Files
syncthing/cmd/syncthing/monitor.go
T
Jakob BorgandRoss Smith II 836045ee87 feat: switch logging framework (#10220)
This updates our logging framework from legacy freetext strings using
the `log` package to structured log entries using `log/slog`. I have
updated all INFO or higher level entries, but not yet DEBUG (😓)... So,
at a high level:

There is a slight change in log levels, effectively adding a new warning
level:

- DEBUG is still debug (ideally not for users but developers, though
this is something we need to work on)
- INFO is still info, though I've added more data here, effectively
making Syncthing more verbose by default (more on this below)
- WARNING is a new log level that is different from the _old_ WARNING
(more below)
- ERROR is what was WARNING before -- problems that must be dealt with,
and also bubbled as a popup in the GUI.

A new feature is that the logging level can be set per package to
something other than just debug or info, and hence I feel that we can
add a bit more things into INFO while moving some (in fact, most)
current INFO level warnings into WARNING. For example, I think it's
justified to get a log of synced files in INFO and sync failures in
WARNING. These are things that have historically been tricky to debug
properly, and having more information by default will be useful to many,
while still making it possible get close to told level of inscrutability
by setting the log level to WARNING. I'd like to get to a stage where
DEBUG is never necessary to just figure out what's going on, as opposed
to trying to narrow down a likely bug.

Code wise:

- Our logging object, generally known as `l` in each package, is now a
new adapter object that provides the old API on top of the newer one.
(This should go away once all old log entries are migrated.) This is
only for `l.Debugln` and `l.Debugf`.
- There is a new level tracker that keeps the log level for each
package.
- There is a nested setup of handlers, since the structure mandated by
`log/slog` is slightly convoluted (imho). We do this because we need to
do formatting at a "medium" level internally so we can buffer log lines
in text format but with separate timestamp and log level for the API/GUI
to consume.
- The `debug` API call becomes a `loglevels` API call, which can set the
log level to `DEBUG`, `INFO`, `WARNING` or `ERROR` per package. The GUI
is updated to handle this.
- Our custom `sync` package provided some debugging of mutexes quite
strongly integrated into the old logging framework, only turned on when
`STTRACE` was set to certain values at startup, etc. It's been a long
time since this has been useful; I removed it.
- The `STTRACE` env var remains and can be used the same way as before,
while additionally permitting specific log levels to be specified,
`STTRACE=model:WARN,scanner:DEBUG`.
- There is a new command line option `--log-level=INFO` to set the
default log level.
- The command line options `--log-flags` and `--verbose` go away, but
are currently retained as hidden & ignored options since we set them by
default in some of our startup examples and Syncthing would otherwise
fail to start.

Sample format messages:

```
2009-02-13 23:31:30 INF A basic info line (attr1="val with spaces" attr2=2 attr3="val\"quote" a=a log.pkg=slogutil)
2009-02-13 23:31:30 INF An info line with grouped values (attr1=val1 foo.attr2=2 foo.bar.attr3=3 a=a log.pkg=slogutil)
2009-02-13 23:31:30 INF An info line with grouped values via logger (foo.attr1=val1 foo.attr2=2 a=a log.pkg=slogutil)
2009-02-13 23:31:30 INF An info line with nested grouped values via logger (bar.foo.attr1=val1 bar.foo.attr2=2 a=a log.pkg=slogutil)
2009-02-13 23:31:30 WRN A warning entry (a=a log.pkg=slogutil)
2009-02-13 23:31:30 ERR An error (a=a log.pkg=slogutil)
```

