Files
seaweedfs/weed/wdclient/exclusive_locks/exclusive_locker.go
T
Chris Lu 902a12fd6f wdclient: bound the wait for a master leader by the caller's context (#11002)
* wdclient: bound the wait for a master leader by the caller's context

WithClient waited on GetMaster with context.Background(), so a caller that
arrived while no master leader was known parked in a 200ms poll loop until one
appeared, whatever deadline it had already set on the RPC. Each retry above it
then left another goroutine in the same wait.

Take the context in WithClient and WithClientCustomGetMaster and hand it to
GetMaster, and stop the retry loop once it is done. The dial keeps
context.Background(): fn brings its own RPC context, so a cancellation seen
here cannot be attributed to the shared connection.

Call sites pass whatever they hold: the request context in the filer's
CollectionList, DeleteCollection and Statistics handlers and in the credential
store's propagation, the operation context in the shell's s3.bucket.delete and
the kafka gateway's broker and filer discovery, and context.Background() where
there is none - the shell commands, the admin dashboard wrapper, and the
exclusive locker's initial lease. The locker's release keeps its own
uncancelled context so a slow unlock cannot turn into a ghost lock.

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU

* wdclient: test that WithClient gives up with the caller's context

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU

* wdclient: cut the master retry backoff short when the caller gives up

util.Retry sleeps unconditionally between attempts, so a transient error
arriving just before the caller's deadline still cost it a full backoff step.
Use the context-aware util.RetryWithBackoff, the same helper the volume lookup
in this file already uses.

Two call sites went with it: the shell's lock-holder lookup builds its three
second bound before WithClient so it also covers finding the leader, as its
comment already promised, and the filer's post-delete collection cleanup goes
back to an uncancelled context - the entry is already gone, so a caller that
hung up must not leave the collection behind.

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU

* wdclient: test that a cancel during backoff ends the retry

Claude-Session: https://claude.ai/code/session_01BjDWtZsCoZY6x4pdDmGWxU
2026-08-27 22:27:45 -07:00

