mirror of
https://github.com/minio/minio.git
synced 2024-12-24 06:05:55 -05:00
Single use DRWMutex, RPCClient refactor and added missing cases to lock-rpc-server (#2560)
This PR contains various fixes for the distributed release: - Use DRWMutex in namespace-lock only for a single Lock()/RLock() call in conformance to server-side rw-locking as implemented in minio/dsync - Implement missing cases in lock-rpc-server to catch Unlock() for active read locks and RUnlock() for an active write lock - Refactor RPCClient to release local mutex while making actual RPC.Call()
This commit is contained in:
parent
780ccc26f7
commit
7f92165c79
@ -86,7 +86,7 @@ func (l *lockServer) LoginHandler(args *RPCLoginArgs, reply *RPCLoginReply) erro
|
||||
return nil
|
||||
}
|
||||
|
||||
// LockHandler - rpc handler for lock operation.
|
||||
// Lock - rpc handler for (single) write lock operation.
|
||||
func (l *lockServer) Lock(args *LockArgs, reply *bool) error {
|
||||
l.mutex.Lock()
|
||||
defer l.mutex.Unlock()
|
||||
@ -98,30 +98,36 @@ func (l *lockServer) Lock(args *LockArgs, reply *bool) error {
|
||||
if !ok {
|
||||
*reply = true
|
||||
l.lockMap[args.Name] = []bool{true}
|
||||
return nil
|
||||
} else {
|
||||
// Either a read or write lock is held on the given name.
|
||||
*reply = false
|
||||
}
|
||||
// Either a read or write lock is held on the given name.
|
||||
*reply = false
|
||||
return nil
|
||||
}
|
||||
|
||||
// UnlockHandler - rpc handler for unlock operation.
|
||||
// Unlock - rpc handler for (single) write unlock operation.
|
||||
func (l *lockServer) Unlock(args *LockArgs, reply *bool) error {
|
||||
l.mutex.Lock()
|
||||
defer l.mutex.Unlock()
|
||||
if err := l.verifyArgs(args); err != nil {
|
||||
return err
|
||||
}
|
||||
_, ok := l.lockMap[args.Name]
|
||||
locksHeld, ok := l.lockMap[args.Name]
|
||||
// No lock is held on the given name, there must be some issue at the lock client side.
|
||||
if !ok {
|
||||
*reply = false
|
||||
return fmt.Errorf("Unlock attempted on an un-locked entity: %s", args.Name)
|
||||
} else if len(locksHeld) == 1 && locksHeld[0] == true {
|
||||
*reply = true
|
||||
delete(l.lockMap, args.Name)
|
||||
return nil
|
||||
} else {
|
||||
*reply = false
|
||||
return fmt.Errorf("Unlock attempted on a read locked entity: %s (%d read locks active)", args.Name, len(locksHeld))
|
||||
}
|
||||
*reply = true
|
||||
delete(l.lockMap, args.Name)
|
||||
return nil
|
||||
}
|
||||
|
||||
// RLock - rpc handler for read lock operation.
|
||||
func (l *lockServer) RLock(args *LockArgs, reply *bool) error {
|
||||
l.mutex.Lock()
|
||||
defer l.mutex.Unlock()
|
||||
@ -142,10 +148,10 @@ func (l *lockServer) RLock(args *LockArgs, reply *bool) error {
|
||||
l.lockMap[args.Name] = append(locksHeld, false)
|
||||
*reply = true
|
||||
}
|
||||
|
||||
return nil
|
||||
}
|
||||
|
||||
// RUnlock - rpc handler for read unlock operation.
|
||||
func (l *lockServer) RUnlock(args *LockArgs, reply *bool) error {
|
||||
l.mutex.Lock()
|
||||
defer l.mutex.Unlock()
|
||||
@ -154,9 +160,13 @@ func (l *lockServer) RUnlock(args *LockArgs, reply *bool) error {
|
||||
}
|
||||
locksHeld, ok := l.lockMap[args.Name]
|
||||
if !ok {
|
||||
*reply = false
|
||||
return fmt.Errorf("RUnlock attempted on an un-locked entity: %s", args.Name)
|
||||
}
|
||||
if len(locksHeld) > 1 {
|
||||
} else if len(locksHeld) == 1 && locksHeld[0] == true {
|
||||
// A write-lock is held, cannot release a read lock
|
||||
*reply = false
|
||||
return fmt.Errorf("RUnlock attempted on a write locked entity: %s", args.Name)
|
||||
} else if len(locksHeld) > 1 {
|
||||
// Remove one of the read locks held.
|
||||
locksHeld = locksHeld[1:]
|
||||
l.lockMap[args.Name] = locksHeld
|
||||
|
@ -75,8 +75,9 @@ type nsParam struct {
|
||||
|
||||
// nsLock - provides primitives for locking critical namespace regions.
