log4js-node / log4js-node Goto Github PK

See this gist. https://gist.github.com/44b39d45e2727d1a31ad

I have seen instances where log messages from different loggers are output in the wrong order:

[2011-10-20 12:02:20.694] [DEBUG] loggerone - log message 1
[2011-10-20 12:02:20.696] [DEBUG] loggertwo - log message 1
[2011-10-20 12:02:20.694] [DEBUG] loggerone - log message 2
[2011-10-20 12:02:20.697] [DEBUG] loggertwo - log message 2

Calling configure seems to disables logging of all messages. Take this example:

var log4js = require('log4js');

log4js.configure({
appenders: [
{ type: 'console', category:'tests' }
],
"levels": {
"tests": "INFO"
}
});
var log = log4js.getLogger('test');
log.warn('This message is not displayed.');

Now you have to define logger like this;

log4js.addAppender(log4js.consoleAppender());

Instead of the old (and less congruent)

log4js.addAppender(consoleAppender());

Thanks!

It's more typical for console logs to go to stderr than stdout, to avoid interfering with "real" output from (for example) a command-line tool.

Simple test case:

log4js = require ('log4js');

logger = log4js.getLogger();

logger.setLevel ('INFO');
log4js.restoreConsole();


logger.info("This line causes exception");

causes the following exception at the logger.info() line:

node.js:201
throw e; // process.nextTick error, or 'error' event on first tick
^
TypeError: Object # has no method '_preLog4js_log'
at Logger. (/Users/nils/Documents/MobileSpan/Devel/git/bin/deploy/node_modules/log4js/lib/appenders/console.js:6:10)
at Logger.emit (events.js:67:17)
at Logger.log (/Users/nils/Documents/MobileSpan/Devel/git/bin/deploy/node_modules/log4js/lib/log4js.js:215:10)
at Logger.info (/Users/nils/Documents/MobileSpan/Devel/git/bin/deploy/node_modules/log4js/lib/log4js.js:233:38)
at Object. (/Users/nils/Documents/MobileSpan/Devel/git/bin/deploy/foo.js:10:8)
at Module._compile (module.js:432:26)
at Object..js (module.js:450:10)
at Module.load (module.js:351:31)
at Function._load (module.js:310:12)
at Array.0 (module.js:470:10)

Am I doing something wrong here?

console.log allows for any number of arguments, e.g. console.log(a,b,c)

Log4js does not. It interprets the second parameter as an exception, and ignores any subsequent.

This isn't a big deal, but log4js replaces the standard console methods, thus breaking any script that uses the standard console.log.

See https://github.com/joyent/node/blob/master/lib/console.js#L60

I'm trying to get the backup files working for log4js using a json config file. It seems to create the backup files, but does not write to them, it only writes to the main log file

My log4js config file: https://gist.github.com/850098#

I created a wrapper module: https://gist.github.com/850104#

And then used the abstraction in other modules: https://gist.github.com/850114#

Can you spot what I'm doing wrong here?

Similar to the way log4j searches Java's classpath for log4j.properties, log4js should look for a log4js.json file when it first starts up.

Here's a quick patch: https://gist.github.com/880747

Basically just lets you specify "layout": {"type": "colored"} in your configuration for nice colorized output.

I did some tests to see how efficient the regex replace method is to format strings (https://github.com/nomiddlename/log4js-node/blob/master/lib/layouts.js#L31).

According to the tests I did simple string concatenation is much faster that regex replace method - around 3.5 times.

