mirror of
https://github.com/restic/restic.git
synced 2025-02-22 23:28:25 +00:00

* feat(backends/s3): add warmup support before repacks and restores This commit introduces basic support for transitioning pack files stored in cold storage to hot storage on S3 and S3-compatible providers. To prevent unexpected behavior for existing users, the feature is gated behind new flags: - `s3.enable-restore`: opt-in flag (defaults to false) - `s3.restore-days`: number of days for the restored objects to remain in hot storage (defaults to `7`) - `s3.restore-timeout`: maximum time to wait for a single restoration (default to `1 day`) - `s3.restore-tier`: retrieval tier at which the restore will be processed. (default to `Standard`) As restoration times can be lengthy, this implementation preemptively restores selected packs to prevent incessant restore-delays during downloads. This is slightly sub-optimal as we could process packs out-of-order (as soon as they're transitioned), but this would really add too much complexity for a marginal gain in speed. To maintain simplicity and prevent resources exhautions with lots of packs, no new concurrency mechanisms or goroutines were added. This just hooks gracefully into the existing routines. **Limitations:** - Tests against the backend were not written due to the lack of cold storage class support in MinIO. Testing was done manually on Scaleway's S3-compatible object storage. If necessary, we could explore testing with LocalStack or mocks, though this requires further discussion. - Currently, this feature only warms up before restores and repacks (prune/copy), as those are the two main use-cases I came across. Support for other commands may be added in future iterations, as long as affected packs can be calculated in advance. - The feature is gated behind a new alpha `s3-restore` feature flag to make it explicit that the feature is still wet behind the ears. - There is no explicit user notification for ongoing pack restorations. While I think it is not necessary because of the opt-in flag, showing some notice may improve usability (but would probably require major refactoring in the progress bar which I didn't want to start). Another possibility would be to add a flag to send restores requests and fail early. See https://github.com/restic/restic/issues/3202 * ui: warn user when files are warming up from cold storage * refactor: remove the PacksWarmer struct It's easier to handle multiple handles in the backend directly, and it may open the door to reducing the number of requests made to the backend in the future.
172 lines
4.1 KiB
Go
172 lines
4.1 KiB
Go
package repository
|
|
|
|
import (
|
|
"context"
|
|
"sync"
|
|
|
|
"github.com/restic/restic/internal/debug"
|
|
"github.com/restic/restic/internal/errors"
|
|
"github.com/restic/restic/internal/feature"
|
|
"github.com/restic/restic/internal/restic"
|
|
"github.com/restic/restic/internal/ui/progress"
|
|
|
|
"golang.org/x/sync/errgroup"
|
|
)
|
|
|
|
type repackBlobSet interface {
|
|
Has(bh restic.BlobHandle) bool
|
|
Delete(bh restic.BlobHandle)
|
|
Len() int
|
|
}
|
|
|
|
type LogFunc func(msg string, args ...interface{})
|
|
|
|
// Repack takes a list of packs together with a list of blobs contained in
|
|
// these packs. Each pack is loaded and the blobs listed in keepBlobs is saved
|
|
// into a new pack. Returned is the list of obsolete packs which can then
|
|
// be removed.
|
|
//
|
|
// The map keepBlobs is modified by Repack, it is used to keep track of which
|
|
// blobs have been processed.
