Un guide pour se connecter

1. Vue d'ensemble

Logback est l'un des frameworks de journalisation les plus utilisés dans la communauté Java. C'est un remplacement de son prédécesseur, Log4j. Logback offre une implémentation plus rapide que Log4j, fournit plus d'options de configuration et une plus grande flexibilité dans l'archivage des anciens fichiers journaux.

Cette introduction présentera l'architecture de Logback et vous montrera comment vous pouvez l'utiliser pour améliorer vos applications.

2. Architecture de connexion

Trois classes comprennent l'architecture Logback; Logger , Appender et Layout .

Un enregistreur est un contexte pour les messages de journal. Il s'agit de la classe avec laquelle les applications interagissent pour créer des messages de journal.

Les appenders placent les messages de journal dans leurs destinations finales. Un Logger peut avoir plus d'un Appender. Nous pensons généralement que les Appenders sont attachés à des fichiers texte, mais Logback est beaucoup plus puissant que cela.

La mise en page prépare les messages pour la sortie. Logback prend en charge la création de classes personnalisées pour le formatage des messages, ainsi que des options de configuration robustes pour celles existantes.

3. Configuration

3.1. Dépendance de Maven

Logback utilise la façade de journalisation simple pour Java (SLF4J) comme interface native. Avant de pouvoir commencer à journaliser les messages, nous devons ajouter Logback et Slf4j à notre pom.xml :

 ch.qos.logback logback-core 1.2.3   org.slf4j slf4j-api 1.7.30 test  

Maven Central possède la dernière version de Logback Core et la version la plus récente de slf4j-api .

3.2. Classpath

Logback requiert également logback-classic.jar sur le chemin d'accès aux classes pour l'exécution.

Nous ajouterons ceci à pom.xml en tant que dépendance de test:

 ch.qos.logback logback-classic 1.2.3  

4. Exemple de base et configuration

Commençons par un exemple rapide d'utilisation de Logback dans une application.

Tout d'abord, nous avons besoin d'un fichier de configuration. Nous allons créer un fichier texte nommé logback.xml et le placer quelque part dans notre chemin de classe :

   %d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n      

Ensuite, nous avons besoin d'une classe simple avec une méthode principale :

public class Example { private static final Logger logger = LoggerFactory.getLogger(Example.class); public static void main(String[] args) { logger.info("Example log from {}", Example.class.getSimpleName()); } }

Cette classe crée un enregistreur et appelle info () pour générer un message de journal.

Lorsque nous exécutons Example, nous voyons notre message enregistré sur la console:

20:34:22.136 [main] INFO Example - Example log from Example

Il est facile de comprendre pourquoi Logback est si populaire; nous sommes prêts à fonctionner en quelques minutes.

Cette configuration et ce code nous donnent quelques conseils sur la façon dont cela fonctionne.

  1. Nous avons un appender nommé STDOUT qui fait référence à un nom de classe ConsoleAppender.
  2. Il existe un modèle qui décrit le format de notre message de journal.
  3. Notre code crée un Logger et nous lui avons transmis notre message via une méthode info () .

Maintenant que nous comprenons les bases, regardons de plus près.

5. Contextes de l' enregistreur

5.1. Créer un contexte

Pour enregistrer un message dans Logback, nous avons initialisé un Logger à partir de SLF4J ou Logback:

private static final Logger logger = LoggerFactory.getLogger(Example.class); 

Et puis l'a utilisé:

logger.info("Example log from {}", Example.class.getSimpleName()); 

C'est notre contexte de journalisation. Lorsque nous l'avons créé, nous avons transmis à LoggerFactory notre classe. Cela donne un nom à l' enregistreur (il existe également une surcharge qui accepte une chaîne).

Les contextes de journalisation existent dans une hiérarchie qui ressemble étroitement à la hiérarchie d'objets Java:

  1. Un enregistreur est un ancêtre lorsque son nom, suivi d'un point, préfixe le nom d'un enregistreur descendant
  2. Un enregistreur est un parent lorsqu'il n'y a pas d'ancêtres entre lui et un enfant

Par exemple, la classe Example ci-dessous se trouve dans le package com.baeldung.logback . Il existe une autre classe nommée ExampleAppender dans le package com.baeldung.logback.appenders .

ExampleAppender's Logger est un enfant de Example's Logger.

Tous les enregistreurs sont des descendants de l'enregistreur racine prédéfini.

Un enregistreur a un niveau, qui peut être défini via la configuration ou avec Logger.setLevel (). La définition du niveau dans le code remplace les fichiers de configuration.

Les niveaux possibles sont, par ordre de priorité: TRACE, DEBUG, INFO, WARN et ERROR.Chaque niveau a une méthode correspondante que nous utilisons pour enregistrer un message à ce niveau.

