2016-03-28 00:52:38 -04:00
|
|
|
/*
|
|
|
|
* Minio Cloud Storage, (C) 2016 Minio, Inc.
|
|
|
|
*
|
|
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
* you may not use this file except in compliance with the License.
|
|
|
|
* You may obtain a copy of the License at
|
|
|
|
*
|
|
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
*
|
|
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
* See the License for the specific language governing permissions and
|
|
|
|
* limitations under the License.
|
|
|
|
*/
|
|
|
|
|
2016-08-18 19:23:42 -04:00
|
|
|
package cmd
|
2016-03-28 00:52:38 -04:00
|
|
|
|
2016-05-31 23:23:31 -04:00
|
|
|
import (
|
|
|
|
"errors"
|
2016-06-19 16:35:26 -04:00
|
|
|
"io"
|
2016-06-21 17:34:11 -04:00
|
|
|
"sync"
|
2016-06-01 19:43:31 -04:00
|
|
|
|
|
|
|
"github.com/klauspost/reedsolomon"
|
2016-07-25 17:17:01 -04:00
|
|
|
"github.com/minio/minio/pkg/bpool"
|
2016-05-31 23:23:31 -04:00
|
|
|
)
|
2016-03-28 00:52:38 -04:00
|
|
|
|
2016-06-24 21:00:34 -04:00
|
|
|
// isSuccessDecodeBlocks - do we have all the blocks to be
|
|
|
|
// successfully decoded?. Input encoded blocks ordered matrix.
|
|
|
|
func isSuccessDecodeBlocks(enBlocks [][]byte, dataBlocks int) bool {
|
2016-06-24 05:06:23 -04:00
|
|
|
// Count number of data and parity blocks that were read.
|
|
|
|
var successDataBlocksCount = 0
|
|
|
|
var successParityBlocksCount = 0
|
2016-06-24 21:00:34 -04:00
|
|
|
for index := range enBlocks {
|
|
|
|
if enBlocks[index] == nil {
|
2016-06-24 05:06:23 -04:00
|
|
|
continue
|
|
|
|
}
|
2016-06-24 21:00:34 -04:00
|
|
|
// block index lesser than data blocks, update data block count.
|
2016-06-24 05:06:23 -04:00
|
|
|
if index < dataBlocks {
|
|
|
|
successDataBlocksCount++
|
|
|
|
continue
|
2016-06-24 21:00:34 -04:00
|
|
|
} // else { // update parity block count.
|
2016-06-24 05:06:23 -04:00
|
|
|
successParityBlocksCount++
|
|
|
|
}
|
2016-08-15 04:25:41 -04:00
|
|
|
// Returns true if we have atleast dataBlocks parity.
|
|
|
|
return successDataBlocksCount == dataBlocks || successDataBlocksCount+successParityBlocksCount >= dataBlocks
|
2016-06-24 05:06:23 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// isSuccessDataBlocks - do we have all the data blocks?
|
2016-06-24 21:00:34 -04:00
|
|
|
// Input encoded blocks ordered matrix.
|
|
|
|
func isSuccessDataBlocks(enBlocks [][]byte, dataBlocks int) bool {
|
2016-06-24 05:06:23 -04:00
|
|
|
// Count number of data blocks that were read.
|
|
|
|
var successDataBlocksCount = 0
|
2016-06-24 21:00:34 -04:00
|
|
|
for index := range enBlocks[:dataBlocks] {
|
|
|
|
if enBlocks[index] == nil {
|
2016-06-24 05:06:23 -04:00
|
|
|
continue
|
|
|
|
}
|
2016-06-24 21:00:34 -04:00
|
|
|
// block index lesser than data blocks, update data block count.
|
2016-06-24 05:06:23 -04:00
|
|
|
if index < dataBlocks {
|
|
|
|
successDataBlocksCount++
|
|
|
|
}
|
|
|
|
}
|
2016-06-24 21:00:34 -04:00
|
|
|
// Returns true if we have atleast the dataBlocks.
