Files
kubernetes/pkg/util/goroutinemap/goroutinemap.go
k8s-merge-robot 5b3b655710 Merge pull request #28153 from jsafrane/flaky-goroutinemap-wait
Automatic merge from submit-queue

Fixed goroutinemap race on Wait()

sync.WaitGroup produces data races when a GoroutineMap is empty and Wait() and Run() are called at the same time.

From sync.WaitGroup:

>  Note that calls with a positive delta that occur when the counter is zero must happen before a Wait.

Fixes #28128
Note that this issue affects only PersistentVolume unit tests.

@saad-ali, PTAL
2016-06-29 19:54:05 -07:00

254 lines
7.5 KiB
Go

/*
Copyright 2016 The Kubernetes Authors.
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
*/
/*
Package goroutinemap implements a data structure for managing go routines
by name. It prevents the creation of new go routines if an existing go routine
with the same name exists.
*/
package goroutinemap
import (
"fmt"
"runtime"
"sync"
"time"
"github.com/golang/glog"
k8sRuntime "k8s.io/kubernetes/pkg/util/runtime"
)
const (
// initialDurationBeforeRetry is the amount of time after an error occurs
// that GoRoutineMap will refuse to allow another operation to start with
// the same operationName (if exponentialBackOffOnError is enabled). Each
// successive error results in a wait 2x times the previous.
initialDurationBeforeRetry time.Duration = 500 * time.Millisecond
// maxDurationBeforeRetry is the maximum amount of time that
// durationBeforeRetry will grow to due to exponential backoff.
maxDurationBeforeRetry time.Duration = 2 * time.Minute
)
// GoRoutineMap defines the supported set of operations.
type GoRoutineMap interface {
// Run adds operationName to the list of running operations and spawns a new
// go routine to execute the operation. If an operation with the same name
// already exists, an error is returned. Once the operation is complete, the
// go routine is terminated and the operationName is removed from the list
// of executing operations allowing a new operation to be started with the
// same name without error.
Run(operationName string, operationFunc func() error) error
// Wait blocks until all operations are completed. This is typically
// necessary during tests - the test should wait until all operations finish
// and evaluate results after that.
Wait()
}
// NewGoRoutineMap returns a new instance of GoRoutineMap.
func NewGoRoutineMap(exponentialBackOffOnError bool) GoRoutineMap {
g := &goRoutineMap{
operations: make(map[string]operation),
exponentialBackOffOnError: exponentialBackOffOnError,
}
g.cond = sync.NewCond(g)
return g
}
type goRoutineMap struct {
operations map[string]operation
exponentialBackOffOnError bool
cond *sync.Cond
sync.Mutex
}
type operation struct {
operationPending bool
lastError error
lastErrorTime time.Time
durationBeforeRetry time.Duration
}
func (grm *goRoutineMap) Run(operationName string, operationFunc func() error) error {
grm.Lock()
defer grm.Unlock()
existingOp, exists := grm.operations[operationName]
if exists {
// Operation with name exists
if existingOp.operationPending {
return newAlreadyExistsError(operationName)
}
if time.Since(existingOp.lastErrorTime) <= existingOp.durationBeforeRetry {
return newExponentialBackoffError(operationName, existingOp)
}
}
grm.operations[operationName] = operation{
operationPending: true,
lastError: existingOp.lastError,
lastErrorTime: existingOp.lastErrorTime,
durationBeforeRetry: existingOp.durationBeforeRetry,
}
go func() (err error) {
// Handle unhandled panics (very unlikely)
defer k8sRuntime.HandleCrash()
// Handle completion of and error, if any, from operationFunc()
defer grm.operationComplete(operationName, &err)
// Handle panic, if any, from operationFunc()
defer recoverFromPanic(operationName, &err)
return operationFunc()
}()
return nil
}
func (grm *goRoutineMap) operationComplete(operationName string, err *error) {
defer grm.cond.Signal()
grm.Lock()
defer grm.Unlock()
if *err == nil || !grm.exponentialBackOffOnError {
// Operation completed without error, or exponentialBackOffOnError disabled
delete(grm.operations, operationName)
if *err != nil {
// Log error
glog.Errorf("operation for %q failed with: %v",
operationName,
*err)
}
} else {
// Operation completed with error and exponentialBackOffOnError Enabled
existingOp := grm.operations[operationName]
if existingOp.durationBeforeRetry == 0 {
existingOp.durationBeforeRetry = initialDurationBeforeRetry
} else {
existingOp.durationBeforeRetry = 2 * existingOp.durationBeforeRetry
if existingOp.durationBeforeRetry > maxDurationBeforeRetry {
existingOp.durationBeforeRetry = maxDurationBeforeRetry
}
}
existingOp.lastError = *err
existingOp.lastErrorTime = time.Now()
existingOp.operationPending = false
grm.operations[operationName] = existingOp
// Log error
glog.Errorf("Operation for %q failed. No retries permitted until %v (durationBeforeRetry %v). error: %v",
operationName,
existingOp.lastErrorTime.Add(existingOp.durationBeforeRetry),
existingOp.durationBeforeRetry,
*err)
}
}
func (grm *goRoutineMap) Wait() {
grm.Lock()
defer grm.Unlock()
for len(grm.operations) > 0 {
grm.cond.Wait()
}
}
func recoverFromPanic(operationName string, err *error) {
if r := recover(); r != nil {
callers := ""
for i := 0; true; i++ {
_, file, line, ok := runtime.Caller(i)
if !ok {
break
}
callers = callers + fmt.Sprintf("%v:%v\n", file, line)
}
*err = fmt.Errorf(
"operation for %q recovered from panic %q. (err=%v) Call stack:\n%v",
operationName,
r,
*err,
callers)
}
}
// alreadyExistsError is the error returned when NewGoRoutine() detects that
// an operation with the given name is already running.
type alreadyExistsError struct {
operationName string
}
var _ error = alreadyExistsError{}
func (err alreadyExistsError) Error() string {
return fmt.Sprintf("Failed to create operation with name %q. An operation with that name is already executing.", err.operationName)
}
func newAlreadyExistsError(operationName string) error {
return alreadyExistsError{operationName}
}
// IsAlreadyExists returns true if an error returned from NewGoRoutine indicates
// that operation with the same name already exists.
func IsAlreadyExists(err error) bool {
switch err.(type) {
case alreadyExistsError:
return true
default:
return false
}
}
// exponentialBackoffError is the error returned when NewGoRoutine() detects
// that the previous operation for given name failed less then
// durationBeforeRetry.
type exponentialBackoffError struct {
operationName string
failedOp operation
}
var _ error = exponentialBackoffError{}
func (err exponentialBackoffError) Error() string {
return fmt.Sprintf(
"Failed to create operation with name %q. An operation with that name failed at %v. No retries permitted until %v (%v). Last error: %q.",
err.operationName,
err.failedOp.lastErrorTime,
err.failedOp.lastErrorTime.Add(err.failedOp.durationBeforeRetry),
err.failedOp.durationBeforeRetry,
err.failedOp.lastError)
}
func newExponentialBackoffError(
operationName string, failedOp operation) error {
return exponentialBackoffError{
operationName: operationName,
failedOp: failedOp,
}
}
// IsExponentialBackoff returns true if an error returned from NewGoRoutine()
// indicates that the previous operation for given name failed less then
// durationBeforeRetry.
func IsExponentialBackoff(err error) bool {
switch err.(type) {
case exponentialBackoffError:
return true
default:
return false
}
}