---------

Co-authored-by: Ross Smith II <ross@smithii.com>
2025-08-07 11:19:36 +02:00

568 lines
14 KiB
Go

// Copyright (C) 2014 The Syncthing Authors.
//
// This Source Code Form is subject to the terms of the Mozilla Public
// License, v. 2.0. If a copy of the MPL was not distributed with this file,
// You can obtain one at https://mozilla.org/MPL/2.0/.
package main
import (
"bufio"
"context"
"fmt"
"io"
"log/slog"
"os"
"os/exec"
"os/signal"
"path/filepath"
"strings"
"sync"
"syscall"
"time"
"github.com/syncthing/syncthing/internal/slogutil"
"github.com/syncthing/syncthing/lib/build"
"github.com/syncthing/syncthing/lib/fs"
"github.com/syncthing/syncthing/lib/locations"
"github.com/syncthing/syncthing/lib/osutil"
"github.com/syncthing/syncthing/lib/svcutil"
)
var (
stdoutFirstLines []string // The first 10 lines of stdout
stdoutLastLines []string // The last 50 lines of stdout
stdoutMut sync.Mutex
)
const (
restartCounts = 4
restartPause = 1 * time.Second
restartLoopThreshold = 60 * time.Second
logFileAutoCloseDelay = 5 * time.Second
logFileMaxOpenTime = time.Minute
panicUploadMaxWait = 30 * time.Second
panicUploadNoticeWait = 10 * time.Second
)
func (c *serveCmd) monitorMain() {
var dst io.Writer = os.Stdout
logFile := locations.Get(locations.LogFile)
if logFile != "-" {
if expanded, err := fs.ExpandTilde(logFile); err == nil {
logFile = expanded
}
var fileDst io.Writer
var err error
open := func(name string) (io.WriteCloser, error) {
return newAutoclosedFile(name, logFileAutoCloseDelay, logFileMaxOpenTime)
}
if c.LogMaxSize > 0 {
fileDst, err = newRotatedFile(logFile, open, int64(c.LogMaxSize), c.LogMaxFiles)
} else {
fileDst, err = open(logFile)
}
if err != nil {
slog.Error("Failed to set up logging to file, proceeding with logging to stdout only", slogutil.Error(err))
} else {
if build.IsWindows {
// Translate line breaks to Windows standard
fileDst = osutil.ReplacingWriter{
Writer: fileDst,
From: '\n',
To: []byte{'\r', '\n'},
}
}
// Log to both stdout and file.
dst = io.MultiWriter(dst, fileDst)
slog.Info("Saved log output", slogutil.FilePath(logFile))
}
}
args := os.Args
binary, err := getBinary(args[0])
if err != nil {
slog.Error("Failed to start the main Syncthing process", slogutil.Error(err))
panic("Error starting the main Syncthing process")
}
var restarts [restartCounts]time.Time
stopSign := make(chan os.Signal, 1)
signal.Notify(stopSign, os.Interrupt, sigTerm)
restartSign := make(chan os.Signal, 1)
sigHup := syscall.Signal(1)
signal.Notify(restartSign, sigHup)
childEnv := childEnv()
first := true
for {
maybeReportPanics()
if t := time.Since(restarts[0]); t < restartLoopThreshold {
slog.Error("Too many restarts; not retrying further", slog.Int("count", restartCounts), slog.Any("interval", t))
os.Exit(svcutil.ExitError.AsInt())
}
copy(restarts[0:], restarts[1:])
restarts[len(restarts)-1] = time.Now()
cmd := exec.Command(binary, args[1:]...)
cmd.Env = childEnv
stderr, err := cmd.StderrPipe()
if err != nil {
panic(err)
}
stdout, err := cmd.StdoutPipe()
if err != nil {
panic(err)
}
slog.Debug("Starting syncthing")
err = cmd.Start()
if err != nil {
slog.Error("Failed to start the main Syncthing process", slogutil.Error(err))
panic("Error starting the main Syncthing process")
}
stdoutMut.Lock()
stdoutFirstLines = make([]string, 0, 10)
stdoutLastLines = make([]string, 0, 50)
stdoutMut.Unlock()
var wg sync.WaitGroup
wg.Add(1)
go func() {
copyStderr(stderr, dst)
wg.Done()
}()
wg.Add(1)
go func() {
copyStdout(stdout, dst)
wg.Done()
}()
exit := make(chan error)
go func() {
wg.Wait()
exit <- cmd.Wait()
}()
stopped := false
select {
case s := <-stopSign:
slog.Info("Received signal; exiting", "signal", s)
cmd.Process.Signal(sigTerm)
err = <-exit
stopped = true
case s := <-restartSign:
slog.Info("Received signal; restarting", "signal", s)
cmd.Process.Signal(sigHup)
err = <-exit
case err = <-exit:
}
if err == nil {
// Successful exit indicates an intentional shutdown
os.Exit(svcutil.ExitSuccess.AsInt())
}
if exiterr, ok := err.(*exec.ExitError); ok {
exitCode := exiterr.ExitCode()
if stopped || c.NoRestart {
os.Exit(exitCode)
}
if exitCode == svcutil.ExitUpgrade.AsInt() {
// Restart the monitor process to release the .old
// binary as part of the upgrade process.
slog.Info("Restarting monitor...")
if err = restartMonitor(binary, args); err != nil {
slog.Error("Failed to restart monitor", slogutil.Error(err))
}
os.Exit(exitCode)
}