|
2016-06-24 05:06:23 -04:00
|
|
|
return successDataBlocksCount >= dataBlocks
|
|
|
|
}
|
|
|
|
|
2016-06-27 01:05:48 -04:00
|
|
|
// Return readable disks slice from which we can read parallelly.
|
2016-06-24 21:00:34 -04:00
|
|
|
func getReadDisks(orderedDisks []StorageAPI, index int, dataBlocks int) (readDisks []StorageAPI, nextIndex int, err error) {
|
|
|
|
readDisks = make([]StorageAPI, len(orderedDisks))
|
|
|
|
dataDisks := 0
|
|
|
|
parityDisks := 0
|
|
|
|
// Count already read data and parity chunks.
|
|
|
|
for i := 0; i < index; i++ {
|
|
|
|
if orderedDisks[i] == nil {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
if i < dataBlocks {
|
|
|
|
dataDisks++
|
|
|
|
} else {
|
|
|
|
parityDisks++
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
// Sanity checks - we should never have this situation.
|
|
|
|
if dataDisks == dataBlocks {
|
2016-08-25 12:39:01 -04:00
|
|
|
return nil, 0, traceError(errUnexpected)
|
2016-06-24 21:00:34 -04:00
|
|
|
}
|
2016-08-15 04:25:41 -04:00
|
|
|
if dataDisks+parityDisks >= dataBlocks {
|
2016-08-25 12:39:01 -04:00
|
|
|
return nil, 0, traceError(errUnexpected)
|
2016-06-24 21:00:34 -04:00
|
|
|
}
|
|
|
|
|
|
|
|
// Find the disks from which next set of parallel reads should happen.
|
|
|
|
for i := index; i < len(orderedDisks); i++ {
|
|
|
|
if orderedDisks[i] == nil {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
if i < dataBlocks {
|
|
|
|
dataDisks++
|
|
|
|
} else {
|
|
|
|
parityDisks++
|
|
|
|
}
|
|
|
|
readDisks[i] = orderedDisks[i]
|
|
|
|
if dataDisks == dataBlocks {
|
|
|
|
return readDisks, i + 1, nil
|
2016-08-15 04:25:41 -04:00
|
|
|
} else if dataDisks+parityDisks == dataBlocks {
|
2016-06-24 21:00:34 -04:00
|
|
|
return readDisks, i + 1, nil
|
|
|
|
}
|
|
|
|
}
|
2016-08-25 12:39:01 -04:00
|
|
|
return nil, 0, traceError(errXLReadQuorum)
|
2016-06-24 21:00:34 -04:00
|
|
|
}
|
|
|
|
|
2016-06-27 16:24:55 -04:00
|
|
|
// parallelRead - reads chunks in parallel from the disks specified in []readDisks.
|
2017-05-16 17:21:52 -04:00
|
|
|
func parallelRead(volume, path string, readDisks, orderedDisks []StorageAPI, enBlocks [][]byte,
|
|
|
|
blockOffset, curChunkSize int64, brVerifiers []bitRotVerifier, pool *bpool.BytePool) {
|
|
|
|
|
2016-06-27 16:24:55 -04:00
|
|
|
// WaitGroup to synchronise the read go-routines.
|
|
|
|
wg := &sync.WaitGroup{}
|
|
|
|
|
|
|
|
// Read disks in parallel.
|
|
|
|
for index := range readDisks {
|
|
|
|
if readDisks[index] == nil {
|
|
|
|
continue
|
|
|
|
}
|
|
|
|
wg.Add(1)
|
|
|
|
// Reads chunk from readDisk[index] in routine.
