1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
|
// Copyright 2016 The Gitea Authors. All rights reserved.
// SPDX-License-Identifier: MIT
package code
import (
"context"
"os"
"runtime/pprof"
"slices"
"sync/atomic"
"time"
"code.gitea.io/gitea/models/db"
repo_model "code.gitea.io/gitea/models/repo"
"code.gitea.io/gitea/modules/graceful"
"code.gitea.io/gitea/modules/indexer/code/bleve"
"code.gitea.io/gitea/modules/indexer/code/elasticsearch"
"code.gitea.io/gitea/modules/indexer/code/internal"
"code.gitea.io/gitea/modules/log"
"code.gitea.io/gitea/modules/process"
"code.gitea.io/gitea/modules/queue"
"code.gitea.io/gitea/modules/setting"
)
var (
indexerQueue *queue.WorkerPoolQueue[*internal.IndexerData]
// globalIndexer is the global indexer, it cannot be nil.
// When the real indexer is not ready, it will be a dummy indexer which will return error to explain it's not ready.
// So it's always safe use it as *globalIndexer.Load() and call its methods.
globalIndexer atomic.Pointer[internal.Indexer]
dummyIndexer *internal.Indexer
)
func init() {
i := internal.NewDummyIndexer()
dummyIndexer = &i
globalIndexer.Store(dummyIndexer)
}
func index(ctx context.Context, indexer internal.Indexer, repoID int64) error {
repo, err := repo_model.GetRepositoryByID(ctx, repoID)
if repo_model.IsErrRepoNotExist(err) {
return indexer.Delete(ctx, repoID)
}
if err != nil {
return err
}
repoTypes := setting.Indexer.RepoIndexerRepoTypes
if len(repoTypes) == 0 {
repoTypes = []string{"sources"}
}
// skip forks from being indexed if unit is not present
if !slices.Contains(repoTypes, "forks") && repo.IsFork {
return nil
}
// skip mirrors from being indexed if unit is not present
if !slices.Contains(repoTypes, "mirrors") && repo.IsMirror {
return nil
}
// skip templates from being indexed if unit is not present
if !slices.Contains(repoTypes, "templates") && repo.IsTemplate {
return nil
}
// skip regular repos from being indexed if unit is not present
if !slices.Contains(repoTypes, "sources") && !repo.IsFork && !repo.IsMirror && !repo.IsTemplate {
return nil
}
sha, err := getDefaultBranchSha(ctx, repo)
if err != nil {
return err
}
changes, err := getRepoChanges(ctx, repo, sha)
if err != nil {
return err
} else if changes == nil {
return nil
}
if err := indexer.Index(ctx, repo, sha, changes); err != nil {
return err
}
return repo_model.UpdateIndexerStatus(ctx, repo, repo_model.RepoIndexerTypeCode, sha)
}
// Init initialize the repo indexer
func Init() {
if !setting.Indexer.RepoIndexerEnabled {
(*globalIndexer.Load()).Close()
return
}
ctx, cancel, finished := process.GetManager().AddTypedContext(context.Background(), "Service: CodeIndexer", process.SystemProcessType, false)
graceful.GetManager().RunAtTerminate(func() {
select {
case <-ctx.Done():
return
default:
}
cancel()
log.Debug("Closing repository indexer")
(*globalIndexer.Load()).Close()
log.Info("PID: %d Repository Indexer closed", os.Getpid())
finished()
})
waitChannel := make(chan time.Duration, 1)
// Create the Queue
switch setting.Indexer.RepoType {
case "bleve", "elasticsearch":
handler := func(items ...*internal.IndexerData) (unhandled []*internal.IndexerData) {
indexer := *globalIndexer.Load()
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 {
unhandled = append(unhandled, indexerData)
if !setting.IsInTesting {
log.Error("Codes indexer handler: index error for repo %v: %v", indexerData.RepoID, err)
}
}
}
return unhandled
}
indexerQueue = queue.CreateUniqueQueue(ctx, "code_indexer", handler)
if indexerQueue == nil {
log.Fatal("Unable to create codes indexer queue")
}
default:
log.Fatal("Unknown codes indexer type; %s", setting.Indexer.RepoType)
}
go func() {
pprof.SetGoroutineLabels(ctx)
start := time.Now()
var (
rIndexer internal.Indexer
existed bool
err error
)
switch setting.Indexer.RepoType {
case "bleve":
log.Info("PID: %d Initializing Repository Indexer at: %s", os.Getpid(), setting.Indexer.RepoPath)
defer func() {
if err := recover(); err != nil {
log.Error("PANIC whilst initializing repository indexer: %v\nStacktrace: %s", err, log.Stack(2))
log.Error("The indexer files are likely corrupted and may need to be deleted")
log.Error("You can completely remove the \"%s\" directory to make Gitea recreate the indexes", setting.Indexer.RepoPath)
}
}()
rIndexer = bleve.NewIndexer(setting.Indexer.RepoPath)
existed, err = rIndexer.Init(ctx)
if err != nil {
cancel()
(*globalIndexer.Load()).Close()
close(waitChannel)
log.Fatal("PID: %d Unable to initialize the bleve Repository Indexer at path: %s Error: %v", os.Getpid(), setting.Indexer.RepoPath, err)
