"io/ioutil"
"log"
"os"
+ "regexp"
"strings"
+ "sync"
"time"
"github.com/curoverse/azure-sdk-for-go/storage"
)
var (
+ azureMaxGetBytes int
azureStorageAccountName string
azureStorageAccountKeyFile string
azureStorageReplication int
- azureWriteRaceInterval time.Duration = 15 * time.Second
- azureWriteRacePollTime time.Duration = time.Second
+ azureWriteRaceInterval = 15 * time.Second
+ azureWriteRacePollTime = time.Second
)
func readKeyFromFile(file string) (string, error) {
"azure-storage-replication",
3,
"Replication level to report to clients when data is stored in an Azure container.")
+ flag.IntVar(
+ &azureMaxGetBytes,
+ "azure-max-get-bytes",
+ BlockSize,
+ fmt.Sprintf("Maximum bytes to request in a single GET request. If smaller than %d, use multiple concurrent range requests to retrieve a block.", BlockSize))
}
// An AzureBlobVolume stores and retrieves blocks in an Azure Blob
replication int
}
+// NewAzureBlobVolume returns a new AzureBlobVolume using the given
+// client and container name. The replication argument specifies the
+// replication level to report when writing data.
func NewAzureBlobVolume(client storage.Client, containerName string, readonly bool, replication int) *AzureBlobVolume {
return &AzureBlobVolume{
azClient: client,
return nil
}
+// Get reads a Keep block that has been stored as a block blob in the
+// container.
+//
+// If the block is younger than azureWriteRaceInterval and is
+// unexpectedly empty, assume a PutBlob operation is in progress, and
+// wait for it to finish writing.
func (v *AzureBlobVolume) Get(loc string) ([]byte, error) {
var deadline time.Time
haveDeadline := false
}
func (v *AzureBlobVolume) get(loc string) ([]byte, error) {
- rdr, err := v.bsClient.GetBlob(v.containerName, loc)
- if err != nil {
- if strings.Contains(err.Error(), "404 Not Found") {
- // "storage: service returned without a response body (404 Not Found)"
- return nil, os.ErrNotExist
+ expectSize := BlockSize
+ if azureMaxGetBytes < BlockSize {
+ // Unfortunately the handler doesn't tell us how long the blob
+ // is expected to be, so we have to ask Azure.
+ props, err := v.bsClient.GetBlobProperties(v.containerName, loc)
+ if err != nil {
+ return nil, v.translateError(err)
}
- return nil, err
+ if props.ContentLength > int64(BlockSize) || props.ContentLength < 0 {
+ return nil, fmt.Errorf("block %s invalid size %d (max %d)", loc, props.ContentLength, BlockSize)
+ }
+ expectSize = int(props.ContentLength)
}
- switch err := err.(type) {
- case nil:
- default:
- log.Printf("ERROR IN Get(): %T %#v", err, err)
- return nil, err
+
+ buf := bufs.Get(expectSize)
+ if expectSize == 0 {
+ return buf, nil
}
- defer rdr.Close()
- buf := bufs.Get(BlockSize)
- n, err := io.ReadFull(rdr, buf)
- switch err {
- case io.EOF, io.ErrUnexpectedEOF:
- return buf[:n], nil
- default:
- bufs.Put(buf)
- return nil, err
+
+ // We'll update this actualSize if/when we get the last piece.
+ actualSize := -1
+ pieces := (expectSize + azureMaxGetBytes - 1) / azureMaxGetBytes
+ errors := make([]error, pieces)
+ var wg sync.WaitGroup
+ wg.Add(pieces)
+ for p := 0; p < pieces; p++ {
+ go func(p int) {
+ defer wg.Done()
+ startPos := p * azureMaxGetBytes
+ endPos := startPos + azureMaxGetBytes
+ if endPos > expectSize {
+ endPos = expectSize
+ }
+ var rdr io.ReadCloser
+ var err error
+ if startPos == 0 && endPos == expectSize {
+ rdr, err = v.bsClient.GetBlob(v.containerName, loc)
+ } else {
+ rdr, err = v.bsClient.GetBlobRange(v.containerName, loc, fmt.Sprintf("%d-%d", startPos, endPos-1))
+ }
+ if err != nil {
+ errors[p] = err
+ return
+ }
+ defer rdr.Close()
+ n, err := io.ReadFull(rdr, buf[startPos:endPos])
+ if pieces == 1 && (err == io.ErrUnexpectedEOF || err == io.EOF) {
+ // If we don't know the actual size,
+ // and just tried reading 64 MiB, it's
+ // normal to encounter EOF.
