Apache logging services logo Apache log4j logo

Log4j 2 API


Although Log4j 2 provides Logger methods that accept Strings and Objects, all of these are ultimately captured in Message objects that are then associated with the log event. Applications are free to construct Messages of their own and pass them to the Logger. Although it may seem more expensive than passing the message format and parameters directly to the event, testing has shown that with modern JVMs the cost of creating and destroying events is minor, especially when complex tasks are encapsulated in the Message instead of the application. In addition, when using the methods that accept Strings and parameters, the underlying Message object will only be created if any configured global filters or the Logger's log level allow the message to be processed.

Consider an application that has a Map object containing {"Name" = "John Doe", "Address" = "123 Main St.", "Phone" = "(999) 555-1212"} and a User object that has a getId method that returns "jdoe". The developer would like to add an informational message that returns "User John Doe has logged in using id jdoe". The way this could be accomplished is by doing:

logger.info("User {} has logged in using id {}", map.get("Name"), user.getId());

While there is nothing inherently wrong with this, as the complexity of the objects and desired output increases this technique becomes harder to use. As an alternative, using Messages allows:

logger.info(new LoggedInMessage(map, user));

In this alternative the formatting is delegated to the LoggedInMessage object's getFormattedMessage method. Although in this alternative a new object is created, none of the methods on the objects passed to the LoggedInMessage are invoked until the LoggedInMessage is formatted. This is especially useful when an Object's toString method does not produce the information you would like to appear in the log.

Another advantage to Messages is that they simplify writing Layouts. In other logging frameworks the Layout must loop through the parameters individually and determine what to do based on what objects are encountered. With Messages the Layout has the option of delegating the formatting to the Message or performing its formatting based on the type of Message encountered.

Borrowing from the earlier example illustrating Markers to identify SQL statements being logged, Messages can also be leveraged. First, the Message is defined.

public class SQLMessage implements Message {
  public enum SQLType {

  private final SQLType type;
  private final String table;
  private final Map<String, String> cols;

  public SQLMessage(SQLType type, String table) {
      this(type, table, null);

  public SQLMessage(SQLType type, String table, Map<String, String> cols) {
      this.type = type;
      this.table = table;
      this.cols = cols;

  public String getFormattedMessage() {
      switch (type) {
          case UPDATE:
            return createUpdateString();
          case QUERY:
            return createQueryString();

  public String getMessageFormat() {
      return type + " " + table;

  public Object getParameters() {
      return cols;

  private String createUpdateString() {

  private String createQueryString() {

  private String formatCols(Map<String, String> cols) {
      StringBuilder sb = new StringBuilder();
      boolean first = true;
      for (Map.Entry<String, String> entry : cols.entrySet()) {
          if (!first) {
              sb.append(", ");
          first = false;
      return sb.toString();

Next we can use the message in our application.

import org.apache.logging.log4j.Logger;
import org.apache.logging.log4j.LogManager;
import java.util.Map;

public class MyApp {

    private Logger logger = LogManager.getLogger(MyApp.class.getName());
    private static final Marker SQL_MARKER = MarkerManager.getMarker("SQL");
    private static final Marker UPDATE_MARKER = MarkerManager.getMarker("SQL_UPDATE", SQL_MARKER);
    private static final Marker QUERY_MARKER = MarkerManager.getMarker("SQL_QUERY", SQL_MARKER);

    public String doQuery(String table) {

        logger.debug(QUERY_MARKER, new SQLMessage(SQLMessage.SQLType.QUERY, table));

        return logger.exit();

    public String doUpdate(String table, Map<String, String> params) {

        logger.debug(UPDATE_MARKER, new SQLMessage(SQLMessage.SQLType.UPDATE, table, parmas);

        return logger.exit();

Notice that in contrast to the prior version of this example, the logger.debug in doUpdate no longer needs to be wrapped in an isDebugEnabled call as creation of the SQLMessage is on the same order of magnitude of performing that check. Furthermore, all the formatting of the SQL columns is now hidden in the SQLMessage instead of having to take place in the business logic. Finally, if desired, Filters and/or Layouts can be written to take special action when an SQLMessage is encountered.


The message pattern passed to a FormattedMessage is first checked to see if it is a valid java.text.MessageFormat pattern. If it is, a MessageFormatMessage is used to format it. If not it is next checked to see if it contains any tokens that are valid format specifiers for String.format(). If so, a StringFormattedMessage is used to format it. Finally, if the pattern doesn't match either of those then a ParameterizedMessage is used to format it.


LocalizedMessage is provided primarily to provide compatibility with Log4j 1.x. Generally, the best approach to localization is to have the client UI render the events in the client's locale.

LocalizedMessage incorporates a ResourceBundle and allows the message pattern parameter to be the key to the message pattern in the bundle. If no bundle is specified, LocalizedMessage will attempt to locate a bundle with the name of the Logger used to log the event. The message retrieved from the bundle will be formatted using a FormattedMessage.


LoggerNameAwareMessage is an interface with a setLoggerName method. This method will be called during event construction so that the Message has the name of the Logger used to log the event when the message is being formatted.


A MapMessage contains a Map of String keys and values. MapMessage implements FormattedMessage and accepts a format specifier of "XML", in which case the Map will be formatted as XML. Otherwise, the Map will be formatted as "key1=value1 key2=value2...".


MessageFormatMessage handles messages that use a conversion format. While this Message has more flexibility than ParameterizedMessage, it is also about two times slower.


A MultiformatMessage will have a getFormats method and a getFormattedMessage method that accepts and array of format Strings. The getFormats method may be called by a Layout to provide it information on what formatting options the Message supports. The Layout may then call getFormattedMessage with one or more for the formats. If the Message doesn't recognize the format name it will simply format the data using its default format. An example of this is the StructuredDataMessage which accepts a format String of "XML" which will cause it to format the event data as XML instead of the RFC 5424 format.


Formats an Object by calling its toString method.


ParameterizedMessage handles messages that contain "{}" in the format to represent replaceable tokens and the replacement parameters.


SimpleMessage contains a String that requires no formatting.


StringFormattedMessage handles messages that use a conversion format that is compliant with java.lang.String.format(). While this Message has more flexibility than ParameterizedMessage, it is also 5 to 10 times slower.


StructuredDataMessage allows applications to add items to a Map as well as set the id to allow a message to be formatted as a Structured Data element in accordance with RFC 5424.


A ThreadDumpMessage, if logged, will generate stack traces for all threads. If running on Java 6+ the stack traces will include any locks that are held.


A TimestampMessage will provide a getTimestamp method that is called during event construction. The timestamp in the Message will be used in lieu of the current timestamp.