Rewrite queue (#24505)

# ⚠️ Breaking

Many deprecated queue config options are removed (actually, they should
have been removed in 1.18/1.19).

If you see the fatal message when starting Gitea: "Please update your
app.ini to remove deprecated config options", please follow the error
messages to remove these options from your app.ini.

Example:

```
2023/05/06 19:39:22 [E] Removed queue option: `[indexer].ISSUE_INDEXER_QUEUE_TYPE`. Use new options in `[queue.issue_indexer]`
2023/05/06 19:39:22 [E] Removed queue option: `[indexer].UPDATE_BUFFER_LEN`. Use new options in `[queue.issue_indexer]`
2023/05/06 19:39:22 [F] Please update your app.ini to remove deprecated config options
```

Many options in `[queue]` are are dropped, including:
`WRAP_IF_NECESSARY`, `MAX_ATTEMPTS`, `TIMEOUT`, `WORKERS`,
`BLOCK_TIMEOUT`, `BOOST_TIMEOUT`, `BOOST_WORKERS`, they can be removed
from app.ini.

# The problem

The old queue package has some legacy problems:

* complexity: I doubt few people could tell how it works.
* maintainability: Too many channels and mutex/cond are mixed together,
too many different structs/interfaces depends each other.
* stability: due to the complexity & maintainability, sometimes there
are strange bugs and difficult to debug, and some code doesn't have test
(indeed some code is difficult to test because a lot of things are mixed
together).
* general applicability: although it is called "queue", its behavior is
not a well-known queue.
* scalability: it doesn't seem easy to make it work with a cluster
without breaking its behaviors.

It came from some very old code to "avoid breaking", however, its
technical debt is too heavy now. It's a good time to introduce a better
"queue" package.

# The new queue package

It keeps using old config and concept as much as possible.

* It only contains two major kinds of concepts:
    * The "base queue": channel, levelqueue, redis
* They have the same abstraction, the same interface, and they are
tested by the same testing code.
* The "WokerPoolQueue", it uses the "base queue" to provide "worker
pool" function, calls the "handler" to process the data in the base
queue.
* The new code doesn't do "PushBack"
* Think about a queue with many workers, the "PushBack" can't guarantee
the order for re-queued unhandled items, so in new code it just does
"normal push"
* The new code doesn't do "pause/resume"
* The "pause/resume" was designed to handle some handler's failure: eg:
document indexer (elasticsearch) is down
* If a queue is paused for long time, either the producers blocks or the
new items are dropped.
* The new code doesn't do such "pause/resume" trick, it's not a common
queue's behavior and it doesn't help much.
* If there are unhandled items, the "push" function just blocks for a
few seconds and then re-queue them and retry.
* The new code doesn't do "worker booster"
* Gitea's queue's handlers are light functions, the cost is only the
go-routine, so it doesn't make sense to "boost" them.
* The new code only use "max worker number" to limit the concurrent
workers.
* The new "Push" never blocks forever
* Instead of creating more and more blocking goroutines, return an error
is more friendly to the server and to the end user.

There are more details in code comments: eg: the "Flush" problem, the
strange "code.index" hanging problem, the "immediate" queue problem.

Almost ready for review.

TODO:

* [x] add some necessary comments during review
* [x] add some more tests if necessary
* [x] update documents and config options
* [x] test max worker / active worker
* [x] re-run the CI tasks to see whether any test is flaky
* [x] improve the `handleOldLengthConfiguration` to provide more
friendly messages
* [x] fine tune default config values (eg: length?)

## Code coverage:

![image](https://user-images.githubusercontent.com/2114189/236620635-55576955-f95d-4810-b12f-879026a3afdf.png)
This commit is contained in:
2023-05-08 19:49:59 +08:00
committed by GitHub
parent cb700aedd1
commit 6f9c278559
100 changed files with 2496 additions and 6858 deletions

View File

@ -273,10 +273,6 @@ func (b *BleveIndexer) Close() {
log.Info("PID: %d Repository Indexer closed", os.Getpid())
}
// SetAvailabilityChangeCallback does nothing
func (b *BleveIndexer) SetAvailabilityChangeCallback(callback func(bool)) {
}
// Ping does nothing
func (b *BleveIndexer) Ping() bool {
return true

View File

@ -42,12 +42,11 @@ var _ Indexer = &ElasticSearchIndexer{}
// ElasticSearchIndexer implements Indexer interface
type ElasticSearchIndexer struct {
client *elastic.Client
indexerAliasName string
available bool
availabilityCallback func(bool)
stopTimer chan struct{}
lock sync.RWMutex
client *elastic.Client
indexerAliasName string
available bool
stopTimer chan struct{}
lock sync.RWMutex
}
type elasticLogger struct {
@ -198,13 +197,6 @@ func (b *ElasticSearchIndexer) init() (bool, error) {
return exists, nil
}
// SetAvailabilityChangeCallback sets callback that will be triggered when availability changes
func (b *ElasticSearchIndexer) SetAvailabilityChangeCallback(callback func(bool)) {
b.lock.Lock()
defer b.lock.Unlock()
b.availabilityCallback = callback
}
// Ping checks if elastic is available
func (b *ElasticSearchIndexer) Ping() bool {
b.lock.RLock()
@ -529,8 +521,4 @@ func (b *ElasticSearchIndexer) setAvailability(available bool) {
}
b.available = available
if b.availabilityCallback != nil {
// Call the callback from within the lock to ensure that the ordering remains correct
b.availabilityCallback(b.available)
}
}