Here are the results:

stringAdd time: 36
stringAdd memory: { rss: 9277440, heapTotal: 3910080, heapUsed: 1682956 }

regex time: 117
regex memory: { rss: 9211904, heapTotal: 3926400, heapUsed: 1914876 }

var format = "%s %s some data string to: %s";
var a = "adfkjafgksjgfksjsgkjsf;gjsgfjs;fdg";
var b = "jadfjadfhjadfkjadfkjakdfjadkfjakdfjakfj";
var c = 5;

var NTIMES = 100000;

// use each test seperately - to test memory.

regexTest();
// arrayJoinTest();
// stringAddTest();

function regexTest() {
    var data = [a, b, c];
    var s = new Date();
    for (var i = 0; i < NTIMES; i++) {
        var k = 0;
        var str = format.replace(/%[sdj]/g, function (match) {
            switch (match) {
            case "%s": return new String(data[k++]);
            case "%d": return new Number(data[k++]);
            case "%j": return JSON.stringify(data[k++]);
            default:
                return match;
            }
        });
    }
    var e = new Date();
    console.log("regex time: %s", e - s);
    console.log("regex memory: %s", require("util").inspect(process.memoryUsage()));
}

function stringAddTest() {
    var data = [a, b, c];
    var s = new Date();
    for (var i = 0; i < NTIMES; i++) {
        f = format.split("%s");
        str = "";
        for (var j = 0, len = f.length - 1; j < len; j++) {
            str += f[j];
            str += data[j];
        }
        str += f[f.length - 1];
    }
    var e = new Date();
    console.log("stringAdd time: %s", e - s);
    console.log("stringAdd memory: %s", require("util").inspect(process.memoryUsage()));
}

function arrayJoinTest() {
    var data = [a, b, c];
    var s = new Date();
    var arr = [ ];
    for (var i = 0; i < NTIMES; i++) {
        f = format.split("%s");
        str = "";
        for (var j = 0, len = f.length - 1; j < len; j++) {
            arr.push(f[j]);
            arr.push(data[j]);
        }
        arr.push(f[f.length - 1]);
        arr.join("");
    }
    var e = new Date();
    console.log("arrayJoin time: %s", e - s);
    console.log("arrayJoin memory: %s", require("util").inspect(process.memoryUsage()));
}

I do see there is a provision for log rolling based on file size which is great. I wanted to check if there currently is or planned release for rolling based on time (yyyy-MM-dd-HH - eg from log4j) so that we could have daily or hourly log files adhering to a pattern that we specify.

I pass a string parameter to isLevelEnabled and it throw an exception:

If I pass logger.levels.DEBUG to isLevelEnabled, it will work well.

DEBUG: current size = 12107
DEBUG: current size = 12170
DEBUG: current size = 12321
DEBUG: current size = 12381
...
...

gets printed on every log message, when I configure a file appender with rolling logs. E.g: I used a JSON like below to pass to log4js.configure();

logAppenderJSON : {
"appenders": [
{
"type": "file",
"absolute": true,
"filename": "/tmp/my-server.log",
"maxLogSize": 100000,
"backups": 5,
"pollInterval" : 15,
"category": "my-server"
}
],
"levels": {
"fad-server": "TRACE"
}
}

I can't understand why u want the logfiles' fileName sort by the most updated index,like xxx.log.1,xxx.log.2.
when u do this:
1、the bak file's name will be updated again if a new file occured,it's waste of time
2、the fileName were always in updating so i can't get a copy of bak file which just the version i want! for example, i want the .log.4 but the file was always changing its content...

Please list "vows" as a devDependency in your package.json so that the npat install will pick it up and the test can be run.

Getting the following error when tried npm install log4js.

[email protected] install /Users/satheesh/Work/LearnNode/node_modules/log4js/node_modules/compress-buffer
make build