Si un enregistreur n'est pas explicitement attribué à un niveau, il hérite du niveau de son ancêtre le plus proche. L'enregistreur racine utilise par défaut DEBUG. Nous verrons comment contourner cela ci-dessous.

5.2. Utiliser un contexte

Créons un exemple de programme qui illustre l'utilisation d'un contexte dans les hiérarchies de journalisation:

ch.qos.logback.classic.Logger parentLogger = (ch.qos.logback.classic.Logger) LoggerFactory.getLogger("com.baeldung.logback"); parentLogger.setLevel(Level.INFO); Logger childlogger = (ch.qos.logback.classic.Logger)LoggerFactory.getLogger("com.baeldung.logback.tests"); parentLogger.warn("This message is logged because WARN > INFO."); parentLogger.debug("This message is not logged because DEBUG < INFO."); childlogger.info("INFO == INFO"); childlogger.debug("DEBUG < INFO"); 

When we run this, we see these messages:

20:31:29.586 [main] WARN com.baeldung.logback - This message is logged because WARN > INFO. 20:31:29.594 [main] INFO com.baeldung.logback.tests - INFO == INFO

We start by retrieving a Logger named com.baeldung.logback and cast it to a ch.qos.logback.classic.Logger.

A Logback context is needed to set the level in the next statement; note that the SLF4J's abstract logger does not implement setLevel().

We set the level of our context to INFO;we then create another logger named com.baeldung.logback.tests.

We log two messages with each context to demonstrate the hierarchy. Logback logs the WARN, and INFO messages and filters the DEBUGmessages.

Now, let's use the root logger:

ch.qos.logback.classic.Logger logger = (ch.qos.logback.classic.Logger)LoggerFactory.getLogger("com.baeldung.logback"); logger.debug("Hi there!"); Logger rootLogger = (ch.qos.logback.classic.Logger)LoggerFactory.getLogger(org.slf4j.Logger.ROOT_LOGGER_NAME); logger.debug("This message is logged because DEBUG == DEBUG."); rootLogger.setLevel(Level.ERROR); logger.warn("This message is not logged because WARN < ERROR."); logger.error("This is logged."); 

We see these messages when we execute this snippet:

20:44:44.241 [main] DEBUG com.baeldung.logback - Hi there! 20:44:44.243 [main] DEBUG com.baeldung.logback - This message is logged because DEBUG == DEBUG. 20:44:44.243 [main] ERROR com.baeldung.logback - This is logged. 

To conclude, we started with a Logger context and printed a DEBUG message.

We then retrieved the root logger using its statically defined name and set its level to ERROR.

And finally, we demonstrated that Logback actually does filter any statement less than an error.

5.3. Parameterized Messages

Unlike the messages in the sample snippets above, most useful log messages required appending Strings. This entails allocating memory, serializing objects, concatenating Strings, and potentially cleaning up the garbage later.

Consider the following message:

log.debug("Current count is " + count); 

We incur the cost of building the message whether the Logger logs the message or not.

Logback offers an alternative with its parameterized messages:

log.debug("Current count is {}", count); 

The braces {} will accept any Object and uses its toString() method to build a message only after verifying that the log message is required.

Let's try some different parameters:

String message = "This is a String"; Integer zero = 0; try { logger.debug("Logging message: {}", message); logger.debug("Going to divide {} by {}", 42, zero); int result = 42 / zero; } catch (Exception e) { logger.error("Error dividing {} by {} ", 42, zero, e); } 

This snippet yields:

21:32:10.311 [main] DEBUG com.baeldung.logback.LogbackTests - Logging message: This is a String 21:32:10.316 [main] DEBUG com.baeldung.logback.LogbackTests - Going to divide 42 by 0 21:32:10.316 [main] ERROR com.baeldung.logback.LogbackTests - Error dividing 42 by 0 java.lang.ArithmeticException: / by zero at com.baeldung.logback.LogbackTests.givenParameters_ValuesLogged(LogbackTests.java:64) ... 

We see how a String, an int, and an Integer can be passed in as parameters.

Also, when an Exception is passed as the last argument to a logging method, Logback will print the stack trace for us.

6. Detailed Configuration

In the previous examples, we were using the 11-line configuration file to we created in section 4 to print log messages to the console. This is Logback's default behavior; if it cannot find a configuration file, it creates a ConsoleAppender and associates it with the root logger.

6.1. Locating Configuration Information

A configuration file can be placed in the classpath and named either logback.xml or logback-test.xml.

Here's how Logback will attempt to find configuration data:

  1. Search for files named logback-test.xml, logback.groovy,or logback.xml in the classpath, in that order.
  2. If the library does not find those files, it will attempt to use Java's ServiceLoader to locate an implementor of the com.qos.logback.classic.spi.Configurator.
  3. Configure itself to log output directly to the console

Note: the current version of Logback does not support Groovy configuration due to there not being a version of Groovy compatible with Java 9.