}
case "elasticsearch":
log.Info("PID: %d Initializing Repository Indexer at: %s", os.Getpid(), setting.Indexer.RepoConnStr)
defer func() {
if err := recover(); err != nil {
log.Error("PANIC whilst initializing repository indexer: %v\nStacktrace: %s", err, log.Stack(2))
log.Error("The indexer files are likely corrupted and may need to be deleted")
log.Error("You can completely remove the \"%s\" index to make Gitea recreate the indexes", setting.Indexer.RepoConnStr)
}
}()
rIndexer = elasticsearch.NewIndexer(setting.Indexer.RepoConnStr, setting.Indexer.RepoIndexerName)
if err != nil {
cancel()
(*globalIndexer.Load()).Close()
close(waitChannel)
log.Fatal("PID: %d Unable to create the elasticsearch Repository Indexer connstr: %s Error: %v", os.Getpid(), setting.Indexer.RepoConnStr, err)
}
existed, err = rIndexer.Init(ctx)
if err != nil {
cancel()
(*globalIndexer.Load()).Close()
close(waitChannel)
log.Fatal("PID: %d Unable to initialize the elasticsearch Repository Indexer connstr: %s Error: %v", os.Getpid(), setting.Indexer.RepoConnStr, err)
}
default:
log.Fatal("PID: %d Unknown Indexer type: %s", os.Getpid(), setting.Indexer.RepoType)
}
globalIndexer.Store(&rIndexer)
// Start processing the queue
go graceful.GetManager().RunWithCancel(indexerQueue)
if !existed { // populate the index because it's created for the first time
go graceful.GetManager().RunWithShutdownContext(populateRepoIndexer)
}
select {
case waitChannel <- time.Since(start):
case <-graceful.GetManager().IsShutdown():
}
close(waitChannel)
}()
if setting.Indexer.StartupTimeout > 0 {
go func() {
pprof.SetGoroutineLabels(ctx)
timeout := setting.Indexer.StartupTimeout
if graceful.GetManager().IsChild() && setting.GracefulHammerTime > 0 {
timeout += setting.GracefulHammerTime
}
select {
case <-graceful.GetManager().IsShutdown():
log.Warn("Shutdown before Repository Indexer completed initialization")
cancel()
(*globalIndexer.Load()).Close()
case duration, ok := <-waitChannel:
if !ok {
log.Warn("Repository Indexer Initialization failed")
cancel()
(*globalIndexer.Load()).Close()
return
}
log.Info("Repository Indexer Initialization took %v", duration)
case <-time.After(timeout):
cancel()
(*globalIndexer.Load()).Close()
log.Fatal("Repository Indexer Initialization Timed-Out after: %v", timeout)
}
}()
}
}
// UpdateRepoIndexer update a repository's entries in the indexer
func UpdateRepoIndexer(repo *repo_model.Repository) {
indexData := &internal.IndexerData{RepoID: repo.ID}
if err := indexerQueue.Push(indexData); err != nil {
log.Error("Update repo index data %v failed: %v", indexData, err)
}
}
// IsAvailable checks if issue indexer is available
func IsAvailable(ctx context.Context) bool {
return (*globalIndexer.Load()).Ping(ctx) == nil
}
// populateRepoIndexer populate the repo indexer with pre-existing data. This
// should only be run when the indexer is created for the first time.
func populateRepoIndexer(ctx context.Context) {
log.Info("Populating the repo indexer with existing repositories")
exist, err := db.IsTableNotEmpty("repository")
if err != nil {
log.Fatal("System error: %v", err)
} else if !exist {
return
}
// if there is any existing repo indexer metadata in the DB, delete it
// since we are starting afresh. Also, xorm requires deletes to have a
// condition, and we want to delete everything, thus 1=1.
if err := db.DeleteAllRecords("repo_indexer_status"); err != nil {
log.Fatal("System error: %v", err)
}
var maxRepoID int64
if maxRepoID, err = db.GetMaxID("repository"); err != nil {
log.Fatal("System error: %v", err)
}
// start with the maximum existing repo ID and work backwards, so that we
// don't include repos that are created after gitea starts; such repos will
// already be added to the indexer, and we don't need to add them again.
for maxRepoID > 0 {
select {
case <-ctx.Done():
log.Info("Repository Indexer population shutdown before completion")
return
default:
}
ids, err := repo_model.GetUnindexedRepos(repo_model.RepoIndexerTypeCode, maxRepoID, 0, 50)
if err != nil {
log.Error("populateRepoIndexer: %v", err)
return
} else if len(ids) == 0 {
break
}
for _, id := range ids {
select {
case <-ctx.Done():
log.Info("Repository Indexer population shutdown before completion")
return
default:
}
if err := indexerQueue.Push(&internal.IndexerData{RepoID: id}); err != nil {
log.Error("indexerQueue.Push: %v", err)
return
}
maxRepoID = id - 1
}
}
log.Info("Done (re)populating the repo indexer with existing repositories")
}
|