|
||||
type nsLock struct {
|
||||
RWLocker
|
||||
ref uint
|
||||
writer RWLocker
|
||||
readerArray []RWLocker
|
||||
ref uint
|
||||
}
|
||||
|
||||
// nsLockMap - namespace lock map, provides primitives to Lock,
|
||||
@ -96,7 +97,7 @@ func (n *nsLockMap) lock(volume, path string, readLock bool) {
|
||||
nsLk, found := n.lockMap[param]
|
||||
if !found {
|
||||
nsLk = &nsLock{
|
||||
RWLocker: func() RWLocker {
|
||||
writer: func() RWLocker {
|
||||
if n.isDist {
|
||||
return dsync.NewDRWMutex(pathutil.Join(volume, path))
|
||||
}
|
||||
@ -107,14 +108,29 @@ func (n *nsLockMap) lock(volume, path string, readLock bool) {
|
||||
n.lockMap[param] = nsLk
|
||||
}
|
||||
nsLk.ref++ // Update ref count here to avoid multiple races.
|
||||
rwlock := nsLk.writer
|
||||
if readLock {
|
||||
rwlock = dsync.NewDRWMutex(pathutil.Join(volume, path))
|
||||
}
|
||||
// Unlock map before Locking NS which might block.
|
||||
n.lockMapMutex.Unlock()
|
||||
|
||||
// Locking here can block.
|
||||
if readLock {
|
||||
nsLk.RLock()
|
||||
rwlock.RLock()
|
||||
|
||||
// Only add (for reader case) to array after RLock() succeeds
|
||||
// (so that we know for sure that element in [0] can be RUnlocked())
|
||||
n.lockMapMutex.Lock()
|
||||
if len(nsLk.readerArray) == 0 {
|
||||
nsLk.readerArray = []RWLocker{rwlock}
|
||||
} else {
|
||||
nsLk.readerArray = append(nsLk.readerArray, rwlock)
|
||||
}
|
||||
n.lockMapMutex.Unlock()
|
||||
|
||||
} else {
|
||||
nsLk.Lock()
|
||||
rwlock.Lock()
|
||||
}
|
||||
}
|
||||
|
||||
@ -127,9 +143,15 @@ func (n *nsLockMap) unlock(volume, path string, readLock bool) {
|
||||
param := nsParam{volume, path}
|
||||
if nsLk, found := n.lockMap[param]; found {
|
||||
if readLock {
|
||||
nsLk.RUnlock()
|
||||
if len(nsLk.readerArray) == 0 {
|
||||
errorIf(errors.New("Length of reader lock array cannot be 0."), "Invalid reader lock array length detected.")
|
||||
}
|
||||
// Release first lock first (FIFO)
|
||||
nsLk.readerArray[0].RUnlock()
|
||||
// And discard first element
|
||||
nsLk.readerArray = nsLk.readerArray[1:]
|
||||
} else {
|
||||
nsLk.Unlock()
|
||||
nsLk.writer.Unlock()
|
||||
}
|
||||
if nsLk.ref == 0 {
|
||||
errorIf(errors.New("Namespace reference count cannot be 0."), "Invalid reference count detected.")
|
||||
@ -138,6 +160,10 @@ func (n *nsLockMap) unlock(volume, path string, readLock bool) {
|
||||
nsLk.ref--
|
||||
}
|
||||
if nsLk.ref == 0 {
|
||||
if len(nsLk.readerArray) != 0 {
|
||||
errorIf(errors.New("Length of reader lock array should be 0 upon deleting map entry."), "Invalid reader lock array length detected.")
|
||||
}
|
||||
|
||||
// Remove from the map if there are no more references.
|
||||
delete(n.lockMap, param)
|
||||
}
|
||||
|
@ -17,16 +17,17 @@
|
||||
package cmd
|
||||
|
||||
import (
|
||||
"errors"
|
||||
"net/rpc"
|
||||
"sync"
|
||||
)
|
||||
|
||||
// RPCClient is a wrapper type for rpc.Client which provides reconnect on first failure.
|
||||
type RPCClient struct {
|
||||
sync.Mutex
|
||||
rpc *rpc.Client
|
||||
node string
|
||||
rpcPath string
|
||||
mu sync.Mutex
|
||||
rpcPrivate *rpc.Client
|
||||
node string
|
||||
rpcPath string
|
||||
}
|
||||
|
||||
// newClient constructs a RPCClient object with node and rpcPath initialized.
|
||||
@ -39,42 +40,80 @@ func newClient(node, rpcPath string) *RPCClient {
|
||||
}
|
||||
}
|
||||
|
||||
// Close closes the underlying socket file descriptor.