6.2. Basic Configuration

Let's take a closer look at our example configuration.

The entire file is in tags.

We see a tag that declares an Appender of type ConsoleAppender, and names it STDOUT. Nested within that tag is an encoder. It has a pattern with what looks like sprintf-style escape codes:

  %d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n  

Last, we see a root tag. This tag sets the root logger to DEBUG mode and associates its output with the Appender named STDOUT:

6.3. Troubleshooting Configuration

Logback configuration files can get complicated, so there are several built-in mechanisms for troubleshooting.

To see debug information as Logback processes the configuration, you can turn on debug logging:

 ... 

Logback will print status information to the console as it processes the configuration:

23:54:23,040 |-INFO in ch.qos.logback.classic.LoggerContext[default] - Found resource [logback-test.xml] at [file:/Users/egoebelbecker/ideaProjects/logback-guide/out/test/resources/logback-test.xml] 23:54:23,230 |-INFO in ch.qos.logback.core.joran.action.AppenderAction - About to instantiate appender of type [ch.qos.logback.core.ConsoleAppender] 23:54:23,236 |-INFO in ch.qos.logback.core.joran.action.AppenderAction - Naming appender as [STDOUT] 23:54:23,247 |-INFO in ch.qos.logback.core.joran.action.NestedComplexPropertyIA - Assuming default type [ch.qos.logback.classic.encoder.PatternLayoutEncoder] for [encoder] property 23:54:23,308 |-INFO in ch.qos.logback.classic.joran.action.RootLoggerAction - Setting level of ROOT logger to DEBUG 23:54:23,309 |-INFO in ch.qos.logback.core.joran.action.AppenderRefAction - Attaching appender named [STDOUT] to Logger[ROOT] 23:54:23,310 |-INFO in ch.qos.logback.classic.joran.action.ConfigurationAction - End of configuration. 23:54:23,313 |-INFO in [email protected] - Registering current configuration as safe fallback point

If warnings or errors are encountered while parsing the configuration file, Logback writes status messages to the console.

There is a second mechanism for printing status information:

  ... 

The StatusListener intercepts status messages and prints them during configuration and while the program is running.

The output from all configuration files is printed, making it useful for locating “rogue” configuration files on the classpath.

6.4. Reloading Configuration Automatically

Reloading logging configuration while an application is running is a powerful troubleshooting tool. Logback makes this possible with the scan parameter:

 ... 

The default behavior is to scan the configuration file for changes every 60 seconds. Modify this interval by adding scanPeriod:

 ... 

We can specify values in milliseconds, seconds, minutes, or hours.

6.5. Modifying Loggers

In our sample file above we set the level of the root logger and associated it with the console Appender.

We can set the level for any logger:

   %d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n        

Let's add this to our classpath and run this code:

Logger foobar = (ch.qos.logback.classic.Logger) LoggerFactory.getLogger("com.baeldung.foobar"); Logger logger = (ch.qos.logback.classic.Logger) LoggerFactory.getLogger("com.baeldung.logback"); Logger testslogger = (ch.qos.logback.classic.Logger) LoggerFactory.getLogger("com.baeldung.logback.tests"); foobar.debug("This is logged from foobar"); logger.debug("This is not logged from logger"); logger.info("This is logged from logger"); testslogger.info("This is not logged from tests"); testslogger.warn("This is logged from tests"); 

We see this output:

00:29:51.787 [main] DEBUG com.baeldung.foobar - This is logged from foobar 00:29:51.789 [main] INFO com.baeldung.logback - This is logged from logger 00:29:51.789 [main] WARN com.baeldung.logback.tests - This is logged from tests 

By not setting the level of our Loggers programmatically, the configuration sets them; com.baeldung.foobar inherits DEBUG from the root logger.

Loggersalso inherit the appender-ref from the root logger. As we'll see below, we can override this.

6.6. Variable Substitution

Logback configuration files support variables. We define variables inside the configuration script or externally. A variable can be specified at any point in a configuration script in place of a value.

For example, here is the configuration for a FileAppender:

  ${LOG_DIR}/tests.log true  %-4relative [%thread] %-5level %logger{35} - %msg%n  

At the top of the configuration, we declared a propertynamed LOG_DIR.Then we used it as part of the path to the file inside the appender definition.

Properties are declared in a tag in configuration scripts. But they are also available from outside sources, such as system properties. We could omit the property declaration in this example and set the value of LOG_DIR on the command line:

$ java -DLOG_DIR=/var/log/application com.baeldung.logback.LogbackTests

We specify the value of the property with ${propertyname}. Logback implements variables as text replacement. Variable substitution can occur at any point in a configuration file where a value can be specified.