|
|
|
|
go func(index int) {
|
|
|
|
defer wg.Done()
|
|
|
|
|
2017-05-16 17:21:52 -04:00
|
|
|
// evaluate if we need to perform bit-rot checking
|
|
|
|
needBitRotVerification := true
|
|
|
|
if brVerifiers[index].isVerified {
|
|
|
|
needBitRotVerification = false
|
|
|
|
// if file has bit-rot, do not reuse disk
|
|
|
|
if brVerifiers[index].hasBitRot {
|
|
|
|
orderedDisks[index] = nil
|
|
|
|
return
|
|
|
|
}
|
2016-06-27 16:24:55 -04:00
|
|
|
}
|
|
|
|
|
2016-07-25 17:17:01 -04:00
|
|
|
buf, err := pool.Get()
|
2016-06-27 16:24:55 -04:00
|
|
|
if err != nil {
|
2016-07-25 17:17:01 -04:00
|
|
|
errorIf(err, "unable to get buffer from byte pool")
|
2016-06-27 16:24:55 -04:00
|
|
|
orderedDisks[index] = nil
|
|
|
|
return
|
|
|
|
}
|
2016-07-25 17:17:01 -04:00
|
|
|
buf = buf[:curChunkSize]
|
2016-06-27 16:24:55 -04:00
|
|
|
|
2017-05-16 17:21:52 -04:00
|
|
|
if needBitRotVerification {
|
|
|
|
_, err = readDisks[index].ReadFileWithVerify(
|
|
|
|
volume, path, blockOffset, buf,
|
|
|
|
brVerifiers[index].algo,
|
|
|
|
brVerifiers[index].checkSum)
|
|
|
|
} else {
|
|
|
|
_, err = readDisks[index].ReadFile(volume, path,
|
|
|
|
blockOffset, buf)
|
|
|
|
}
|
|
|
|
|
|
|
|
// if bit-rot verification was done, store the
|
|
|
|
// result of verification so we can skip
|
|
|
|
// re-doing it next time
|
|
|
|
if needBitRotVerification {
|
|
|
|
brVerifiers[index].isVerified = true
|
|
|
|
_, ok := err.(hashMismatchError)
|
|
|
|
brVerifiers[index].hasBitRot = ok
|
|
|
|
}
|
|
|
|
|
2016-07-25 17:17:01 -04:00
|
|
|
if err != nil {
|
|
|
|
orderedDisks[index] = nil
|
|
|
|
return
|
|
|
|
}
|
|
|
|
enBlocks[index] = buf
|
2016-06-27 16:24:55 -04:00
|
|
|
}(index)
|
|
|
|
}
|
|
|
|
|
|
|
|
// Waiting for first routines to finish.
|
|
|
|
wg.Wait()
|
|
|
|
}
|
|
|
|
|
2017-05-16 17:21:52 -04:00
|
|
|
// erasureReadFile - read bytes from erasure coded files and writes to
|
|
|
|
// given writer. Erasure coded files are read block by block as per
|
|
|
|
// given erasureInfo and data chunks are decoded into a data
|
|
|
|
// block. Data block is trimmed for given offset and length, then
|
|
|
|
// written to given writer. This function also supports bit-rot
|
|
|
|
// detection by verifying checksum of individual block's checksum.
|
|
|
|
func erasureReadFile(writer io.Writer, disks []StorageAPI, volume, path string,
|
|
|
|
offset, length, totalLength, blockSize int64, dataBlocks, parityBlocks int,
|
|
|
|
checkSums []string, algo HashAlgo, pool *bpool.BytePool) (int64, error) {
|
|
|
|
|
2016-07-07 04:30:34 -04:00
|
|
|
// Offset and length cannot be negative.
|
|
|
|
if offset < 0 || length < 0 {
|
2016-08-25 12:39:01 -04:00
|
|
|
return 0, traceError(errUnexpected)
|
2016-07-07 04:30:34 -04:00
|
|
|
}
|
|
|
|
|
2016-07-20 04:30:30 -04:00
|
|
|
// Can't request more data than what is available.