sh: make: command not found
npm ERR! error installing [email protected] Error: [email protected] install: make build
npm ERR! error installing [email protected] sh "-c" "make build" failed with 127
npm ERR! error installing [email protected] at ChildProcess. (/usr/local/lib/node_modules/npm/lib/utils/exec.js:49:20)
npm ERR! error installing [email protected] at ChildProcess.emit (events.js:70:17)
npm ERR! error installing [email protected] at maybeExit (child_process.js:359:16)
npm ERR! error installing [email protected] at Process.onexit (child_process.js:395:5)
npm ERR! error installing [email protected] Error: [email protected] install: make build
npm ERR! error installing [email protected] sh "-c" "make build" failed with 127
npm ERR! error installing [email protected] at ChildProcess. (/usr/local/lib/node_modules/npm/lib/utils/exec.js:49:20)
npm ERR! error installing [email protected] at ChildProcess.emit (events.js:70:17)
npm ERR! error installing [email protected] at maybeExit (child_process.js:359:16)
npm ERR! error installing [email protected] at Process.onexit (child_process.js:395:5)
npm ERR! error installing [email protected] Error: ENOENT, no such file or directory '/Users/satheesh/Work/LearnNode/node_modules/log4js/node_modules/___async.npm/package/.gitmodules'
npm ERR! [email protected] install: make build
npm ERR! sh "-c" "make build" failed with 127
npm ERR!
npm ERR! Failed at the [email protected] install script.
npm ERR! This is most likely a problem with the compress-buffer package,
npm ERR! not with npm itself.
npm ERR! Tell the author that this fails on your system:
npm ERR! make build
npm ERR! You can get their info via:
npm ERR! npm owner ls compress-buffer
npm ERR! There is likely additional logging output above.
npm ERR!
npm ERR! System Darwin 11.2.0
npm ERR! command "node" "/usr/local/bin/npm" "install" "log4js"
npm ERR! cwd /Users/satheesh/Work/LearnNode
npm ERR! node -v v0.6.3
npm ERR! npm -v 1.1.0-alpha-2
npm ERR! code ELIFECYCLE
npm ERR! message [email protected] install: make build
npm ERR! message sh "-c" "make build" failed with 127
npm ERR!
npm ERR! Additional logging details can be found in:
npm ERR! /Users/satheesh/Work/LearnNode/npm-debug.log
npm not ok

On startup in node > v0.4.x you get the following message on stdout:
The "sys" module is now called "util". It should have a similar interface.

File logtest.js:

var log4js = require('log4js');
var logger = log4js.getLogger('test');

log4js.configure('l4js.json', {reloadSecs: 3});

setInterval(function() { logger.info('Info test'); logger.debug('Debug test'); }, 1000);

File l4js.json:

{
"levels": {
        "test": "INFO"
}
}

Result:
[2011-08-29 05:45:35.039] [INFO] test - Info test
[2011-08-29 05:45:36.039] [INFO] test - Info test
[2011-08-29 05:45:37.039] [INFO] test - Info test
[2011-08-29 05:45:38.039] [INFO] test - Info test
[2011-08-29 05:45:39.039] [INFO] test - Info test
[2011-08-29 05:45:40.039] [INFO] test - Info test
[2011-08-29 05:45:40.040] [DEBUG] test - Debug test
[2011-08-29 05:45:41.039] [INFO] test - Info test
[2011-08-29 05:45:41.039] [DEBUG] test - Debug test
[2011-08-29 05:45:42.039] [INFO] test - Info test

I put a call in rollThatLog at the top to dump the logFile before "rolling":

[2011-07-21 18:50:21.560] [INFO] console - {"path":"/usr/local/hapa/logs/event_server/event_log.log","fd":6,"writable":true,"flags":"a","encoding":"utf8","mode":420,"busy":false,"_queue":[],"drainable":true}

Now after roll that log, notice the new log fd is null:

[2011-07-21 18:50:21.561] [INFO] console - {"path":"/usr/local/hapa/logs/event_server/event_log.log","fd":null,"writable":true,"flags":"a","encoding":"utf8","mode":420,"busy":true,"_queue":[]}

After this rotation, all subsequent logs still are going to the original log file, which is named .1 ....

It wouls be nice to have an option to configure log4js to not overwrite console.

In log4js-node, is there a way to automatically show the file and line number that logs a message?

I found

http://code.google.com/p/aost/source/browse/trunk/tools/firefox-plugin/trump/chrome/content/logger.js?r=858 - should I do something custom like this?

Sorry, I did look through existing tickets and documentation but my search came up no hits. It just seems like an obvious thing to want.

