Merge pull request #9006 from snitm/thin-pool-improvements

Thin pool improvements

daemon/graphdriver/devmapper/README.md

@@ -100,6 +100,25 @@ Here is the list of supported options:
 
     ``docker -d --storage-opt dm.mountopt=nodiscard``
 
+ *  `dm.thinpooldev`
+
+    Specifies a custom blockdevice to use for the thin pool.
+
+    If using a block device for device mapper storage, ideally lvm2
+    would be used to create/manage the thin-pool volume that is then
+    handed to docker to exclusively create/manage the thin and thin
+    snapshot volumes needed for it's containers.  Managing the thin-pool
+    outside of docker makes for the most feature-rich method of having
+    docker utilize device mapper thin provisioning as the backing
+    storage for docker's containers.  lvm2-based thin-pool management
+    feature highlights include: automatic or interactive thin-pool
+    resize support, dynamically change thin-pool features, automatic
+    thinp metadata checking when lvm2 activates the thin-pool, etc.
+
+    Example use:
+
+    ``docker -d --storage-opt dm.thinpooldev=/dev/mapper/thin-pool``
+
  *  `dm.datadev`
 
     Specifies a custom blockdevice to use for data for the thin pool.

daemon/graphdriver/devmapper/deviceset.go

@@ -84,6 +84,7 @@ type DeviceSet struct {
 	metadataDevice       string
 	doBlkDiscard         bool
 	thinpBlockSize       uint32
+	thinPoolDevice       string
 }
 
 type DiskUsage struct {
@@ -150,8 +151,11 @@ func (devices *DeviceSet) oldMetadataFile() string {
 }
 
 func (devices *DeviceSet) getPoolName() string {
+	if devices.thinPoolDevice == "" {
 		return devices.devicePrefix + "-pool"
 	}
+	return devices.thinPoolDevice
+}
 
 func (devices *DeviceSet) getPoolDevName() string {
 	return getDevName(devices.getPoolName())
@@ -411,7 +415,22 @@ func (devices *DeviceSet) setupBaseImage() error {
 		}
 	}
 
-	log.Debugf("Initializing base device-manager snapshot")
+	if devices.thinPoolDevice != "" && oldInfo == nil {
+		_, transactionId, dataUsed, _, _, _, err := devices.poolStatus()
+		if err != nil {
+			return err
+		}
+		if dataUsed != 0 {
+			return fmt.Errorf("Unable to take ownership of thin-pool (%s) that already has used data blocks",
+				devices.thinPoolDevice)
+		}
+		if transactionId != 0 {
+			return fmt.Errorf("Unable to take ownership of thin-pool (%s) with non-zero transaction Id",
+				devices.thinPoolDevice)
+		}
+	}
+
+	log.Debugf("Initializing base device-mapper thin volume")
 
 	id := devices.NextDeviceId
 
@@ -430,7 +449,7 @@ func (devices *DeviceSet) setupBaseImage() error {
 		return err
 	}
 
-	log.Debugf("Creating filesystem on base device-manager snapshot")
+	log.Debugf("Creating filesystem on base device-mapper thin volume")
 
 	if err = devices.activateDeviceIfNeeded(info); err != nil {
 		return err
@@ -605,7 +624,7 @@ func (devices *DeviceSet) initDevmapper(doInit bool) error {
 	devices.devicePrefix = fmt.Sprintf("docker-%d:%d-%d", major(sysSt.Dev), minor(sysSt.Dev), sysSt.Ino)
 	log.Debugf("Generated prefix: %s", devices.devicePrefix)
 
-	// Check for the existence of the device <prefix>-pool
+	// Check for the existence of the thin-pool device
 	log.Debugf("Checking for existence of the pool '%s'", devices.getPoolName())
 	info, err := devicemapper.GetInfo(devices.getPoolName())
 	if info == nil {
@@ -624,7 +643,7 @@ func (devices *DeviceSet) initDevmapper(doInit bool) error {
 	createdLoopback := false
 
 	// If the pool doesn't exist, create it
-	if info.Exists == 0 {
+	if info.Exists == 0 && devices.thinPoolDevice == "" {
 		log.Debugf("Pool doesn't exist. Creating it.")
 
 		var (
@@ -988,9 +1007,11 @@ func (devices *DeviceSet) Shutdown() error {
 	}
 
 	devices.Lock()
+	if devices.thinPoolDevice == "" {
 		if err := devices.deactivatePool(); err != nil {
 			log.Debugf("Shutdown deactivate pool , error: %s", err)
 		}
+	}
 
 	devices.saveDeviceSetMetaData()
 	devices.Unlock()
@@ -1275,6 +1296,8 @@ func NewDeviceSet(root string, doInit bool, options []string) (*DeviceSet, error
 			devices.metadataDevice = val
 		case "dm.datadev":
 			devices.dataDevice = val
+		case "dm.thinpooldev":
+			devices.thinPoolDevice = strings.TrimPrefix(val, "/dev/mapper/")
 		case "dm.blkdiscard":
 			foundBlkDiscard = true
 			devices.doBlkDiscard, err = strconv.ParseBool(val)
@@ -1294,7 +1317,7 @@ func NewDeviceSet(root string, doInit bool, options []string) (*DeviceSet, error
 	}
 
 	// By default, don't do blk discard hack on raw devices, its rarely useful and is expensive
-	if !foundBlkDiscard && devices.dataDevice != "" {
+	if !foundBlkDiscard && (devices.dataDevice != "" || devices.thinPoolDevice != "") {
 		devices.doBlkDiscard = false
 	}
 

pkg/devicemapper/devmapper.go

@@ -384,7 +384,8 @@ func CreatePool(poolName string, dataFile, metadataFile *os.File, poolBlockSize
 	}
 
 	var cookie uint = 0
-	if err := task.SetCookie(&cookie, 0); err != nil {
+	var flags uint16 = DmUdevDisableSubsystemRulesFlag | DmUdevDisableDiskRulesFlag | DmUdevDisableOtherRulesFlag
+	if err := task.SetCookie(&cookie, flags); err != nil {
 		return fmt.Errorf("Can't set cookie %s", err)
 	}
 	defer UdevWait(cookie)

pkg/devicemapper/devmapper_wrapper.go

@@ -82,6 +82,12 @@ const (
 	LoNameSize       = C.LO_NAME_SIZE
 )
 
+const (
+	DmUdevDisableSubsystemRulesFlag = C.DM_UDEV_DISABLE_SUBSYSTEM_RULES_FLAG
+	DmUdevDisableDiskRulesFlag      = C.DM_UDEV_DISABLE_DISK_RULES_FLAG
+	DmUdevDisableOtherRulesFlag     = C.DM_UDEV_DISABLE_OTHER_RULES_FLAG
+)
+
 var (
 	DmGetLibraryVersion    = dmGetLibraryVersionFct
 	DmGetNextTarget        = dmGetNextTargetFct