|
||||
func (rpcClient *RPCClient) Close() error {
|
||||
rpcClient.Lock()
|
||||
defer rpcClient.Unlock()
|
||||
// If rpc client has not connected yet there is nothing to close.
|
||||
if rpcClient.rpc == nil {
|
||||
return nil
|
||||
// clearRPCClient clears the pointer to the rpc.Client object in a safe manner
|
||||
func (rpcClient *RPCClient) clearRPCClient() {
|
||||
rpcClient.mu.Lock()
|
||||
rpcClient.rpcPrivate = nil
|
||||
rpcClient.mu.Unlock()
|
||||
}
|
||||
|
||||
// getRPCClient gets the pointer to the rpc.Client object in a safe manner
|
||||
func (rpcClient *RPCClient) getRPCClient() *rpc.Client {
|
||||
rpcClient.mu.Lock()
|
||||
rpcLocalStack := rpcClient.rpcPrivate
|
||||
rpcClient.mu.Unlock()
|
||||
return rpcLocalStack
|
||||
}
|
||||
|
||||
// dialRPCClient tries to establish a connection to the server in a safe manner
|
||||
func (rpcClient *RPCClient) dialRPCClient() (*rpc.Client, error) {
|
||||
rpcClient.mu.Lock()
|
||||
defer rpcClient.mu.Unlock()
|
||||
// After acquiring lock, check whether another thread may not have already dialed and established connection
|
||||
if rpcClient.rpcPrivate != nil {
|
||||
return rpcClient.rpcPrivate, nil
|
||||
}
|
||||
// Reset rpcClient.rpc to allow for subsequent calls to use a new
|
||||
// (socket) connection.
|
||||
clnt := rpcClient.rpc
|
||||
rpcClient.rpc = nil
|
||||
return clnt.Close()
|
||||
rpc, err := rpc.DialHTTPPath("tcp", rpcClient.node, rpcClient.rpcPath)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
} else if rpc == nil {
|
||||
return nil, errors.New("No valid RPC Client created after dial")
|
||||
}
|
||||
rpcClient.rpcPrivate = rpc
|
||||
return rpcClient.rpcPrivate, nil
|
||||
}
|
||||
|
||||
// Call makes a RPC call to the remote endpoint using the default codec, namely encoding/gob.
|
||||
func (rpcClient *RPCClient) Call(serviceMethod string, args interface{}, reply interface{}) error {
|
||||
rpcClient.Lock()
|
||||
defer rpcClient.Unlock()
|
||||
|
||||
// Make a copy below so that we can safely (continue to) work with the rpc.Client.
|
||||
// Even in the case the two threads would simultaneously find that the connection is not initialised,
|
||||
// they would both attempt to dial and only one of them would succeed in doing so.
|
||||
rpcLocalStack := rpcClient.getRPCClient()
|
||||
|
||||
// If the rpc.Client is nil, we attempt to (re)connect with the remote endpoint.
|
||||
if rpcClient.rpc == nil {
|
||||
clnt, err := rpc.DialHTTPPath("tcp", rpcClient.node, rpcClient.rpcPath)
|
||||
if rpcLocalStack == nil {
|
||||
var err error
|
||||
rpcLocalStack, err = rpcClient.dialRPCClient()
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
rpcClient.rpc = clnt
|
||||
}
|
||||
|
||||
// If the RPC fails due to a network-related error, then we reset
|
||||
// rpc.Client for a subsequent reconnect.
|
||||
err := rpcClient.rpc.Call(serviceMethod, args, reply)
|
||||
err := rpcLocalStack.Call(serviceMethod, args, reply)
|
||||
if IsRPCError(err) {
|
||||
rpcClient.rpc = nil
|
||||
rpcClient.clearRPCClient()
|
||||
}
|
||||
return err
|
||||
}
|
||||
|
||||
// Close closes the underlying socket file descriptor.
|
||||
func (rpcClient *RPCClient) Close() error {
|
||||
|
||||
// See comment above for making a copy on local stack
|
||||
rpcLocalStack := rpcClient.getRPCClient()
|
||||
|
||||
// If rpc client has not connected yet there is nothing to close.
|
||||
if rpcLocalStack == nil {
|
||||
return nil
|
||||
}
|
||||
|
||||
// Reset rpcClient.rpc to allow for subsequent calls to use a new
|
||||
// (socket) connection.
|
||||
rpcClient.clearRPCClient()
|
||||
return rpcLocalStack.Close()
|
||||
}
|
||||
|
||||
// IsRPCError returns true if the error value is due to a network related
|
||||
|
Loading…
Reference in New Issue
Block a user