distribution/storage/layerreader.go

175 lines
3.6 KiB
Go
Raw Normal View History

Initial implementation of registry LayerService This change contains the initial implementation of the LayerService to power layer push and pulls on the storagedriver. The interfaces presented in this package will be used by the http application to drive most features around efficient pulls and resumable pushes. The file storage/layer.go defines the interface interactions. LayerService is the root type and supports methods to access Layer and LayerUpload objects. Pull operations are supported with LayerService.Fetch and push operations are supported with LayerService.Upload and LayerService.Resume. Reads and writes of layers are split between Layer and LayerUpload, respectively. LayerService is implemented internally with the layerStore object, which takes a storagedriver.StorageDriver and a pathMapper instance. LayerUploadState is currently exported and will likely continue to be as the interaction between it and layerUploadStore are better understood. Likely, the layerUploadStore lifecycle and implementation will be deferred to the application. Image pushes pulls will be implemented in a similar manner without the discrete, persistent upload. Much of this change is in place to get something running and working. Caveats of this change include the following: 1. Layer upload state storage is implemented on the local filesystem, separate from the storage driver. This must be replaced with using the proper backend and other state storage. This can be removed when we implement resumable hashing and tarsum calculations to avoid backend roundtrips. 2. Error handling is rather bespoke at this time. The http API implementation should really dictate the error return structure for the future, so we intend to refactor this heavily to support these errors. We'd also like to collect production data to understand how failures happen in the system as a while before moving to a particular edict around error handling. 3. The layerUploadStore, which manages layer upload storage and state is not currently exported. This will likely end up being split, with the file management portion being pointed at the storagedriver and the state storage elsewhere. 4. Access Control provisions are nearly completely missing from this change. There are details around how layerindex lookup works that are related with access controls. As the auth portions of the new API take shape, these provisions will become more clear. Please see TODOs for details and individual recommendations.
2014-11-18 00:29:42 +00:00
package storage
import (
"bufio"
"fmt"
"io"
"os"
"time"
"github.com/docker/docker-registry/digest"
Initial implementation of registry LayerService This change contains the initial implementation of the LayerService to power layer push and pulls on the storagedriver. The interfaces presented in this package will be used by the http application to drive most features around efficient pulls and resumable pushes. The file storage/layer.go defines the interface interactions. LayerService is the root type and supports methods to access Layer and LayerUpload objects. Pull operations are supported with LayerService.Fetch and push operations are supported with LayerService.Upload and LayerService.Resume. Reads and writes of layers are split between Layer and LayerUpload, respectively. LayerService is implemented internally with the layerStore object, which takes a storagedriver.StorageDriver and a pathMapper instance. LayerUploadState is currently exported and will likely continue to be as the interaction between it and layerUploadStore are better understood. Likely, the layerUploadStore lifecycle and implementation will be deferred to the application. Image pushes pulls will be implemented in a similar manner without the discrete, persistent upload. Much of this change is in place to get something running and working. Caveats of this change include the following: 1. Layer upload state storage is implemented on the local filesystem, separate from the storage driver. This must be replaced with using the proper backend and other state storage. This can be removed when we implement resumable hashing and tarsum calculations to avoid backend roundtrips. 2. Error handling is rather bespoke at this time. The http API implementation should really dictate the error return structure for the future, so we intend to refactor this heavily to support these errors. We'd also like to collect production data to understand how failures happen in the system as a while before moving to a particular edict around error handling. 3. The layerUploadStore, which manages layer upload storage and state is not currently exported. This will likely end up being split, with the file management portion being pointed at the storagedriver and the state storage elsewhere. 4. Access Control provisions are nearly completely missing from this change. There are details around how layerindex lookup works that are related with access controls. As the auth portions of the new API take shape, these provisions will become more clear. Please see TODOs for details and individual recommendations.
2014-11-18 00:29:42 +00:00
)
// layerReadSeeker implements Layer and provides facilities for reading and
// seeking.
type layerReader struct {
layerStore *layerStore
rc io.ReadCloser
brd *bufio.Reader
name string // repo name of this layer
digest digest.Digest
Initial implementation of registry LayerService This change contains the initial implementation of the LayerService to power layer push and pulls on the storagedriver. The interfaces presented in this package will be used by the http application to drive most features around efficient pulls and resumable pushes. The file storage/layer.go defines the interface interactions. LayerService is the root type and supports methods to access Layer and LayerUpload objects. Pull operations are supported with LayerService.Fetch and push operations are supported with LayerService.Upload and LayerService.Resume. Reads and writes of layers are split between Layer and LayerUpload, respectively. LayerService is implemented internally with the layerStore object, which takes a storagedriver.StorageDriver and a pathMapper instance. LayerUploadState is currently exported and will likely continue to be as the interaction between it and layerUploadStore are better understood. Likely, the layerUploadStore lifecycle and implementation will be deferred to the application. Image pushes pulls will be implemented in a similar manner without the discrete, persistent upload. Much of this change is in place to get something running and working. Caveats of this change include the following: 1. Layer upload state storage is implemented on the local filesystem, separate from the storage driver. This must be replaced with using the proper backend and other state storage. This can be removed when we implement resumable hashing and tarsum calculations to avoid backend roundtrips. 2. Error handling is rather bespoke at this time. The http API implementation should really dictate the error return structure for the future, so we intend to refactor this heavily to support these errors. We'd also like to collect production data to understand how failures happen in the system as a while before moving to a particular edict around error handling. 3. The layerUploadStore, which manages layer upload storage and state is not currently exported. This will likely end up being split, with the file management portion being pointed at the storagedriver and the state storage elsewhere. 4. Access Control provisions are nearly completely missing from this change. There are details around how layerindex lookup works that are related with access controls. As the auth portions of the new API take shape, these provisions will become more clear. Please see TODOs for details and individual recommendations.