|
|
|
|
if offset+length > totalLength {
|
2016-08-25 12:39:01 -04:00
|
|
|
return 0, traceError(errUnexpected)
|
2016-07-20 04:30:30 -04:00
|
|
|
}
|
|
|
|
|
2017-05-16 17:21:52 -04:00
|
|
|
// chunkSize is the amount of data that needs to be read from
|
|
|
|
// each disk at a time.
|
2016-07-25 17:17:01 -04:00
|
|
|
chunkSize := getChunkSize(blockSize, dataBlocks)
|
|
|
|
|
2017-05-16 17:21:52 -04:00
|
|
|
brVerifiers := make([]bitRotVerifier, len(disks))
|
|
|
|
for i := range brVerifiers {
|
|
|
|
brVerifiers[i].algo = algo
|
|
|
|
brVerifiers[i].checkSum = checkSums[i]
|
|
|
|
}
|
2016-06-22 12:05:03 -04:00
|
|
|
|
|
|
|
// Total bytes written to writer
|
2017-02-24 12:20:40 -05:00
|
|
|
var bytesWritten int64
|
2016-06-22 12:05:03 -04:00
|
|
|
|
2016-07-20 04:30:30 -04:00
|
|
|
startBlock := offset / blockSize
|
|
|
|
endBlock := (offset + length) / blockSize
|
|
|
|
|
|
|
|
// curChunkSize = chunk size for the current block in the for loop below.
|
|
|
|
// curBlockSize = block size for the current block in the for loop below.
|
|
|
|
// curChunkSize and curBlockSize can change for the last block if totalLength%blockSize != 0
|
|
|
|
curChunkSize := chunkSize
|
|
|
|
curBlockSize := blockSize
|
2016-06-22 12:05:03 -04:00
|
|
|
|
|
|
|
// For each block, read chunk from each disk. If we are able to read all the data disks then we don't
|
|
|
|
// need to read parity disks. If one of the data disk is missing we need to read DataBlocks+1 number
|
|
|
|
// of disks. Once read, we Reconstruct() missing data if needed and write it to the given writer.
|
2016-07-20 04:30:30 -04:00
|
|
|
for block := startBlock; block <= endBlock; block++ {
|
2016-07-25 17:17:01 -04:00
|
|
|
// Mark all buffers as unused at the start of the loop so that the buffers
|
|
|
|
// can be reused.
|
|
|
|
pool.Reset()
|
|
|
|
|
2016-06-24 21:00:34 -04:00
|
|
|
// Each element of enBlocks holds curChunkSize'd amount of data read from its corresponding disk.
|
2016-07-16 11:35:30 -04:00
|
|
|
enBlocks := make([][]byte, len(disks))
|
2016-06-24 21:00:34 -04:00
|
|
|
|
2016-07-20 04:30:30 -04:00
|
|
|
if ((offset + bytesWritten) / blockSize) == (totalLength / blockSize) {
|
|
|
|
// This is the last block for which curBlockSize and curChunkSize can change.
|
|
|
|
// For ex. if totalLength is 15M and blockSize is 10MB, curBlockSize for
|
|
|
|
// the last block should be 5MB.
|
|
|
|
curBlockSize = totalLength % blockSize
|
|
|
|
curChunkSize = getChunkSize(curBlockSize, dataBlocks)
|
2016-06-21 17:34:11 -04:00
|
|
|
}
|
2016-06-21 00:40:10 -04:00
|
|
|
|
2016-06-24 05:06:23 -04:00
|
|
|
// NOTE: That for the offset calculation we have to use chunkSize and
|
|
|
|
// not curChunkSize. If we use curChunkSize for offset calculation
|
|
|
|
// then it can result in wrong offset for the last block.
|
|
|
|
blockOffset := block * chunkSize
|
2016-05-31 23:23:31 -04:00
|
|
|
|
2016-06-24 21:00:34 -04:00
|
|
|
// nextIndex - index from which next set of parallel reads
|
|
|
|
// should happen.