+ } else if err != nil {
+ errors[p] = err
+ }
+ if p == pieces-1 {
+ actualSize = startPos + n
+ }
+ }(p)
}
+ wg.Wait()
+ for _, err := range errors {
+ if err != nil {
+ bufs.Put(buf)
+ return nil, v.translateError(err)
+ }
+ }
+ return buf[:actualSize], nil
}
+// Compare the given data with existing stored data.
func (v *AzureBlobVolume) Compare(loc string, expect []byte) error {
rdr, err := v.bsClient.GetBlob(v.containerName, loc)
if err != nil {
- return err
+ return v.translateError(err)
}
defer rdr.Close()
return compareReaderWithBuf(rdr, expect, loc[:32])
}
+// Put stores a Keep block as a block blob in the container.
func (v *AzureBlobVolume) Put(loc string, block []byte) error {
if v.readonly {
return MethodDisabledError
return v.bsClient.CreateBlockBlobFromReader(v.containerName, loc, uint64(len(block)), bytes.NewReader(block))
}
+// Touch updates the last-modified property of a block blob.
func (v *AzureBlobVolume) Touch(loc string) error {
if v.readonly {
return MethodDisabledError
})
}
+// Mtime returns the last-modified property of a block blob.
func (v *AzureBlobVolume) Mtime(loc string) (time.Time, error) {
props, err := v.bsClient.GetBlobProperties(v.containerName, loc)
if err != nil {
return time.Parse(time.RFC1123, props.LastModified)
}
+// IndexTo writes a list of Keep blocks that are stored in the
+// container.
func (v *AzureBlobVolume) IndexTo(prefix string, writer io.Writer) error {
params := storage.ListBlobsParameters{
Prefix: prefix,
if err != nil {
return err
}
+ if !v.isKeepBlock(b.Name) {
+ continue
+ }
if b.Properties.ContentLength == 0 && t.Add(azureWriteRaceInterval).After(time.Now()) {
// A new zero-length blob is probably
// just a new non-empty blob that
}
}
+// Delete a Keep block.
func (v *AzureBlobVolume) Delete(loc string) error {
if v.readonly {
return MethodDisabledError
})
}
+// Status returns a VolumeStatus struct with placeholder data.
func (v *AzureBlobVolume) Status() *VolumeStatus {
return &VolumeStatus{
DeviceNum: 1,
}
}
+// String returns a volume label, including the container name.
func (v *AzureBlobVolume) String() string {
return fmt.Sprintf("azure-storage-container:%+q", v.containerName)
}
+// Writable returns true, unless the -readonly flag was on when the
+// volume was added.
func (v *AzureBlobVolume) Writable() bool {
return !v.readonly
}
+// Replication returns the replication level of the container, as
+// specified by the -azure-storage-replication argument.
func (v *AzureBlobVolume) Replication() int {
return v.replication
}
+
+// If possible, translate an Azure SDK error to a recognizable error
+// like os.ErrNotExist.
+func (v *AzureBlobVolume) translateError(err error) error {
+ switch {
+ case err == nil:
+ return err
+ case strings.Contains(err.Error(), "404 Not Found"):
+ // "storage: service returned without a response body (404 Not Found)"
+ return os.ErrNotExist
+ default:
+ return err
+ }
+}
+
+var keepBlockRegexp = regexp.MustCompile(`^[0-9a-f]{32}$`)
+
+func (v *AzureBlobVolume) isKeepBlock(s string) bool {
+ return keepBlockRegexp.MatchString(s)
+}
projects / arvados.git / blobdiff