Debug: make indexing timeout configurable

build/builder.go

@@ -1088,8 +1088,11 @@ func (b *Builder) writeShard(fn string, ib *zoekt.IndexBuilder) (*finishedShard,
 		return nil, err
 	}
 
-	log.Printf("finished %s: %d index bytes (overhead %3.1f)", fn, fi.Size(),
-		float64(fi.Size())/float64(ib.ContentSize()+1))
+	log.Printf("finished shard %s: %d index bytes (overhead %3.1f), %d files processed \n",
+		fn,
+		fi.Size(),
+		float64(fi.Size())/float64(ib.ContentSize()+1),
+		ib.NumFiles())
 
 	return &finishedShard{f.Name(), fn}, nil
 }

cmd/zoekt-sourcegraph-indexserver/index.go

@@ -25,9 +25,7 @@ import (
 	sglog "github.com/sourcegraph/log"
 )
 
-// indexTimeout defines how long the indexserver waits before
-// killing an indexing job.
-const indexTimeout = 1*time.Hour + 30*time.Minute // an index should never take longer than an hour and a half
+const defaultIndexingTimeout = 1*time.Hour + 30*time.Minute
 
 // IndexOptions are the options that Sourcegraph can set via it's search
 // configuration endpoint.
@@ -163,6 +161,9 @@ type gitIndexConfig struct {
 	// The primary purpose of this configuration option is to be able to provide a stub
 	// implementation for this in our test suite. All other callers should use build.Options.FindRepositoryMetadata().
 	findRepositoryMetadata func(args *indexArgs) (repository *zoekt.Repository, metadata *zoekt.IndexMetadata, ok bool, err error)
+
+	// timeout defines how long the index server waits before killing an indexing job.
+	timeout time.Duration
 }
 
 func gitIndex(c gitIndexConfig, o *indexArgs, sourcegraph Sourcegraph, l sglog.Logger) error {
@@ -182,8 +183,7 @@ func gitIndex(c gitIndexConfig, o *indexArgs, sourcegraph Sourcegraph, l sglog.L
 	}
 
 	buildOptions := o.BuildOptions()
-
-	ctx, cancel := context.WithTimeout(context.Background(), indexTimeout)
+	ctx, cancel := context.WithTimeout(context.Background(), c.timeout)
 	defer cancel()
 
 	gitDir, err := tmpGitDir(o.Name)

cmd/zoekt-sourcegraph-indexserver/main.go

@@ -82,7 +82,7 @@ var (
 		Help: "A histogram of latencies for indexing a repository.",
 		Buckets: prometheus.ExponentialBucketsRange(
 			(100 * time.Millisecond).Seconds(),
-			(40*time.Minute + indexTimeout).Seconds(), // add an extra 40 minutes to account for the time it takes to clone the repo
+			(40*time.Minute + defaultIndexingTimeout).Seconds(), // add an extra 40 minutes to account for the time it takes to clone the repo
 			20),
 	}, []string{
 		"state", // state is an indexState
@@ -188,6 +188,9 @@ type Server struct {
 	hostname string
 
 	mergeOpts mergeOpts
+
+	// timeout defines how long the index server waits before killing an indexing job.
+	timeout time.Duration
 }
 
 var debug = log.New(io.Discard, "", log.LstdFlags)
@@ -585,6 +588,7 @@ func (s *Server) Index(args *indexArgs) (state indexState, err error) {
 		findRepositoryMetadata: func(args *indexArgs) (repository *zoekt.Repository, metadata *zoekt.IndexMetadata, ok bool, err error) {
 			return args.BuildOptions().FindRepositoryMetadata()
 		},
+		timeout: s.timeout,
 	}
 
 	err = gitIndex(c, args, s.Sourcegraph, s.logger)
@@ -1369,6 +1373,11 @@ func newServer(conf rootConfig) (*Server, error) {
 		debug.Printf("skipping generating symbols metadata for: %s", joinStringSet(reposShouldSkipSymbolsCalculation, ", "))
 	}
 
+	indexingTimeout := getEnvWithDefaultDuration("INDEXING_TIMEOUT", defaultIndexingTimeout)
+	if indexingTimeout != defaultIndexingTimeout {
+		debug.Printf("using configured indexing timeout: %s", indexingTimeout)
+	}
+
 	var sg Sourcegraph
 	if rootURL.IsAbs() {
 		var batchSize int
@@ -1432,6 +1441,7 @@ func newServer(conf rootConfig) (*Server, error) {
 			minAgeDays:      conf.minAgeDays,
 			maxPriority:     conf.maxPriority,
 		},
+		timeout: indexingTimeout,
 	}, err
 }
 

gitindex/index.go

@@ -529,15 +529,19 @@ func indexGitRepo(opts Options, config gitIndexConfig) error {
 
 	var names []string
 	fileKeys := map[string][]fileKey{}
+	totalFiles := 0
+
 	for key := range repos {
 		n := key.FullPath()
 		fileKeys[n] = append(fileKeys[n], key)
 		names = append(names, n)
+		totalFiles++
 	}
 
 	sort.Strings(names)
 	names = uniq(names)
 
+	log.Printf("attempting to index %d total files", totalFiles)
 	for _, name := range names {
 		keys := fileKeys[name]
 

indexbuilder.go

@@ -230,6 +230,11 @@ func (b *IndexBuilder) ContentSize() uint32 {
 	return b.contentPostings.endByte + b.namePostings.endByte
 }
 
+// NumFiles returns the number of files added to this builder
+func (b *IndexBuilder) NumFiles() int {
+	return len(b.contentStrings)
+}
+
 // NewIndexBuilder creates a fresh IndexBuilder. The passed in
 // Repository contains repo metadata, and may be set to nil.
 func NewIndexBuilder(r *Repository) (*IndexBuilder, error) {

merge.go

@@ -82,7 +82,7 @@ func builderWriteAll(fn string, ib *IndexBuilder) error {
 		return err
 	}
 
-	log.Printf("finished %s: %d index bytes (overhead %3.1f)", fn, fi.Size(),
+	log.Printf("finished shard %s: %d index bytes (overhead %3.1f)", fn, fi.Size(),
 		float64(fi.Size())/float64(ib.ContentSize()+1))
 
 	return nil