}
func newBufferedPool(ctx context.Context, bufsize int, n int, f worker.Func) (chan worker.Job, chan worker.Job, *worker.Pool) {
inCh := make(chan worker.Job, bufsize)
outCh := make(chan worker.Job, bufsize)
return inCh, outCh, worker.New(ctx, n, f, inCh, outCh)
}
func TestPool(t *testing.T) {
inCh, outCh, p := newBufferedPool(context.TODO(), 200, concurrency, square)
for i := 0; i < 150; i++ {
inCh <- worker.Job{Data: i}
}
close(inCh)
p.Wait()
for res := range outCh {
if res.Error != nil {
t.Errorf("unexpected error for job %v received: %v", res.Data, res.Error)
continue
}
n := res.Data.(int)
m := res.Result.(int)
if m != n*n {
t.Errorf("wrong value for job %d returned: want %d, got %d", n, n*n, m)
}
}
}
func TestPoolErrors(t *testing.T) {
inCh, outCh, p := newBufferedPool(context.TODO(), 200, concurrency, square)
for i := 0; i < 150; i++ {
inCh <- worker.Job{Data: i + 1900}
}
close(inCh)
p.Wait()
for res := range outCh {
n := res.Data.(int)
if n > 2000 {
if res.Error == nil {
t.Errorf("expected error not found, result is %v", res)
continue
}
if res.Error != errTooLarge {
t.Errorf("unexpected error found, result is %v", res)
}
continue
} else {
if res.Error != nil {
t.Errorf("unexpected error for job %v received: %v", res.Data, res.Error)
continue
}
}
m := res.Result.(int)
if m != n*n {
t.Errorf("wrong value for job %d returned: want %d, got %d", n, n*n, m)
}
}
}