View File

@ -44,7 +44,6 @@ type SearchResultLanguages struct {
// Indexer defines an interface to index and search code contents
type Indexer interface {
Ping() bool
SetAvailabilityChangeCallback(callback func(bool))
Index(ctx context.Context, repo *repo_model.Repository, sha string, changes *repoChanges) error
Delete(repoID int64) error
Search(ctx context.Context, repoIDs []int64, language, keyword string, page, pageSize int, isMatch bool) (int64, []*SearchResult, []*SearchResultLanguages, error)
@ -81,7 +80,7 @@ type IndexerData struct {
RepoID int64
}
var indexerQueue queue.UniqueQueue
var indexerQueue *queue.WorkerPoolQueue[*IndexerData]
func index(ctx context.Context, indexer Indexer, repoID int64) error {
repo, err := repo_model.GetRepositoryByID(ctx, repoID)
@ -137,37 +136,45 @@ func Init() {
// Create the Queue
switch setting.Indexer.RepoType {
case "bleve", "elasticsearch":
handler := func(data ...queue.Data) []queue.Data {
handler := func(items ...*IndexerData) (unhandled []*IndexerData) {
idx, err := indexer.get()
if idx == nil || err != nil {
log.Error("Codes indexer handler: unable to get indexer!")
return data
return items
}
unhandled := make([]queue.Data, 0, len(data))
for _, datum := range data {
indexerData, ok := datum.(*IndexerData)
if !ok {
log.Error("Unable to process provided datum: %v - not possible to cast to IndexerData", datum)
continue
}
for _, indexerData := range items {
log.Trace("IndexerData Process Repo: %d", indexerData.RepoID)
// FIXME: it seems there is a bug in `CatFileBatch` or `nio.Pipe`, which will cause the process to hang forever in rare cases
/*
sync.(*Cond).Wait(cond.go:70)
github.com/djherbis/nio/v3.(*PipeReader).Read(sync.go:106)
bufio.(*Reader).fill(bufio.go:106)
bufio.(*Reader).ReadSlice(bufio.go:372)
bufio.(*Reader).collectFragments(bufio.go:447)
bufio.(*Reader).ReadString(bufio.go:494)
code.gitea.io/gitea/modules/git.ReadBatchLine(batch_reader.go:149)
code.gitea.io/gitea/modules/indexer/code.(*BleveIndexer).addUpdate(bleve.go:214)
code.gitea.io/gitea/modules/indexer/code.(*BleveIndexer).Index(bleve.go:296)
code.gitea.io/gitea/modules/indexer/code.(*wrappedIndexer).Index(wrapped.go:74)
code.gitea.io/gitea/modules/indexer/code.index(indexer.go:105)
*/
if err := index(ctx, indexer, indexerData.RepoID); err != nil {
if !setting.IsInTesting {
log.Error("indexer index error for repo %v: %v", indexerData.RepoID, err)
}
if indexer.Ping() {
if !idx.Ping() {
log.Error("Code indexer handler: indexer is unavailable.")
unhandled = append(unhandled, indexerData)
continue
}
// Add back to queue
unhandled = append(unhandled, datum)
if !setting.IsInTesting {
log.Error("Codes indexer handler: index error for repo %v: %v", indexerData.RepoID, err)
}
}
}
return unhandled
}
indexerQueue = queue.CreateUniqueQueue("code_indexer", handler, &IndexerData{})
indexerQueue = queue.CreateUniqueQueue("code_indexer", handler)
if indexerQueue == nil {
log.Fatal("Unable to create codes indexer queue")
}
@ -224,18 +231,6 @@ func Init() {
indexer.set(rIndexer)
if queue, ok := indexerQueue.(queue.Pausable); ok {
rIndexer.SetAvailabilityChangeCallback(func(available bool) {
if !available {
log.Info("Code index queue paused")
queue.Pause()
} else {
log.Info("Code index queue resumed")
queue.Resume()
}
})
}
// Start processing the queue
go graceful.GetManager().RunWithShutdownFns(indexerQueue.Run)

View File

@ -56,16 +56,6 @@ func (w *wrappedIndexer) get() (Indexer, error) {
return w.internal, nil
}
// SetAvailabilityChangeCallback sets callback that will be triggered when availability changes
func (w *wrappedIndexer) SetAvailabilityChangeCallback(callback func(bool)) {
indexer, err := w.get()
if err != nil {
log.Error("Failed to get indexer: %v", err)
return
}
indexer.SetAvailabilityChangeCallback(callback)
}
// Ping checks if elastic is available
func (w *wrappedIndexer) Ping() bool {
indexer, err := w.get()