}
if c.NoRestart {
os.Exit(svcutil.ExitError.AsInt())
}
slog.Info("Syncthing exited", slogutil.Error(err))
time.Sleep(restartPause)
if first {
// Let the next child process know that this is not the first time
// it's starting up.
childEnv = append(childEnv, "STRESTART=yes")
first = false
}
}
}
func getBinary(args0 string) (string, error) {
e, err := os.Executable()
if err == nil {
return e, nil
}
// Check if args0 cuts it
e, lerr := exec.LookPath(args0)
if lerr == nil {
return e, nil
}
return "", err
}
func copyStderr(stderr io.Reader, dst io.Writer) {
br := bufio.NewReader(stderr)
var panicFd *os.File
defer func() {
if panicFd != nil {
_ = panicFd.Close()
maybeReportPanics()
}
}()
for {
line, err := br.ReadString('\n')
if err != nil {
return
}
dst.Write([]byte(line))
if panicFd == nil && (strings.HasPrefix(line, "panic:") || strings.HasPrefix(line, "fatal error:")) {
panicFd, err = os.Create(locations.GetTimestamped(locations.PanicLog))
if err != nil {
slog.Error("Failed to create panic log", slogutil.Error(err))
continue
}
slog.Error("Panic detected, writing to file", slogutil.FilePath(panicFd.Name()))
slog.Info("Please check for existing issues with similar panic message at https://github.com/syncthing/syncthing/issues/")
slog.Info("If no issue with similar panic message exists, please create a new issue with the panic log attached")
stdoutMut.Lock()
for _, line := range stdoutFirstLines {
panicFd.WriteString(line)
}
panicFd.WriteString("...\n")
for _, line := range stdoutLastLines {
panicFd.WriteString(line)
}
stdoutMut.Unlock()
panicFd.WriteString("Panic at " + time.Now().Format(time.RFC3339) + "\n")
}
if panicFd != nil {
panicFd.WriteString(line)
}
}
}
func copyStdout(stdout io.Reader, dst io.Writer) {
br := bufio.NewReader(stdout)
for {
line, err := br.ReadString('\n')
if err != nil {
return
}
stdoutMut.Lock()
if len(stdoutFirstLines) < cap(stdoutFirstLines) {
stdoutFirstLines = append(stdoutFirstLines, line)
} else {
if l := len(stdoutLastLines); l == cap(stdoutLastLines) {
stdoutLastLines = stdoutLastLines[:l-1]
}
stdoutLastLines = append(stdoutLastLines, line)
}
stdoutMut.Unlock()
dst.Write([]byte(line))
}
}
func restartMonitor(binary string, args []string) error {
// Set the STRESTART environment variable to indicate to the next
// process that this is a restart and not initial start. This prevents
// opening the browser on startup.
os.Setenv("STRESTART", "yes")
if !build.IsWindows {
// syscall.Exec is the cleanest way to restart on Unixes as it
// replaces the current process with the new one, keeping the pid and
// controlling terminal and so on
return restartMonitorUnix(binary, args)
}
// but it isn't supported on Windows, so there we start a normal
// exec.Command and return.
return restartMonitorWindows(binary, args)
}
func restartMonitorUnix(binary string, args []string) error {
return syscall.Exec(binary, args, os.Environ())
}
func restartMonitorWindows(binary string, args []string) error {
cmd := exec.Command(binary, args[1:]...)
// Retain the standard streams
cmd.Stderr = os.Stderr
cmd.Stdout = os.Stdout
cmd.Stdin = os.Stdin
return cmd.Start()
}
// rotatedFile keeps a set of rotating logs. There will be the base file plus up
// to maxFiles rotated ones, each ~ maxSize bytes large.
type rotatedFile struct {
name string
create createFn
maxSize int64 // bytes
maxFiles int
currentFile io.WriteCloser
currentSize int64
}
type createFn func(name string) (io.WriteCloser, error)
func newRotatedFile(name string, create createFn, maxSize int64, maxFiles int) (*rotatedFile, error) {
var size int64
if info, err := os.Lstat(name); err != nil {
if !os.IsNotExist(err) {
return nil, err
}
size = 0
} else {
size = info.Size()
}
writer, err := create(name)
if err != nil {
return nil, err
}
return &rotatedFile{
name: name,
create: create,
maxSize: maxSize,
maxFiles: maxFiles,
currentFile: writer,
currentSize: size,
}, nil
}
func (r *rotatedFile) Write(bs []byte) (int, error) {
// Check if we're about to exceed the max size, and if so close this
// file so we'll start on a new one.
if r.currentSize+int64(len(bs)) > r.maxSize {
r.currentFile.Close()
r.currentSize = 0
r.rotate()
f, err := r.create(r.name)
if err != nil {
return 0, err
}
r.currentFile = f
}
n, err := r.currentFile.Write(bs)
r.currentSize += int64(n)
return n, err
}
func (r *rotatedFile) rotate() {
// The files are named "name", "name.0", "name.1", ...
