refactoring

3 files changed, 112 insertions, 63 deletions

diff --git a/weed/util/chunk_cache/chunk_cache.go b/weed/util/chunk_cache/chunk_cache.go
index 7c4a77304..232e57a55 100644
--- a/weed/util/chunk_cache/chunk_cache.go
+++ b/weed/util/chunk_cache/chunk_cache.go
@@ -1,52 +1,39 @@
 package chunk_cache
 
 import (
-	"fmt"
-	"path"
-	"sort"
 	"sync"
 
 	"github.com/chrislusf/seaweedfs/weed/glog"
-	"github.com/chrislusf/seaweedfs/weed/storage"
 	"github.com/chrislusf/seaweedfs/weed/storage/needle"
 )
 
+const (
+	memCacheSizeLimit = 1024 * 1024
+)
+
 // a global cache for recently accessed file chunks
 type ChunkCache struct {
-	memCache   *ChunkCacheInMemory
-	diskCaches []*ChunkCacheVolume
+	memCache  *ChunkCacheInMemory
+	diskCache *OnDiskCacheLayer
 	sync.RWMutex
 }
 
 func NewChunkCache(maxEntries int64, dir string, diskSizeMB int64, segmentCount int) *ChunkCache {
-	c := &ChunkCache{
-		memCache: NewChunkCacheInMemory(maxEntries),
-	}
 
 	volumeCount, volumeSize := int(diskSizeMB/30000), int64(30000)
 	if volumeCount < segmentCount {
 		volumeCount, volumeSize = segmentCount, diskSizeMB/int64(segmentCount)
 	}
 
-	for i := 0; i < volumeCount; i++ {
-		fileName := path.Join(dir, fmt.Sprintf("cache_%d", i))
-		diskCache, err := LoadOrCreateChunkCacheVolume(fileName, volumeSize*1024*1024)
-		if err != nil {
-			glog.Errorf("failed to add cache %s : %v", fileName, err)
-		} else {
-			c.diskCaches = append(c.diskCaches, diskCache)
-		}
+	c := &ChunkCache{
+		memCache:  NewChunkCacheInMemory(maxEntries),
+		diskCache: NewOnDiskCacheLayer(dir, "cache", volumeCount, volumeSize),
 	}
 
-	// keep newest cache to the front
-	sort.Slice(c.diskCaches, func(i, j int) bool {
-		return c.diskCaches[i].lastModTime.After(c.diskCaches[j].lastModTime)
-	})
-
 	return c
 }
 
-func (c *ChunkCache) GetChunk(fileId string) (data []byte) {
+func (c *ChunkCache) GetChunk(fileId string, chunkSize uint64) (data []byte) {
 	if c == nil {
 		return
 	}
@@ -54,12 +41,15 @@ func (c *ChunkCache) GetChunk(fileId string) (data []byte) {
 	c.RLock()
 	defer c.RUnlock()
 
-	return c.doGetChunk(fileId)
+	return c.doGetChunk(fileId, chunkSize)
 }
 
-func (c *ChunkCache) doGetChunk(fileId string) (data []byte) {
-	if data = c.memCache.GetChunk(fileId); data != nil {
-		return data
+func (c *ChunkCache) doGetChunk(fileId string, chunkSize uint64) (data []byte) {
+
+	if chunkSize < memCacheSizeLimit {
+		if data = c.memCache.GetChunk(fileId); data != nil {
+			return data
+		}
 	}
 
 	fid, err := needle.ParseFileIdFromString(fileId)
@@ -67,20 +57,9 @@ func (c *ChunkCache) doGetChunk(fileId string) (data []byte) {
 		glog.Errorf("failed to parse file id %s", fileId)
 		return nil
 	}
-	for _, diskCache := range c.diskCaches {
-		data, err = diskCache.GetNeedle(fid.Key)
-		if err == storage.ErrorNotFound {
-			continue
-		}
-		if err != nil {
-			glog.Errorf("failed to read cache file %s id %s", diskCache.fileName, fileId)
-			continue
-		}
-		if len(data) != 0 {
-			return
-		}
-	}
-	return nil
+
+	return c.diskCache.getChunk(fid.Key)
+
 }
 
 func (c *ChunkCache) SetChunk(fileId string, data []byte) {
@@ -95,22 +74,8 @@ func (c *ChunkCache) SetChunk(fileId string, data []byte) {
 
 func (c *ChunkCache) doSetChunk(fileId string, data []byte) {
 
-	c.memCache.SetChunk(fileId, data)
-
-	if len(c.diskCaches) == 0 {
-		return
-	}
-
-	if c.diskCaches[0].fileSize+int64(len(data)) > c.diskCaches[0].sizeLimit {
-		t, resetErr := c.diskCaches[len(c.diskCaches)-1].Reset()
-		if resetErr != nil {
-			glog.Errorf("failed to reset cache file %s", c.diskCaches[len(c.diskCaches)-1].fileName)
-			return
-		}
-		for i := len(c.diskCaches) - 1; i > 0; i-- {
-			c.diskCaches[i] = c.diskCaches[i-1]
-		}
-		c.diskCaches[0] = t
+	if len(data) < memCacheSizeLimit {
+		c.memCache.SetChunk(fileId, data)
 	}
 
 	fid, err := needle.ParseFileIdFromString(fileId)
@@ -118,7 +83,8 @@ func (c *ChunkCache) doSetChunk(fileId string, data []byte) {
 		glog.Errorf("failed to parse file id %s", fileId)
 		return
 	}
-	c.diskCaches[0].WriteNeedle(fid.Key, data)
+
+	c.diskCache.setChunk(fid.Key, data)
 