7. Appenders

Loggers pass LoggingEvents to Appenders.Appenders do the actual work of logging. We usually think of logging as something that goes to a file or the console, but Logback is capable of much more. Logback-core provides several useful appenders.

7.1. ConsoleAppender

We've seen ConsoleAppenderin action already. Despite its name, ConsoleAppender appends messages to System.outor System.err.

It uses an OutputStreamWriter to buffer the I/O, so directing it to System.err does not result in unbuffered writing.

7.2. FileAppender

FileAppenderappends messages to a file. It supports a broad range of configuration parameters. Let's add a file appender to our basic configuration:

    %d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n    tests.log true  %-4relative [%thread] %-5level %logger{35} - %msg%n          

The FileAppender is configured with a file name via . The tag instructs the Appenderto append to an existing file rather than truncating it. If we run the test several times, we see that the logging output is appended to the same file.

If we re-run our test from above, messages from com.baeldung.logback.tests go to both the console and to a file named tests.log. The descendant logger inherits the root logger's association with the ConsoleAppender with its association with FileAppender. Appenders are cumulative.

We can override this behavior:

Setting additivity to falsedisables the default behavior. Tests will not log to the console, and neither will any of its descendants.

7.3. RollingFileAppender

Oftentimes, appending log messages to the same file is not the behavior we need. We want files to “roll” based on time, log file size, or a combination of both.

For this, we have RollingFileAppender:

  ${LOG_FILE}.log   ${LOG_FILE}.%d{yyyy-MM-dd}.gz  30 3GB   %-4relative [%thread] %-5level %logger{35} - %msg%n   

A RollingFileAppenderhas a RollingPolicy.In this sample configuration, we see a TimeBasedRollingPolicy.

Similar to the FileAppender, we configured this appender with a file name. We declared a property and used it for this because we'll be reusing the file name below.

We define a fileNamePattern inside the RollingPolicy.This pattern defines not just the name of files, but also how often to roll them. TimeBasedRollingPolicyexamines the pattern and rolls at the most finely defined period.

For example:

   ${LOG_DIR}/${LOG_FILE}.log  ${LOG_DIR}/%d{yyyy/MM}/${LOG_FILE}.gz 3GB 

The active log file is /var/logs/application/LogFile.This file rolls over at the beginning of each month into /Current Year/Current Month/LogFile.gzand RollingFileAppender createsa new active file.

When the total size of archived files reaches 3GB, RollingFileAppenderdeletes archives on a first-in-first-out basis.

There are codes for a week, hour, minute, second, and even millisecond. Logback has a reference here.

RollingFileAppenderalso has built-in support for compressing files. It compresses our rolled files because named our them LogFile.gz.

TimeBasedPolicyisn't our only option for rolling files. Logback also offers SizeAndTimeBasedRollingPolicy,which will roll based on current log file size as well as time. It also offers a FixedWindowRollingPolicywhich rolls log file names each time the logger is started.

We can also write our own RollingPolicy.

7.4. Custom Appenders

We can create custom appenders by extending one of Logback's base appender classes. We have a tutorial for creating custom appenders here.

8. Layouts

Layouts format log messages. Like the rest of Logback, Layoutsare extensible and we can create our own. However, the default PatternLayout offers what most applications need and then some.

We've used PatternLayout in all of our examples so far:

 %d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n  

This configuration script contains the configuration for PatternLayoutEncoder.We pass an Encoder to our Appender,and this encoder uses the PatternLayout to format the messages.

The text in the tag defines how log messages are formatting. PatternLayout implements a large variety of conversion words and format modifiers for creating patterns.

Let's break this one down. PatternLayout recognizes conversion words with a %, so the conversions in our pattern generate:

  • %d{HH:mm:ss.SSS} – a timestamp with hours, minutes, seconds and milliseconds
  • [%thread] – the thread name generating the log message, surrounded by square brackets
  • %-5level – the level of the logging event, padded to 5 characters
  • %logger{36} – the name of the logger, truncated to 35 characters
  • %msg%n – the log messages followed by the platform dependent line separator character

So we see messages similar to this:

21:32:10.311 [main] DEBUG com.baeldung.logback.LogbackTests - Logging message: This is a String

An exhaustive list of conversion words and format modifiers can be found here.

9. Conclusion

In this extensive guide, we covered the fundamentals of using Logback in an application.

Nous avons examiné les trois principaux composants de l'architecture de Logback: les enregistreurs, les appendeurs et la mise en page. Logback a de puissants scripts de configuration, que nous avons utilisés pour manipuler des composants pour filtrer et formater les messages. Nous avons également examiné les deux ajouts de fichiers les plus couramment utilisés pour créer, déplacer, organiser et compresser des fichiers journaux.

Comme d'habitude, des extraits de code peuvent être trouvés sur GitHub.