172 lines
4.9 KiB
Go

package exclusive_locks
import (
"context"
"sync"
"sync/atomic"
"time"
"github.com/seaweedfs/seaweedfs/weed/glog"
"github.com/seaweedfs/seaweedfs/weed/pb/master_pb"
"github.com/seaweedfs/seaweedfs/weed/wdclient"
)
const (
RenewInterval = 4 * time.Second
SafeRenewInterval = 3 * time.Second
InitLockInterval = 1 * time.Second
// bounds each lease and renew RPC attempt so an unresponsive master
// cannot hold l.mu (and block ReleaseLock) indefinitely
rpcTimeout = 3 * time.Second
)
type ExclusiveLocker struct {
token int64
lockTsNs int64
isLocked atomic.Bool
masterClient *wdclient.MasterClient
lockName string
message string
clientName string
// serializes renew and release RPCs: a renewal in flight during a release
// would re-create the lock on the master and leave it held until expiry
mu sync.Mutex
// Each lock has and only has one goroutine
renewGoroutineRunning atomic.Bool
}
func NewExclusiveLocker(masterClient *wdclient.MasterClient, lockName string) *ExclusiveLocker {
return &ExclusiveLocker{
masterClient: masterClient,
lockName: lockName,
}
}
func (l *ExclusiveLocker) IsLocked() bool {
return l.isLocked.Load()
}
func (l *ExclusiveLocker) GetToken() (token int64, lockTsNs int64) {
for time.Unix(0, atomic.LoadInt64(&l.lockTsNs)).Add(SafeRenewInterval).Before(time.Now()) {
// wait until now is within the safe lock period, no immediate renewal to change the token
time.Sleep(100 * time.Millisecond)
}
return atomic.LoadInt64(&l.token), atomic.LoadInt64(&l.lockTsNs)
}
func (l *ExclusiveLocker) RequestLock(clientName string) {
if l.isLocked.Load() {
return
}
// retry to get the lease
for {
if err := l.masterClient.WithClient(context.Background(), false, func(client master_pb.SeaweedClient) error {
attemptCtx, cancel := context.WithTimeout(context.Background(), rpcTimeout)
defer cancel()
resp, err := client.LeaseAdminToken(attemptCtx, &master_pb.LeaseAdminTokenRequest{
PreviousToken: atomic.LoadInt64(&l.token),
PreviousLockTime: atomic.LoadInt64(&l.lockTsNs),
LockName: l.lockName,
ClientName: clientName,
})
if err == nil {
atomic.StoreInt64(&l.token, resp.Token)
atomic.StoreInt64(&l.lockTsNs, resp.LockTsNs)
}
return err
}); err != nil {
glog.V(2).Infof("Failed to acquire lock %s: %v", l.lockName, err)
time.Sleep(InitLockInterval)
} else {
break
}
}
l.mu.Lock()
l.clientName = clientName
l.isLocked.Store(true)
l.mu.Unlock()
glog.V(1).Infof("Acquired lock %s", l.lockName)
// Each lock has and only has one goroutine
if l.renewGoroutineRunning.CompareAndSwap(false, true) {
// start a goroutine to renew the lease
go func() {
ctx2, cancel2 := context.WithCancel(context.Background())
defer cancel2()
for {
if err := l.renewLease(ctx2); err != nil {
glog.Warningf("Failed to renew lock %s: %v", l.lockName, err)
// clear the running flag before isLocked, so a RequestLock that
// reacquires (once isLocked is false) starts a replacement renewer
l.renewGoroutineRunning.Store(false)
l.isLocked.Store(false)
return
}
time.Sleep(RenewInterval)
}
}()
}
}
func (l *ExclusiveLocker) renewLease(ctx context.Context) error {
l.mu.Lock()
defer l.mu.Unlock()
if !l.isLocked.Load() {
return nil
}
return l.masterClient.WithClient(ctx, false, func(client master_pb.SeaweedClient) error {
attemptCtx, cancel := context.WithTimeout(ctx, rpcTimeout)
defer cancel()
resp, err := client.LeaseAdminToken(attemptCtx, &master_pb.LeaseAdminTokenRequest{
PreviousToken: atomic.LoadInt64(&l.token),
PreviousLockTime: atomic.LoadInt64(&l.lockTsNs),
LockName: l.lockName,
ClientName: l.clientName,
Message: l.message,
})
if err == nil {
atomic.StoreInt64(&l.token, resp.Token)
atomic.StoreInt64(&l.lockTsNs, resp.LockTsNs)
glog.V(2).Infof("Renewed lock %s: ts %d token %d", l.lockName, l.lockTsNs, l.token)
}
return err
})
}
func (l *ExclusiveLocker) ReleaseLock() {
l.mu.Lock()
defer l.mu.Unlock()
l.isLocked.Store(false)
l.clientName = ""
prevToken := atomic.LoadInt64(&l.token)
prevLockTsNs := atomic.LoadInt64(&l.lockTsNs)
ctx, cancel := context.WithCancel(context.Background())
defer cancel()
// single unbounded attempt: a release cut short by a deadline leaves the
// lock held until it expires, turning a slow unlock into a ghost lock
l.masterClient.WithClient(ctx, false, func(client master_pb.SeaweedClient) error {
client.ReleaseAdminToken(ctx, &master_pb.ReleaseAdminTokenRequest{
PreviousToken: prevToken,
PreviousLockTime: prevLockTsNs,
LockName: l.lockName,
})
return nil
})
// compare on clear: a RequestLock racing a slow release must not have its
// fresh token zeroed
atomic.CompareAndSwapInt64(&l.token, prevToken, 0)
atomic.CompareAndSwapInt64(&l.lockTsNs, prevLockTsNs, 0)
}
func (l *ExclusiveLocker) SetMessage(message string) {
l.message = message
}