Collector implementation

This commit is contained in:
2026-03-14 20:07:22 +01:00
parent 4393ae2726
commit 6ca296b2e8
16 changed files with 3052 additions and 0 deletions

130
cmd/collector/main.go Normal file
View File

@@ -0,0 +1,130 @@
package main
import (
"bufio"
"context"
"flag"
"log"
"net"
"os"
"os/signal"
"path/filepath"
"strings"
"syscall"
pb "git.ipng.ch/ipng/nginx-logtail/proto/logtailpb"
"google.golang.org/grpc"
)
func main() {
listen := flag.String("listen", ":9090", "gRPC listen address")
logPaths := flag.String("logs", "", "comma-separated log file paths/globs to tail")
logsFile := flag.String("logs-file", "", "file containing one log path/glob per line")
source := flag.String("source", hostname(), "name for this collector (default: hostname)")
v4prefix := flag.Int("v4prefix", 24, "IPv4 prefix length for client bucketing")
v6prefix := flag.Int("v6prefix", 48, "IPv6 prefix length for client bucketing")
flag.Parse()
patterns := collectPatterns(*logPaths, *logsFile)
if len(patterns) == 0 {
log.Fatal("collector: no log paths specified; use --logs or --logs-file")
}
paths := expandGlobs(patterns)
if len(paths) == 0 {
log.Fatal("collector: no log files matched the specified patterns")
}
log.Printf("collector: tailing %d file(s)", len(paths))
ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
defer stop()
// Shared channel: tailer → store. Buffer absorbs ~20s of peak load.
ch := make(chan LogRecord, 200_000)
store := NewStore(*source)
go store.Run(ch)
tailer := NewMultiTailer(paths, *v4prefix, *v6prefix, ch)
go tailer.Run(ctx)
lis, err := net.Listen("tcp", *listen)
if err != nil {
log.Fatalf("collector: failed to listen on %s: %v", *listen, err)
}
grpcServer := grpc.NewServer()
pb.RegisterLogtailServiceServer(grpcServer, NewServer(store, *source))
go func() {
log.Printf("collector: gRPC server listening on %s (source=%s)", *listen, *source)
if err := grpcServer.Serve(lis); err != nil {
log.Printf("collector: gRPC server stopped: %v", err)
}
}()
<-ctx.Done()
log.Printf("collector: shutting down")
grpcServer.GracefulStop()
close(ch)
}
// collectPatterns merges patterns from --logs (comma-separated) and --logs-file.
func collectPatterns(logPaths, logsFile string) []string {
var patterns []string
for _, p := range strings.Split(logPaths, ",") {
if p = strings.TrimSpace(p); p != "" {
patterns = append(patterns, p)
}
}
if logsFile != "" {
f, err := os.Open(logsFile)
if err != nil {
log.Fatalf("collector: cannot open --logs-file %s: %v", logsFile, err)
}
defer f.Close()
sc := bufio.NewScanner(f)
for sc.Scan() {
if p := strings.TrimSpace(sc.Text()); p != "" && !strings.HasPrefix(p, "#") {
patterns = append(patterns, p)
}
}
}
return patterns
}
// expandGlobs expands any glob patterns and returns deduplicated concrete paths.
func expandGlobs(patterns []string) []string {
seen := make(map[string]struct{})
var paths []string
for _, pat := range patterns {
matches, err := filepath.Glob(pat)
if err != nil {
log.Printf("collector: invalid glob %q: %v", pat, err)
continue
}
if len(matches) == 0 {
// Keep the path even if it doesn't exist yet; the tailer will retry.
log.Printf("collector: pattern %q matched no files, will watch for creation", pat)
if _, ok := seen[pat]; !ok {
seen[pat] = struct{}{}
paths = append(paths, pat)
}
continue
}
for _, m := range matches {
if _, ok := seen[m]; !ok {
seen[m] = struct{}{}
paths = append(paths, m)
}
}
}
return paths
}
func hostname() string {
h, err := os.Hostname()
if err != nil {
return "unknown"
}
return h
}