forgejo/modules/queue/unique_queue_disk_channel.go
Giteabot b1e68f39e7
Properly flush unique queues on startup (#23154) (#23201)
Backport #23154

There have been a number of reports of PRs being blocked whilst being
checked which have been difficult to debug. In investigating #23050 I
have realised that whilst the Warn there is somewhat of a miscall there
was a real bug in the way that the LevelUniqueQueue was being restored
on start-up of the PersistableChannelUniqueQueue.

Next there is a conflict in the setting of the internal leveldb queue
name - This wasn't being set so it was being overridden by other unique
queues.

This PR fixes these bugs and adds a testcase.

Thanks to @brechtvl  for noticing the second issue.

Fix #23050
and others

Signed-off-by: Andrew Thornton <art27@cantab.net>
Co-authored-by: zeripath <art27@cantab.net>
Co-authored-by: techknowlogick <techknowlogick@gitea.io>
Co-authored-by: delvh <leon@kske.dev>
2023-03-06 22:35:07 +08:00

336 lines
10 KiB
Go

// Copyright 2020 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package queue
import (
"context"
"runtime/pprof"
"sync"
"time"
"code.gitea.io/gitea/modules/log"
)
// PersistableChannelUniqueQueueType is the type for persistable queue
const PersistableChannelUniqueQueueType Type = "unique-persistable-channel"
// PersistableChannelUniqueQueueConfiguration is the configuration for a PersistableChannelUniqueQueue
type PersistableChannelUniqueQueueConfiguration struct {
Name string
DataDir string
BatchLength int
QueueLength int
Timeout time.Duration
MaxAttempts int
Workers int
MaxWorkers int
BlockTimeout time.Duration
BoostTimeout time.Duration
BoostWorkers int
}
// PersistableChannelUniqueQueue wraps a channel queue and level queue together
//
// Please note that this Queue does not guarantee that a particular
// task cannot be processed twice or more at the same time. Uniqueness is
// only guaranteed whilst the task is waiting in the queue.
type PersistableChannelUniqueQueue struct {
channelQueue *ChannelUniqueQueue
delayedStarter
lock sync.Mutex
closed chan struct{}
}
// NewPersistableChannelUniqueQueue creates a wrapped batched channel queue with persistable level queue backend when shutting down
// This differs from a wrapped queue in that the persistent queue is only used to persist at shutdown/terminate
func NewPersistableChannelUniqueQueue(handle HandlerFunc, cfg, exemplar interface{}) (Queue, error) {
configInterface, err := toConfig(PersistableChannelUniqueQueueConfiguration{}, cfg)
if err != nil {
return nil, err
}
config := configInterface.(PersistableChannelUniqueQueueConfiguration)
queue := &PersistableChannelUniqueQueue{
closed: make(chan struct{}),
}
wrappedHandle := func(data ...Data) (failed []Data) {
for _, unhandled := range handle(data...) {
if fail := queue.PushBack(unhandled); fail != nil {
failed = append(failed, fail)
}
}
return failed
}
channelUniqueQueue, err := NewChannelUniqueQueue(wrappedHandle, ChannelUniqueQueueConfiguration{
WorkerPoolConfiguration: WorkerPoolConfiguration{
QueueLength: config.QueueLength,
BatchLength: config.BatchLength,
BlockTimeout: config.BlockTimeout,
BoostTimeout: config.BoostTimeout,
BoostWorkers: config.BoostWorkers,
MaxWorkers: config.MaxWorkers,
Name: config.Name + "-channel",
},
Workers: config.Workers,
}, exemplar)
if err != nil {
return nil, err
}
// the level backend only needs temporary workers to catch up with the previously dropped work
levelCfg := LevelUniqueQueueConfiguration{
ByteFIFOQueueConfiguration: ByteFIFOQueueConfiguration{
WorkerPoolConfiguration: WorkerPoolConfiguration{