Best, Martin.

log4js 0.4.2 hasn't been published yet.
There is a fix in 0.4.2 which I need. Can you publish the new version to NPM?

Thanks,
Samet

I just started using log4js-node. My projects use console.log and a lot of those calls have multiple arguments like

console.log(obj1, obj2, 'string');

log4js concats the arguments without spacers, console.log added spaces.

The behaviour should be that log4js adds spaces.

Please can you change "web" to "url" in /log4js/package.json
see https://github.com/Pita/etherpad-lite/issues/188#issuecomment-2537708

• "web": "http://github.com/csausdev/log4js-node/issues"
• "url": "http://github.com/csausdev/log4js-node/issues"

i followed the example ,and with this 2 line code

logger.error('Cheese is too ripe!');
logger.fatal('Cheese was breeding ground for listeria.');

i have a .log file in my disk,but i found the 2 message are on one line,not have a new line on the second log item:

[the .log file:]
"
[2012-06-28 14:24:41.101] [ERROR] cheese - Cheese is too ripe![2012-06-28 14:24:41.108] [FATAL] cheese - Cheese was breeding ground for listeria.

"

the examples provided do not work.

The only way to get a file appender is through JSON.

Is the npmversion in sync with the version in github?

I installed log4js:

ryppi install log4js

I've created my custom config:

log4js.json:

{
"appenders": [
{
"type": "console",
"layout": {
"type": "basic"
}
},
{
"type": "file",
"filename": "log_file.log",
"maxLogSize": 1024,
"backups": 3,
"pollInterval": 15
}
]
}

My node version:

node -v
v0.5.10

When I run application that does:

log4js.configure(__dirname + '/log4js.json', {
"reloadSecs": 10
});

It gives me:

Error: Problem reading log4js config ... . Error was "use fs.watch api instead".

In some earlier version of node was giving something about 'undefined is not a function' within fs.watchFile.

I run Node on Window 7 (both Cygwin built and windows binary give same error).

node tests.js does not execute any tests, but does not fail either.

Found this error:
[2011-10-18 02:55:04.590] [ERROR] console - Error: ENOENT, No such file or directory '/foo/bar/node1/logs/foobar.log.9'
at Object.renameSync (fs.js:320:18)
at rollThatLog (/foo/bar/node_modules/log4js/lib/appenders/file.js:40:24)
at StatWatcher. (/foo/bar/node_modules/log4js/lib/appenders/file.js:29:21)

The interesting thing is that the error occured on foobar.log.4, but it's complaining about foobar.log.9, which exists btw, so I'm not sure why it's complaining about 'No such file or directory'.

The files at /foo/bar/node1/logs/:
-rw-r--r-- 1 octocat octocat 0 Oct 18 02:55 foobar.log
-rw-r--r-- 1 octocat octocat 0 Oct 18 02:55 foobar.log.1
-rw-r--r-- 1 octocat octocat 0 Oct 9 09:18 foobar.log.10
-rw-r--r-- 1 octocat octocat 0 Oct 18 02:55 foobar.log.2
-rw-r--r-- 1 octocat octocat 104858178 Oct 18 02:55 foobar.log.4
-rw-r--r-- 1 octocat octocat 121278872 Oct 17 12:47 foobar.log.5
-rw-r--r-- 1 octocat octocat 104858259 Oct 14 18:39 foobar.log.6
-rw-r--r-- 1 octocat octocat 403161013 Oct 14 12:15 foobar.log.8
-rw-r--r-- 1 octocat octocat 109284476 Oct 11 16:59 foobar.log.9

log4js.json:
{
"appenders": [{
"type": "logLevelFilter",
"level": "DEBUG",
"appender": {
"type": "file",
"filename": "/foo/bar/node1/logs/foobar.log",
"maxLogSize": 104857600,
"backups": 10,
"pollInterval": 15
}
}]
}

The file logger was stopping after about four messages.

Workaround was to bypass the queued logging.