|
|
|
|
nextIndex := 0
|
|
|
|
|
2016-06-27 16:24:55 -04:00
|
|
|
for {
|
2016-06-24 21:00:34 -04:00
|
|
|
// readDisks - disks from which we need to read in parallel.
|
|
|
|
var readDisks []StorageAPI
|
|
|
|
var err error
|
2016-07-16 11:35:30 -04:00
|
|
|
// get readable disks slice from which we can read parallelly.
|
|
|
|
readDisks, nextIndex, err = getReadDisks(disks, nextIndex, dataBlocks)
|
2016-06-24 21:00:34 -04:00
|
|
|
if err != nil {
|
2016-06-27 16:24:55 -04:00
|
|
|
return bytesWritten, err
|
2016-06-19 16:35:26 -04:00
|
|
|
}
|
2016-07-16 11:35:30 -04:00
|
|
|
// Issue a parallel read across the disks specified in readDisks.
|
2017-05-16 17:21:52 -04:00
|
|
|
parallelRead(volume, path, readDisks, disks, enBlocks, blockOffset, curChunkSize, brVerifiers, pool)
|
2016-07-16 11:35:30 -04:00
|
|
|
if isSuccessDecodeBlocks(enBlocks, dataBlocks) {
|
2016-06-27 16:24:55 -04:00
|
|
|
// If enough blocks are available to do rs.Reconstruct()
|
|
|
|
break
|
|
|
|
}
|
2016-07-16 11:35:30 -04:00
|
|
|
if nextIndex == len(disks) {
|
2016-06-27 16:24:55 -04:00
|
|
|
// No more disks to read from.
|
2016-08-25 12:39:01 -04:00
|
|
|
return bytesWritten, traceError(errXLReadQuorum)
|
2016-06-24 21:00:34 -04:00
|
|
|
}
|
2016-07-16 11:35:30 -04:00
|
|
|
// We do not have enough enough data blocks to reconstruct the data
|
|
|
|
// hence continue the for-loop till we have enough data blocks.
|
2016-06-24 21:00:34 -04:00
|
|
|
}
|
2016-06-24 05:06:23 -04:00
|
|
|
|
2016-06-24 21:00:34 -04:00
|
|
|
// If we have all the data blocks no need to decode, continue to write.
|
2016-07-16 11:35:30 -04:00
|
|
|
if !isSuccessDataBlocks(enBlocks, dataBlocks) {
|
2016-06-22 12:05:03 -04:00
|
|
|
// Reconstruct the missing data blocks.
|
2016-07-16 11:35:30 -04:00
|
|
|
if err := decodeData(enBlocks, dataBlocks, parityBlocks); err != nil {
|
2016-06-19 16:35:26 -04:00
|
|
|
return bytesWritten, err
|
2016-05-31 23:23:31 -04:00
|
|
|
}
|
2016-05-28 18:13:15 -04:00
|
|
|
}
|
2016-05-31 23:23:31 -04:00
|
|
|
|
2016-07-20 04:30:30 -04:00
|
|
|
// Offset in enBlocks from where data should be read from.
|
2017-02-24 12:20:40 -05:00
|
|
|
var enBlocksOffset int64
|
2016-07-19 02:56:16 -04:00
|
|
|
|
2016-07-20 04:30:30 -04:00
|
|
|
// Total data to be read from enBlocks.
|
|
|
|
enBlocksLength := curBlockSize
|
2016-07-19 02:56:16 -04:00
|
|
|
|
2016-07-20 04:30:30 -04:00
|
|
|
// If this is the start block then enBlocksOffset might not be 0.
|
2016-06-19 16:35:26 -04:00
|
|
|
if block == startBlock {
|
2016-07-20 04:30:30 -04:00
|
|
|
enBlocksOffset = offset % blockSize
|
|
|
|
enBlocksLength -= enBlocksOffset
|
2016-06-19 16:35:26 -04:00
|
|
|
}
|
2016-05-29 18:38:14 -04:00
|
|
|
|
2016-07-20 04:30:30 -04:00
|
|
|
remaining := length - bytesWritten
|
|
|
|
if remaining < enBlocksLength {
|
2016-06-22 12:05:03 -04:00
|
|
|
// We should not send more data than what was requested.
