aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
author霍晓栋 <huoxd@jiedaibao.com>2016-07-02 21:34:30 +0800
committer霍晓栋 <huoxd@jiedaibao.com>2016-07-02 21:34:30 +0800
commit2e74fb60c8d051911066d492d8f90123f98354ee (patch)
tree6966ced682450edbd565d323ceebec22f910365c
parent339856eb783d4ba6597609d1dbe50bb575fb08df (diff)
downloadseaweedfs-2e74fb60c8d051911066d492d8f90123f98354ee.tar.xz
seaweedfs-2e74fb60c8d051911066d492d8f90123f98354ee.zip
add data integrity checking
-rw-r--r--weed/storage/needle_read_write.go14
-rw-r--r--weed/storage/volume.go54
2 files changed, 66 insertions, 2 deletions
diff --git a/weed/storage/needle_read_write.go b/weed/storage/needle_read_write.go
index 2f26147d6..9499c825d 100644
--- a/weed/storage/needle_read_write.go
+++ b/weed/storage/needle_read_write.go
@@ -226,7 +226,7 @@ func ReadNeedleHeader(r *os.File, version Version, offset int64) (n *Needle, bod
//n should be a needle already read the header
//the input stream will read until next file entry
-func (n *Needle) ReadNeedleBody(r *os.File, version Version, offset int64, bodyLength uint32) (err error) {
+func (n *Needle) ReadNeedleBody(r *os.File, version Version, offset int64, bodyLength uint32, verifyCheckSum bool) (err error) {
if bodyLength <= 0 {
return nil
}
@@ -238,6 +238,12 @@ func (n *Needle) ReadNeedleBody(r *os.File, version Version, offset int64, bodyL
}
n.Data = bytes[:n.Size]
n.Checksum = NewCRC(n.Data)
+ if verifyCheckSum {
+ checksum := util.BytesToUint32(bytes[n.Size : n.Size+NeedleChecksumSize])
+ if n.Checksum.Value() != checksum {
+ err = fmt.Errorf("CRC check failed")
+ }
+ }
case Version2:
bytes := make([]byte, bodyLength)
if _, err = r.ReadAt(bytes, offset); err != nil {
@@ -245,6 +251,12 @@ func (n *Needle) ReadNeedleBody(r *os.File, version Version, offset int64, bodyL
}
n.readNeedleDataVersion2(bytes[0:n.Size])
n.Checksum = NewCRC(n.Data)
+ if verifyCheckSum {
+ checksum := util.BytesToUint32(bytes[n.Size : n.Size+NeedleChecksumSize])
+ if n.Checksum.Value() != checksum {
+ err = fmt.Errorf("CRC check failed")
+ }
+ }
default:
err = fmt.Errorf("Unsupported Version! (%d)", version)
}
diff --git a/weed/storage/volume.go b/weed/storage/volume.go
index d40bdc565..b0ab9e008 100644
--- a/weed/storage/volume.go
+++ b/weed/storage/volume.go
@@ -46,6 +46,57 @@ func loadVolumeWithoutIndex(dirname string, collection string, id VolumeId, need
e = v.load(false, false, needleMapKind)
return
}
+func verifyIndexFileIntegrity(indexFile *os.File) (indexSize int64, err error) {
+ var fi os.FileInfo
+ if fi, err = indexFile.Stat(); err != nil {
+ return
+ } else if indexSize = fi.Size(); indexSize != 0 && indexSize%16 != 0 {
+ err = fmt.Errorf("index file %s's size is %d bytes, maybe corrupted", indexFile.Name(), fi.Size())
+ return
+ }
+ return
+}
+func readIndexEntryAtOffset(indexFile *os.File, offset int64, v Version) (bytes []byte, err error) {
+ if offset < 0 {
+ err = fmt.Errorf("offset %d for index file %s is invalid", offset, indexFile.Name())
+ return
+ }
+ bytes = make([]byte, 16)
+ _, err = indexFile.ReadAt(bytes, offset)
+ return
+}
+func verifyNeedleIntegrity(datFile *os.File, v Version, offset int64, key uint64) error {
+ if n, bodyLength, err := ReadNeedleHeader(datFile, v, offset); err != nil {
+ return fmt.Errorf("can not read needle header: %s", err.Error())
+ } else {
+ if n.Id != key {
+ return fmt.Errorf("index key %#x does not match needle's Id %#x", key, n.Id)
+ } else {
+ if err := n.ReadNeedleBody(datFile, v, offset+int64(NeedleHeaderSize), bodyLength, true); err != nil {
+ return fmt.Errorf("dat file %s's body reading failed: %s", datFile.Name(), err.Error())
+ }
+ }
+ }
+ return nil
+}
+func volumeDataIntegrityChecking(v *Volume, indexFile *os.File) {
+ var indexSize int64
+ var e error
+ if indexSize, e = verifyIndexFileIntegrity(indexFile); e != nil {
+ glog.V(0).Infof("verifyIndexFileIntegrity failed %s", e.Error())
+ v.readOnly = true
+ }
+ var lastIdxEntry []byte
+ if lastIdxEntry, e = readIndexEntryAtOffset(indexFile, indexSize-16, v.Version()); e != nil {
+ glog.V(0).Infof("readLastIndexEntry failed %s", e.Error())
+ v.readOnly = true
+ }
+ key, offset, _ := idxFileEntry(lastIdxEntry)
+ if e = verifyNeedleIntegrity(v.dataFile, v.Version(), int64(offset)*NeedlePaddingSize, key); e != nil {
+ glog.V(0).Infof("verifyNeedleIntegrity failed %s", e.Error())
+ v.readOnly = true
+ }
+}
func (v *Volume) FileName() (fileName string) {
if v.Collection == "" {
fileName = path.Join(v.dir, v.Id.String())
@@ -105,6 +156,7 @@ func (v *Volume) load(alsoLoadIndex bool, createDatIfMissing bool, needleMapKind
return fmt.Errorf("cannot write Volume Index %s.idx: %v", fileName, e)
}
}
+ volumeDataIntegrityChecking(v, indexFile)
switch needleMapKind {
case NeedleMapInMemory:
glog.V(0).Infoln("loading index file", fileName+".idx", "readonly", v.readOnly)
@@ -332,7 +384,7 @@ func ScanVolumeFile(dirname string, collection string, id VolumeId,
}
for n != nil {
if readNeedleBody {
- if err = n.ReadNeedleBody(v.dataFile, version, offset+int64(NeedleHeaderSize), rest); err != nil {
+ if err = n.ReadNeedleBody(v.dataFile, version, offset+int64(NeedleHeaderSize), rest, false); err != nil {
glog.V(0).Infof("cannot read needle body: %v", err)
//err = fmt.Errorf("cannot read needle body: %v", err)
//return