Guess is that either Node or Windows is suppressing the drain events - the code as written implies that there will always be a drain event after a write, and after first open.

With more time, I'd have provided a replacement with a timer.

Travis-ci reports that the build fails once the GELF support test was merged in. Looks like it was taking too long.
See:
http://travis-ci.org/#!/csausdev/log4js-node/builds/514258

Looks like the test doesn't handle timeouts, or the log message never gets sent.

Hi,

I'm using etherpad-lite (https://github.com/Pita/etherpad-lite) which uses node.js and log4js-node. Due to some problem with socket code, I got a very large logfile and I realized that it was time to set up log rotation. I started to look into how this was done with log4js and I found that there is built-in logrolling, I just need to configure it.

Great. So since the application isn't configuring this anywhere, I followed the directions and created a log4js.json file with the following content straight from the example:

{
  "appenders": [
    {
      "type": "file",
      "filename": "/var/log/etherpad-lite/etherpad-lite.log",
      "maxLogSize": 1024,
      "backups": 3,
      "pollInterval": 15
    }
  ]
}

and then I puzzled over where I was supposed to put it. The documentation says the CWD is the first place looked, so I tried to figure out where exactly the CWD was. I decided the most logical place was the CWD of the node process, so I put it there (based on the value from /proc//cwd). It didn't seem to fire there. My current log file is larger than 1024 (by the way, what unit is that? bytes? megabytes?), so I expected it to rotate, but it didn't do anything. I plopped the log4js.json in various other locations that made less and less sense, including until I gave up.

Am I misunderstanding how this should work or is there a bug here?

thanks!

Hi,

I thought to work on something that would look like this:

Synopsis

consider your unit-test to contain a test like this:

test("my operation emits logs properly", function(done) {

  var config = 
      { category: "my-operation"
      , maxEntries : 200   //keeps in memory maximum 200 entries per category 
      };

      testAppender = log4js.addAppender( log4js.unitestsAppender(config), "my-operation" ),
      myAutits = testAppender.getAudits();

  //this operation is assumed to send an entry to getLogger("my-operation")
  operationThatShouldEmitLogs(function(err, result){
      log4js.removeAppender( testAppender);

      assert.ok( 
        myAutits.contains(
          { level    : /INFO|WARN/
          , text     : /(Operation succeeded|Operation failed)/
          , category : "my-operation"
          }
        )
      );
});

or

var vows = require('vows')
   , claim = require('assert')
   ;

vows.describe(
   "my great utility"
).addBatch(
   { "my operation" :
     { topic: 
        function(){
           var cb = this.callback;
           var config = 
               { category: "my-operation"
               , maxEntries : 200   //keeps in memory maximum 200 entries per category 
               };

           testAppender = log4js.addAppender( log4js.unitestsAppender(config), "my-operation" ),
           myAutits = testAppender.getAudits();

           //this operation is assumed to send an entry to getLogger("my-operation")
           operationThatShouldEmitLogs( function(err,result) {
                  log4js.removeAppender( testAppender);
                  cb(err,result);
            });     
        }
     , "should emit loggs properly" :
        function(err, audit){
           assert.isUndefined(err);
           assert.isObject(audit);
           assert.ok( 
             myAutits.contains(
               { level    : /INFO|WARN/
               , text     : /(Operation succeeded|Operation failed)/
               , category : "my-operation"
               }
             )
           );
        }
     }
  }
);

API

I'm not final on the API design, and would love to talk about it with somebody
what I got now - and I'm not sure if its fully compatible with the underlying infra (log4js) is:

• unitestsAppender
  -- implements Logger
  -- getAudits(0) - returns unitTestsAudits module
• unitTestsAudits
  -- contains(query), where query can optionally specify level, text and category
  (why category? in case you audit few together)

Technicalities

I looked on ./lib/appenders/console - it looks quite easy to add an appender that will gather the entries in memory, so they can later be searched for assertions.
My real problems are -

• technically how to do it with GIT and GITHUB...
• I can't run the unit-tests - perhaps because I'm a windows user and the npm install fiddles with dependencies that are not cross-platform.