QueueLength: config.QueueLength,
BatchLength: config.BatchLength,
BlockTimeout: 1 * time.Second,
BoostTimeout: 5 * time.Minute,
BoostWorkers: 1,
MaxWorkers: 5,
Name: config.Name + "-level",
},
Workers: 0,
},
DataDir: config.DataDir,
QueueName: config.Name + "-level",
}
queue.channelQueue = channelUniqueQueue.(*ChannelUniqueQueue)
levelQueue, err := NewLevelUniqueQueue(func(data ...Data) []Data {
for _, datum := range data {
err := queue.Push(datum)
if err != nil && err != ErrAlreadyInQueue {
log.Error("Unable push to channelled queue: %v", err)
}
}
return nil
}, levelCfg, exemplar)
if err == nil {
queue.delayedStarter = delayedStarter{
internal: levelQueue.(*LevelUniqueQueue),
name: config.Name,
}
_ = GetManager().Add(queue, PersistableChannelUniqueQueueType, config, exemplar)
return queue, nil
}
if IsErrInvalidConfiguration(err) {
// Retrying ain't gonna make this any better...
return nil, ErrInvalidConfiguration{cfg: cfg}
}
queue.delayedStarter = delayedStarter{
cfg: levelCfg,
underlying: LevelUniqueQueueType,
timeout: config.Timeout,
maxAttempts: config.MaxAttempts,
name: config.Name,
}
_ = GetManager().Add(queue, PersistableChannelUniqueQueueType, config, exemplar)
return queue, nil
}
// Name returns the name of this queue
func (q *PersistableChannelUniqueQueue) Name() string {
return q.delayedStarter.name
}
// Push will push the indexer data to queue
func (q *PersistableChannelUniqueQueue) Push(data Data) error {
return q.PushFunc(data, nil)
}
// PushFunc will push the indexer data to queue
func (q *PersistableChannelUniqueQueue) PushFunc(data Data, fn func() error) error {
select {
case <-q.closed:
return q.internal.(UniqueQueue).PushFunc(data, fn)
default:
return q.channelQueue.PushFunc(data, fn)
}
}
// PushBack will push the indexer data to queue
func (q *PersistableChannelUniqueQueue) PushBack(data Data) error {
select {
case <-q.closed:
if pbr, ok := q.internal.(PushBackable); ok {
return pbr.PushBack(data)
}
return q.internal.Push(data)
default:
return q.channelQueue.Push(data)
}
}
// Has will test if the queue has the data
func (q *PersistableChannelUniqueQueue) Has(data Data) (bool, error) {
// This is more difficult...
has, err := q.channelQueue.Has(data)
if err != nil || has {
return has, err
}
q.lock.Lock()
defer q.lock.Unlock()
if q.internal == nil {
return false, nil
}
return q.internal.(UniqueQueue).Has(data)
}
// Run starts to run the queue
func (q *PersistableChannelUniqueQueue) Run(atShutdown, atTerminate func(func())) {
pprof.SetGoroutineLabels(q.channelQueue.baseCtx)
log.Debug("PersistableChannelUniqueQueue: %s Starting", q.delayedStarter.name)
q.lock.Lock()
if q.internal == nil {
err := q.setInternal(atShutdown, func(data ...Data) []Data {
for _, datum := range data {
err := q.Push(datum)
if err != nil && err != ErrAlreadyInQueue {
log.Error("Unable push to channelled queue: %v", err)
}
}
return nil
}, q.channelQueue.exemplar)
q.lock.Unlock()
if err != nil {
log.Fatal("Unable to create internal queue for %s Error: %v", q.Name(), err)
return
}
} else {
q.lock.Unlock()
}
atShutdown(q.Shutdown)
atTerminate(q.Terminate)
_ = q.channelQueue.AddWorkers(q.channelQueue.workers, 0)
if luq, ok := q.internal.(*LevelUniqueQueue); ok && !luq.IsEmpty() {
// Just run the level queue - we shut it down once it's flushed
go luq.Run(func(_ func()) {}, func(_ func()) {})
go func() {
_ = luq.Flush(0)
for !luq.IsEmpty() {
_ = luq.Flush(0)
select {
case <-time.After(100 * time.Millisecond):
case <-luq.shutdownCtx.Done():
if luq.byteFIFO.Len(luq.terminateCtx) > 0 {
log.Warn("LevelUniqueQueue: %s shut down before completely flushed", luq.Name())