2014-11-18 00:29:42 +00:00
path string
createdAt time.Time
// offset is the current read offset
offset int64
// size is the total layer size, if available.
size int64
closedErr error // terminal error, if set, reader is closed
}
var _ Layer = &layerReader{}
func (lrs *layerReader) Name() string {
return lrs.name
}
func (lrs *layerReader) Digest() digest.Digest {
return lrs.digest
Initial implementation of registry LayerService This change contains the initial implementation of the LayerService to power layer push and pulls on the storagedriver. The interfaces presented in this package will be used by the http application to drive most features around efficient pulls and resumable pushes. The file storage/layer.go defines the interface interactions. LayerService is the root type and supports methods to access Layer and LayerUpload objects. Pull operations are supported with LayerService.Fetch and push operations are supported with LayerService.Upload and LayerService.Resume. Reads and writes of layers are split between Layer and LayerUpload, respectively. LayerService is implemented internally with the layerStore object, which takes a storagedriver.StorageDriver and a pathMapper instance. LayerUploadState is currently exported and will likely continue to be as the interaction between it and layerUploadStore are better understood. Likely, the layerUploadStore lifecycle and implementation will be deferred to the application. Image pushes pulls will be implemented in a similar manner without the discrete, persistent upload. Much of this change is in place to get something running and working. Caveats of this change include the following: 1. Layer upload state storage is implemented on the local filesystem, separate from the storage driver. This must be replaced with using the proper backend and other state storage. This can be removed when we implement resumable hashing and tarsum calculations to avoid backend roundtrips. 2. Error handling is rather bespoke at this time. The http API implementation should really dictate the error return structure for the future, so we intend to refactor this heavily to support these errors. We'd also like to collect production data to understand how failures happen in the system as a while before moving to a particular edict around error handling. 3. The layerUploadStore, which manages layer upload storage and state is not currently exported. This will likely end up being split, with the file management portion being pointed at the storagedriver and the state storage elsewhere. 4. Access Control provisions are nearly completely missing from this change. There are details around how layerindex lookup works that are related with access controls. As the auth portions of the new API take shape, these provisions will become more clear. Please see TODOs for details and individual recommendations.
2014-11-18 00:29:42 +00:00
}
func (lrs *layerReader) CreatedAt() time.Time {
return lrs.createdAt
}
func (lrs *layerReader) Read(p []byte) (n int, err error) {
if err := lrs.closed(); err != nil {
return 0, err
}
rd, err := lrs.reader()
if err != nil {
return 0, err
}
n, err = rd.Read(p)
lrs.offset += int64(n)
// Simulate io.EOR error if we reach filesize.
if err == nil && lrs.offset >= lrs.size {
err = io.EOF
}
// TODO(stevvooe): More error checking is required here. If the reader
// times out for some reason, we should reset the reader so we re-open the
// connection.
return n, err
}
func (lrs *layerReader) Seek(offset int64, whence int) (int64, error) {
if err := lrs.closed(); err != nil {
return 0, err
}
var err error
newOffset := lrs.offset
switch whence {
case os.SEEK_CUR:
newOffset += int64(whence)
case os.SEEK_END:
newOffset = lrs.size + int64(whence)
case os.SEEK_SET:
newOffset = int64(whence)
}
if newOffset < 0 {
err = fmt.Errorf("cannot seek to negative position")
} else if newOffset >= lrs.size {
err = fmt.Errorf("cannot seek passed end of layer")
} else {
if lrs.offset != newOffset {
lrs.resetReader()
}
// No problems, set the offset.
lrs.offset = newOffset
}
return lrs.offset, err
}
// Close the layer. Should be called when the resource is no longer needed.
func (lrs *layerReader) Close() error {
if lrs.closedErr != nil {
return lrs.closedErr
}
// TODO(sday): Must export this error.
lrs.closedErr = fmt.Errorf("layer closed")
// close and release reader chain
if lrs.rc != nil {
lrs.rc.Close()
lrs.rc = nil
}
lrs.brd = nil
return lrs.closedErr
}
// reader prepares the current reader at the lrs offset, ensuring its buffered
// and ready to go.
func (lrs *layerReader) reader() (io.Reader, error) {
if err := lrs.closed(); err != nil {
return nil, err
}
if lrs.rc != nil {
return lrs.brd, nil
}
// If we don't have a reader, open one up.
rc, err := lrs.layerStore.driver.ReadStream(lrs.path, uint64(lrs.offset))
if err != nil {
return nil, err
}
lrs.rc = rc
if lrs.brd == nil {
// TODO(stevvooe): Set an optimal buffer size here. We'll have to
// understand the latency characteristics of the underlying network to
// set this correctly, so we may want to leave it to the driver. For
// out of process drivers, we'll have to optimize this buffer size for
// local communication.
lrs.brd = bufio.NewReader(lrs.rc)
} else {
lrs.brd.Reset(lrs.rc)
}
return lrs.brd, nil
}
// resetReader resets the reader, forcing the read method to open up a new
// connection and rebuild the buffered reader. This should be called when the
// offset and the reader will become out of sync, such as during a seek
// operation.
func (lrs *layerReader) resetReader() {
if err := lrs.closed(); err != nil {
return
}
if lrs.rc != nil {
lrs.rc.Close()
lrs.rc = nil
}
}
func (lrs *layerReader) closed() error {
return lrs.closedErr
}