 }
 
@@ -128,7 +94,5 @@ func (c *ChunkCache) Shutdown() {
 	}
 	c.Lock()
 	defer c.Unlock()
-	for _, diskCache := range c.diskCaches {
-		diskCache.Shutdown()
-	}
+	c.diskCache.shutdown()
 }
diff --git a/weed/util/chunk_cache/chunk_cache_on_disk_test.go b/weed/util/chunk_cache/chunk_cache_on_disk_test.go
index f93daf5a7..63bcba2be 100644
--- a/weed/util/chunk_cache/chunk_cache_on_disk_test.go
+++ b/weed/util/chunk_cache/chunk_cache_on_disk_test.go
@@ -23,6 +23,7 @@ func TestOnDisk(t *testing.T) {
 	type test_data struct {
 		data   []byte
 		fileId string
+		size   uint64
 	}
 	testData := make([]*test_data, writeCount)
 	for i := 0; i < writeCount; i++ {
@@ -31,12 +32,13 @@ func TestOnDisk(t *testing.T) {
 		testData[i] = &test_data{
 			data:   buff,
 			fileId: fmt.Sprintf("1,%daabbccdd", i+1),
+			size:   uint64(len(buff)),
 		}
 		cache.SetChunk(testData[i].fileId, testData[i].data)
 	}
 
 	for i := 0; i < writeCount; i++ {
-		data := cache.GetChunk(testData[i].fileId)
+		data := cache.GetChunk(testData[i].fileId, testData[i].size)
 		if bytes.Compare(data, testData[i].data) != 0 {
 			t.Errorf("failed to write to and read from cache: %d", i)
 		}
@@ -47,7 +49,7 @@ func TestOnDisk(t *testing.T) {
 	cache = NewChunkCache(0, tmpDir, totalDiskSizeMb, segmentCount)
 
 	for i := 0; i < writeCount; i++ {
-		data := cache.GetChunk(testData[i].fileId)
+		data := cache.GetChunk(testData[i].fileId, testData[i].size)
 		if bytes.Compare(data, testData[i].data) != 0 {
 			t.Errorf("failed to write to and read from cache: %d", i)
 		}
diff --git a/weed/util/chunk_cache/on_disk_cache_layer.go b/weed/util/chunk_cache/on_disk_cache_layer.go
new file mode 100644
index 000000000..065188ac3
--- /dev/null
+++ b/weed/util/chunk_cache/on_disk_cache_layer.go
@@ -0,0 +1,83 @@
+package chunk_cache
+
+import (
+	"fmt"
+	"path"
+	"sort"
+
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/storage"
+	"github.com/chrislusf/seaweedfs/weed/storage/types"
+)
+
+type OnDiskCacheLayer struct {
+	diskCaches []*ChunkCacheVolume
+}
+
+func NewOnDiskCacheLayer(dir, namePrefix string, volumeCount int, volumeSize int64) *OnDiskCacheLayer{
+	c := &OnDiskCacheLayer{}
+	for i := 0; i < volumeCount; i++ {
+		fileName := path.Join(dir, fmt.Sprintf("%s_%d", namePrefix, i))
+		diskCache, err := LoadOrCreateChunkCacheVolume(fileName, volumeSize*1024*1024)
+		if err != nil {
+			glog.Errorf("failed to add cache %s : %v", fileName, err)
+		} else {
+			c.diskCaches = append(c.diskCaches, diskCache)
+		}
+	}
+
+	// keep newest cache to the front
+	sort.Slice(c.diskCaches, func(i, j int) bool {
+		return c.diskCaches[i].lastModTime.After(c.diskCaches[j].lastModTime)
+	})
+
+	return c
+}
+
+func (c *OnDiskCacheLayer) setChunk(needleId types.NeedleId, data []byte) {
+
+	if c.diskCaches[0].fileSize+int64(len(data)) > c.diskCaches[0].sizeLimit {
+		t, resetErr := c.diskCaches[len(c.diskCaches)-1].Reset()
+		if resetErr != nil {
+			glog.Errorf("failed to reset cache file %s", c.diskCaches[len(c.diskCaches)-1].fileName)
+			return
+		}
+		for i := len(c.diskCaches) - 1; i > 0; i-- {
+			c.diskCaches[i] = c.diskCaches[i-1]
+		}
+		c.diskCaches[0] = t
+	}
+
+	c.diskCaches[0].WriteNeedle(needleId, data)
+
+}
+
+func (c *OnDiskCacheLayer) getChunk(needleId types.NeedleId) (data []byte){
+
+	var err error
+
+	for _, diskCache := range c.diskCaches {
+		data, err = diskCache.GetNeedle(needleId)
+		if err == storage.ErrorNotFound {
+			continue
+		}
+		if err != nil {
+			glog.Errorf("failed to read cache file %s id %d", diskCache.fileName, needleId)
+			continue
+		}
+		if len(data) != 0 {
+			return
+		}
+	}
+
+	return nil
+
+}
+
+func (c *OnDiskCacheLayer) shutdown(){
+
+	for _, diskCache := range c.diskCaches {
+		diskCache.Shutdown()
+	}
+
+}