// "name.(r.maxFiles-1)". Increase the numbers on the
// suffixed ones.
for i := r.maxFiles - 1; i > 0; i-- {
from := numberedFile(r.name, i-1)
to := numberedFile(r.name, i)
err := os.Rename(from, to)
if err != nil && !os.IsNotExist(err) {
fmt.Println("LOG: Rotating logs:", err)
}
}
// Rename the base to base.0
err := os.Rename(r.name, numberedFile(r.name, 0))
if err != nil && !os.IsNotExist(err) {
fmt.Println("LOG: Rotating logs:", err)
}
}
// numberedFile adds the number between the file name and the extension.
func numberedFile(name string, num int) string {
ext := filepath.Ext(name) // contains the dot
withoutExt := name[:len(name)-len(ext)]
return fmt.Sprintf("%s.%d%s", withoutExt, num, ext)
}
// An autoclosedFile is an io.WriteCloser that opens itself for appending on
// Write() and closes itself after an interval of no writes (closeDelay) or
// when the file has been open for too long (maxOpenTime). A call to Write()
// will return any error that happens on the resulting Open() call too. Errors
// on automatic Close() calls are silently swallowed...
type autoclosedFile struct {
name string // path to write to
closeDelay time.Duration // close after this long inactivity
maxOpenTime time.Duration // or this long after opening
fd io.WriteCloser // underlying WriteCloser
opened time.Time // timestamp when the file was last opened
closed chan struct{} // closed on Close(), stops the closerLoop
closeTimer *time.Timer // fires closeDelay after a write
mut sync.Mutex
}
func newAutoclosedFile(name string, closeDelay, maxOpenTime time.Duration) (*autoclosedFile, error) {
f := &autoclosedFile{
name: name,
closeDelay: closeDelay,
maxOpenTime: maxOpenTime,
closed: make(chan struct{}),
closeTimer: time.NewTimer(time.Minute),
}
f.mut.Lock()
defer f.mut.Unlock()
if err := f.ensureOpenLocked(); err != nil {
return nil, err
}
go f.closerLoop()
return f, nil
}
func (f *autoclosedFile) Write(bs []byte) (int, error) {
f.mut.Lock()
defer f.mut.Unlock()
// Make sure the file is open for appending
if err := f.ensureOpenLocked(); err != nil {
return 0, err
}
// If we haven't run into the maxOpenTime, postpone close for another
// closeDelay
if time.Since(f.opened) < f.maxOpenTime {
f.closeTimer.Reset(f.closeDelay)
}
return f.fd.Write(bs)
}
func (f *autoclosedFile) Close() error {
f.mut.Lock()
defer f.mut.Unlock()
// Stop the timer and closerLoop() routine
f.closeTimer.Stop()
close(f.closed)
// Close the file, if it's open
if f.fd != nil {
return f.fd.Close()
}
return nil
}
// Must be called with f.mut held!
func (f *autoclosedFile) ensureOpenLocked() error {
if f.fd != nil {
// File is already open
return nil
}
// We open the file for write only, and create it if it doesn't exist.
flags := os.O_WRONLY | os.O_CREATE | os.O_APPEND
fd, err := os.OpenFile(f.name, flags, 0o644)
if err != nil {
return err
}
f.fd = fd
f.opened = time.Now()
return nil
}
func (f *autoclosedFile) closerLoop() {
for {
select {
case <-f.closeTimer.C:
// Close the file when the timer expires.
f.mut.Lock()
if f.fd != nil {
f.fd.Close() // errors, schmerrors
f.fd = nil
}
f.mut.Unlock()
case <-f.closed:
return
}
}
}
// Returns the desired child environment, properly filtered and added to.
func childEnv() []string {
var env []string
for _, str := range os.Environ() {
if strings.HasPrefix(str, "STNORESTART=") {
continue
}
if strings.HasPrefix(str, "STMONITORED=") {
continue
}
env = append(env, str)
}
env = append(env, "STMONITORED=yes")
return env
}
// maybeReportPanics tries to figure out if crash reporting is on or off,
// and reports any panics it can find if it's enabled. We spend at most
// panicUploadMaxWait uploading panics...
func maybeReportPanics() {
// Try to get a config to see if/where panics should be reported.
cfg, err := loadOrDefaultConfig()
if err != nil {
slog.Error("Couldn't load config; not reporting crash")
return
}
// Bail if we're not supposed to report panics.
opts := cfg.Options()
if !opts.CREnabled {
return
}
// Set up a timeout on the whole operation.
ctx, cancel := context.WithTimeout(context.Background(), panicUploadMaxWait)
defer cancel()
// Print a notice if the upload takes a long time.
go func() {
select {
case <-ctx.Done():
return
case <-time.After(panicUploadNoticeWait):
slog.Warn("Uploading crash reports is taking a while, please wait")
}
}()
// Report the panics.
dir := locations.GetBaseDir(locations.ConfigBaseDir)
uploadPanicLogs(ctx, opts.CRURL, dir)
}