}
return
}
}
log.Debug("LevelUniqueQueue: %s flushed so shutting down", luq.Name())
luq.Shutdown()
GetManager().Remove(luq.qid)
}()
} else {
log.Debug("PersistableChannelUniqueQueue: %s Skipping running the empty level queue", q.delayedStarter.name)
_ = q.internal.Flush(0)
q.internal.(*LevelUniqueQueue).Shutdown()
GetManager().Remove(q.internal.(*LevelUniqueQueue).qid)
}
}
// Flush flushes the queue
func (q *PersistableChannelUniqueQueue) Flush(timeout time.Duration) error {
return q.channelQueue.Flush(timeout)
}
// FlushWithContext flushes the queue
func (q *PersistableChannelUniqueQueue) FlushWithContext(ctx context.Context) error {
return q.channelQueue.FlushWithContext(ctx)
}
// IsEmpty checks if a queue is empty
func (q *PersistableChannelUniqueQueue) IsEmpty() bool {
return q.channelQueue.IsEmpty()
}
// IsPaused will return if the pool or queue is paused
func (q *PersistableChannelUniqueQueue) IsPaused() bool {
return q.channelQueue.IsPaused()
}
// Pause will pause the pool or queue
func (q *PersistableChannelUniqueQueue) Pause() {
q.channelQueue.Pause()
}
// Resume will resume the pool or queue
func (q *PersistableChannelUniqueQueue) Resume() {
q.channelQueue.Resume()
}
// IsPausedIsResumed will return a bool indicating if the pool or queue is paused and a channel that will be closed when it is resumed
func (q *PersistableChannelUniqueQueue) IsPausedIsResumed() (paused, resumed <-chan struct{}) {
return q.channelQueue.IsPausedIsResumed()
}
// Shutdown processing this queue
func (q *PersistableChannelUniqueQueue) Shutdown() {
log.Trace("PersistableChannelUniqueQueue: %s Shutting down", q.delayedStarter.name)
q.lock.Lock()
select {
case <-q.closed:
q.lock.Unlock()
return
default:
if q.internal != nil {
q.internal.(*LevelUniqueQueue).Shutdown()
}
close(q.closed)
q.lock.Unlock()
}
log.Trace("PersistableChannelUniqueQueue: %s Cancelling pools", q.delayedStarter.name)
q.internal.(*LevelUniqueQueue).baseCtxCancel()
q.channelQueue.baseCtxCancel()
log.Trace("PersistableChannelUniqueQueue: %s Waiting til done", q.delayedStarter.name)
q.channelQueue.Wait()
q.internal.(*LevelUniqueQueue).Wait()
// Redirect all remaining data in the chan to the internal channel
close(q.channelQueue.dataChan)
log.Trace("PersistableChannelUniqueQueue: %s Redirecting remaining data", q.delayedStarter.name)
countOK, countLost := 0, 0
for data := range q.channelQueue.dataChan {
err := q.internal.(*LevelUniqueQueue).Push(data)
if err != nil {
log.Error("PersistableChannelUniqueQueue: %s Unable redirect %v due to: %v", q.delayedStarter.name, data, err)
countLost++
} else {
countOK++
}
}
if countLost > 0 {
log.Warn("PersistableChannelUniqueQueue: %s %d will be restored on restart, %d lost", q.delayedStarter.name, countOK, countLost)
} else if countOK > 0 {
log.Warn("PersistableChannelUniqueQueue: %s %d will be restored on restart", q.delayedStarter.name, countOK)
}
log.Trace("PersistableChannelUniqueQueue: %s Done Redirecting remaining data", q.delayedStarter.name)
log.Debug("PersistableChannelUniqueQueue: %s Shutdown", q.delayedStarter.name)
}
// Terminate this queue and close the queue
func (q *PersistableChannelUniqueQueue) Terminate() {
log.Trace("PersistableChannelUniqueQueue: %s Terminating", q.delayedStarter.name)
q.Shutdown()
q.lock.Lock()
defer q.lock.Unlock()
if q.internal != nil {
q.internal.(*LevelUniqueQueue).Terminate()
}
q.channelQueue.baseCtxFinished()
log.Debug("PersistableChannelUniqueQueue: %s Terminated", q.delayedStarter.name)
}
func init() {
queuesMap[PersistableChannelUniqueQueueType] = NewPersistableChannelUniqueQueue
}