|
2016-07-20 04:30:30 -04:00
|
|
|
enBlocksLength = remaining
|
2016-06-19 16:35:26 -04:00
|
|
|
}
|
2016-06-24 05:06:23 -04:00
|
|
|
|
2016-06-22 15:55:23 -04:00
|
|
|
// Write data blocks.
|
2016-07-20 04:30:30 -04:00
|
|
|
n, err := writeDataBlocks(writer, enBlocks, dataBlocks, enBlocksOffset, enBlocksLength)
|
2016-06-19 16:35:26 -04:00
|
|
|
if err != nil {
|
|
|
|
return bytesWritten, err
|
|
|
|
}
|
2016-06-24 05:06:23 -04:00
|
|
|
|
|
|
|
// Update total bytes written.
|
2016-06-22 15:55:23 -04:00
|
|
|
bytesWritten += n
|
2016-07-20 04:30:30 -04:00
|
|
|
|
|
|
|
if bytesWritten == length {
|
|
|
|
// Done writing all the requested data.
|
|
|
|
break
|
|
|
|
}
|
2016-05-31 23:23:31 -04:00
|
|
|
}
|
2016-06-19 16:35:26 -04:00
|
|
|
|
2016-06-24 05:06:23 -04:00
|
|
|
// Success.
|
2016-06-19 16:35:26 -04:00
|
|
|
return bytesWritten, nil
|
2016-03-28 00:52:38 -04:00
|
|
|
}
|
2016-06-02 04:49:46 -04:00
|
|
|
|
|
|
|
// decodeData - decode encoded blocks.
|
|
|
|
func decodeData(enBlocks [][]byte, dataBlocks, parityBlocks int) error {
|
2016-06-24 05:06:23 -04:00
|
|
|
// Initialized reedsolomon.
|
2016-06-02 04:49:46 -04:00
|
|
|
rs, err := reedsolomon.New(dataBlocks, parityBlocks)
|
|
|
|
if err != nil {
|
2016-08-25 12:39:01 -04:00
|
|
|
return traceError(err)
|
2016-06-02 04:49:46 -04:00
|
|
|
}
|
2016-06-24 05:06:23 -04:00
|
|
|
|
|
|
|
// Reconstruct encoded blocks.
|
2016-06-02 04:49:46 -04:00
|
|
|
err = rs.Reconstruct(enBlocks)
|
|
|
|
if err != nil {
|
2016-08-25 12:39:01 -04:00
|
|
|
return traceError(err)
|
2016-06-02 04:49:46 -04:00
|
|
|
}
|
2016-06-24 05:06:23 -04:00
|
|
|
|
2016-06-02 04:49:46 -04:00
|
|
|
// Verify reconstructed blocks (parity).
|
|
|
|
ok, err := rs.Verify(enBlocks)
|
|
|
|
if err != nil {
|
2016-08-25 12:39:01 -04:00
|
|
|
return traceError(err)
|
2016-06-02 04:49:46 -04:00
|
|
|
}
|
|
|
|
if !ok {
|
|
|
|
// Blocks cannot be reconstructed, corrupted data.
|
2016-11-15 21:14:23 -05:00
|
|
|
err = errors.New("Verification failed after reconstruction, data likely corrupted")
|
2016-08-25 12:39:01 -04:00
|
|
|
return traceError(err)
|
2016-06-02 04:49:46 -04:00
|
|
|
}
|
2016-06-24 05:06:23 -04:00
|
|
|
|
|
|
|
// Success.
|
2016-06-02 04:49:46 -04:00
|
|
|
return nil
|
|
|
|
}
|