|
|
func Repack(
|
|
ctx context.Context,
|
|
repo restic.Repository,
|
|
dstRepo restic.Repository,
|
|
packs restic.IDSet,
|
|
keepBlobs repackBlobSet,
|
|
p *progress.Counter,
|
|
logf LogFunc,
|
|
) (obsoletePacks restic.IDSet, err error) {
|
|
debug.Log("repacking %d packs while keeping %d blobs", len(packs), keepBlobs.Len())
|
|
|
|
if logf == nil {
|
|
logf = func(_ string, _ ...interface{}) {}
|
|
}
|
|
|
|
if repo == dstRepo && dstRepo.Connections() < 2 {
|
|
return nil, errors.New("repack step requires a backend connection limit of at least two")
|
|
}
|
|
|
|
wg, wgCtx := errgroup.WithContext(ctx)
|
|
|
|
dstRepo.StartPackUploader(wgCtx, wg)
|
|
wg.Go(func() error {
|
|
var err error
|
|
obsoletePacks, err = repack(wgCtx, repo, dstRepo, packs, keepBlobs, p, logf)
|
|
return err
|
|
})
|
|
|
|
if err := wg.Wait(); err != nil {
|
|
return nil, err
|
|
}
|
|
return obsoletePacks, nil
|
|
}
|
|
|
|
func repack(
|
|
ctx context.Context,
|
|
repo restic.Repository,
|
|
dstRepo restic.Repository,
|
|
packs restic.IDSet,
|
|
keepBlobs repackBlobSet,
|
|
p *progress.Counter,
|
|
logf LogFunc,
|
|
) (obsoletePacks restic.IDSet, err error) {
|
|
wg, wgCtx := errgroup.WithContext(ctx)
|
|
|
|
if feature.Flag.Enabled(feature.S3Restore) {
|
|
job, err := repo.StartWarmup(ctx, packs)
|
|
if err != nil {
|
|
return nil, err
|
|
}
|
|
if job.HandleCount() != 0 {
|
|
logf("warming up %d packs from cold storage, this may take a while...", job.HandleCount())
|
|
if err := job.Wait(ctx); err != nil {
|
|
return nil, err
|
|
}
|
|
}
|
|
}
|
|
|
|
var keepMutex sync.Mutex
|
|
downloadQueue := make(chan restic.PackBlobs)
|
|
wg.Go(func() error {
|
|
defer close(downloadQueue)
|
|
for pbs := range repo.ListPacksFromIndex(wgCtx, packs) {
|
|
var packBlobs []restic.Blob
|
|
keepMutex.Lock()
|
|
// filter out unnecessary blobs
|
|
for _, entry := range pbs.Blobs {
|
|
h := restic.BlobHandle{ID: entry.ID, Type: entry.Type}
|
|
if keepBlobs.Has(h) {
|
|
packBlobs = append(packBlobs, entry)
|
|
}
|
|
}
|
|
keepMutex.Unlock()
|
|
|
|
select {
|
|
case downloadQueue <- restic.PackBlobs{PackID: pbs.PackID, Blobs: packBlobs}:
|
|
case <-wgCtx.Done():
|
|
return wgCtx.Err()
|
|
}
|
|
}
|
|
return wgCtx.Err()
|
|
})
|
|
|
|
worker := func() error {
|
|
for t := range downloadQueue {
|
|
err := repo.LoadBlobsFromPack(wgCtx, t.PackID, t.Blobs, func(blob restic.BlobHandle, buf []byte, err error) error {
|
|
if err != nil {
|
|
// a required blob couldn't be retrieved
|
|
return err
|
|
}
|
|
|
|
keepMutex.Lock()
|
|
// recheck whether some other worker was faster
|
|
shouldKeep := keepBlobs.Has(blob)
|
|
if shouldKeep {
|
|
keepBlobs.Delete(blob)
|
|
}
|
|
keepMutex.Unlock()
|
|
|
|
if !shouldKeep {
|
|
return nil
|
|
}
|
|
|
|
// We do want to save already saved blobs!
|
|
_, _, _, err = dstRepo.SaveBlob(wgCtx, blob.Type, buf, blob.ID, true)
|
|
if err != nil {
|
|
return err
|
|
}
|
|
|
|
debug.Log(" saved blob %v", blob.ID)
|
|
return nil
|
|
})
|
|
if err != nil {
|
|
return err
|
|
}
|
|
p.Add(1)
|
|
}
|
|
return nil
|
|
}
|
|
|
|
// as packs are streamed the concurrency is limited by IO
|
|
// reduce by one to ensure that uploading is always possible
|
|
repackWorkerCount := int(repo.Connections() - 1)
|
|
if repo != dstRepo {
|
|
// no need to share the upload and download connections for different repositories
|
|
repackWorkerCount = int(repo.Connections())
|
|
}
|
|
for i := 0; i < repackWorkerCount; i++ {
|
|
wg.Go(worker)
|
|
}
|
|
|
|
if err := wg.Wait(); err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
if err := dstRepo.Flush(ctx); err != nil {
|
|
return nil, err
|
|
}
|
|
|
|
return packs, nil
|
|
}
|