Logging is indeed very important when developing software applications, because logs help us to debug, monitor, and analyze our software applications. There are several go packages available for logging.
- log
- logrus
- zap
- zerolog
- glog
1. log
This is the logging package that comes with the standard Go Library. This has very basic logging capabilities. So, this package is good for logging in small applications.
Basic Logging
log.Print("Hello, log!")
log.Println("Hello with newline")
log.Printf("Hello, %s!", "formatted")
Output:
2024/04/28 11:28:37 Hello, log!
2024/04/28 11:28:37 Hello with newline
2024/04/28 11:28:37 Hello, formatted!
Log Prefix and Flags
The log package allows you to set a prefix to each log entry and control the logging format via flags.
log.SetPrefix("prefix: ")
log.SetFlags(log.Ldate | log.Ltime | log.Lshortfile)
log.Println("This is a log with prefix-1")
log.Println("This is a log with prefix-2")
Output:
prefix: 2024/04/28 11:33:17 main.go:33: This is a log with prefix-1
prefix: 2024/04/28 11:33:17 main.go:34: This is a log with prefix-2
Levels of Severity
The log package doesn’t support log levels directly, but you can simulate them using different prefixes or by defining custom logging functions.
func info(args ...interface{}) {
log.SetPrefix("INFO: ")
log.Print(args...)
}
func error(args ...interface{}) {
log.SetPrefix("ERROR: ")
log.Print(args...)
}
Fatal and Panic Logging
log.Fatal* functions will log the message and then call os.Exit(1).
log.Panic* functions will log the message and then panic
log.Fatal("This is a fatal error")
log.Panic("This is a panic")
2. logrus
logrus is a powerful logging package which extends the capabilities of log package. logrus package can be used for logging in more complex applications. Some benefits of logrus are as below.
- Structured Logging: Logrus allows for logging in a structured format, particularly in JSON format.
- Log Levels: Logrus supports multiple logging levels (debug, info, warn, error, fatal, and panic), making it easy to output messages according to their severity.
- Hooks: You can add hooks to send logs to external systems or to trigger actions based on certain log messages.
Basic Logging
logrus.Info("This is an info message")
logrus.Warn("This is a warning message")
logrus.Error("This is an error message")
Output:
INFO[0000] This is an info message
WARN[0000] This is a warning message
ERRO[0000] This is an error message
Structured Logging
Use structured logging to include additional fields in the log entries.
logrus.WithFields(logrus.Fields{
"username": "Sumudu Liyanage",
"id": 123,
}).Info("User logged in")
Output:
INFO[0000] User logged in id=123 username="Sumudu Liyanage"
Customizing Output
Logrus allows you to customize the output format. For instance, to log in JSON format.
logrus.SetFormatter(&logrus.JSONFormatter{})
logrus.WithFields(logrus.Fields{
"username": "Sumudu Liyanage",
"id": 123,
}).Info("User logged in")
Output:
{"id":123,"level":"info","msg":"User logged in","time":"2024-04-28T11:56:35+05:30","username":"Sumudu Liyanage"}
Hooks
You can add hooks to integrate with other systems. For example, adding a hook to send error-level logs to a monitoring service.
3. zap
zap is a high performance logging library. It's designed for applications that require fast, structured logging with minimal overhead. Choosing zap over logrus will be because of the speed.
logger, err := zap.NewProduction()
if err != nil {
panic(err)
}
logger.Info("This is an info message", zap.String("key", "value"), zap.Int("number", 1))
Output:
{"level":"info","ts":1714287036.2461052,"caller":"hello-world/main.go:14","msg":"This is an info message","key":"value","number":1}
4. zerolog
zerolog is another high-performance logging library for Go, specifically designed for zero-allocation JSON logging. zerolog minimizes memory allocations by writing JSON log events directly to output (like io.Writer), avoiding the overhead of temporary objects.
// Set global log level
zerolog.SetGlobalLevel(zerolog.InfoLevel)
// Configure zerolog to write to stdout with human-friendly formatting
log.Logger = log.Output(zerolog.ConsoleWriter{Out: os.Stdout})
// Logging with context
log.Info().Str("role", "myrole").Msg("This is an info message")
log.Debug().Int("id", 123).Msg("This is a debug message that won't be displayed")
log.Error().Msg("This is an error message")
Output:
12:38PM INF This is an info message role=myrole
12:38PM ERR This is an error message
glog
glog offers a robust feature set for handling verbosity-based logging, which is quite useful for detailed control over log output granularity.
Unlike traditional log levels (info, warn, error), glog uses verbosity levels represented by integers. Higher numbers indicate more verbose logs, allowing developers to specify in great detail how much logging should be emitted.
/ Basic logging
glog.Info("This is an informational message")
glog.Warning("This is a warning message")
glog.Error("This is an error message")
// Conditional logging based on verbosity level
if glog.V(2) {
glog.Info("This is a verbose level 2 message")
}
Is fmt also a log package?
The fmt package in Go (Golang) is not primarily a logging package but is used for formatting and printing output. It's part of the Go standard library and provides functions to format strings, numbers, and other data into strings and to print them to standard output (your console) or other outputs.
Thanks for reading!
Top comments (3)
This is cool! I didn't know so many logging libraries existed. I'll definitely try some :)
Just a quick warning, when you write about Levels of Severity, your code sample won't compile.
erroris a type in Go, so compilation will exit with an error. (Kinda like howfunc int()would fail).Thanks for this!
You can enable syntax highlight for your post as follows:
triple-backtick Go
code
triple-backtick
You could have talked about Go slog interface
go.dev/blog/slog