GITHUB

I'm still trying to take my first successful steps with github, with cloning a repo, working on a solution and sumbitting a pull request that is not editing a file on the github web editor...

Windows

I can offer help in making component cross-platform, the minuet we figure how to run it on windows, or how to ignore it if it's just wont...

I'd appreciate some help here... :)

Running example-connect-logger.js from HEAD results in exception:

TypeError: Object #<Object> has no method 'isLevelEnabled'
    at Object.connectLogger (.../log4js/lib/connect-logger.js:47:17)
    at HTTPServer.<anonymous> (.../log4js/example-connect-logger.js:9:20)
    at HTTPServer.configure (.../express/lib/http.js:484:52)
    at Object.<anonymous> (.../log4js/example-connect-logger.js:8:5)
    at Module._compile (module.js:423:26)
    at Object..js (module.js:429:10)
    at Module.load (module.js:339:31)
    at Function._load (module.js:298:12)
    at Array.<anonymous> (module.js:442:10)
    at EventEmitter._tickCallback (node.js:175:26)

This is because require("connect-logger").getLogger() awaits a Logger instance as a first argument, but receives log4js module instance (with this).

When fixing this (for example, by removing "(this)" part in logj4s.js), it starts to work, but produces misformatted results like this:

[2011-07-18 20:08:14.445] [INFO] cheese - 1
'2''7''.''0''.''0''.''1'' ''-'' ''-'' ''"''G''E''T'' ''/'' ''H''T''T''P''/''1''.''1''"'' ''2''0''0'' ''1''1'' ''"''u''n''d''e''f''i''n''e''d''"'' ''"''l''w''p''-''r''e''q''u''e''s''t''/''5''.''8''3''4'' ''l''i''b''w''w''w''-''p''e''r''l''/''5''.''8''3''7''"'

This is due to calls like thislogger.log(level, format(fmt, req, res)), but Logger.prototype.log expecting an array. I.e. either thislogger.log should be called with [format(...)] array, or Logger.prototype.log (and level-named shortcuts) should be modified to accept variadic arguments.

It would be great to have an SMTP appender so I can get alerts for fatals. Of course, that means endorsing a particular email library.

Which prevents it from running under node 0.5.x:

node.js:198
        throw e; // process.nextTick error, or 'error' event on first tick
              ^
Error: require.paths is removed. Use node_modules folders, or the NODE_PATH environment variable instead.
    at Function.<anonymous> (module.js:360:11)
    at findConfiguration (/Users/mark/work/foo/node_modules/log4js/lib/log4js.js:253:37)
    at Object.<anonymous> (/Users/mark/work/foo/node_modules/log4js/lib/log4js.js:387:11)
    at Module._compile (module.js:416:26)
    at Object..js (module.js:422:10)
    at Module.load (module.js:335:31)
    at Function._load (module.js:294:12)
    at Module.require (module.js:341:17)
    at require (module.js:352:17)
    at Object.<anonymous> (/Users/mark/work/foo/main.js:9:14)

So i have multiple files in my project. Let's say I have 2 files.

In one file I require log4js and then configure it using a file. Now, If I require log4js in another file, I lose all the configuration. This kinda breaks node's module system. I need to reconfigure it. However, as I do that I somehow lose my console configs. I like to output console to file as well and without colors but it always comes up with colors.

Any ideas? Thanks!!

So I have custom config file log4js-config.json

{
"appenders": [
{
"type": "console",
"layout": {
"type": "basic"
}
},
{
"type": "file",
"filename": "log/application_server.log"
}
]
}

This line:

log4js.configure(__dirname + '/log4js-config.json', {
"reloadSecs": 10
});

Will reload my config on any change but also each 10 seconds reload the default config (overwriting my settings - basic layout instead of colorized one (btw it doesn't work on Windows so you might consider some coloring library))

Could you please add this to the npm registry?

http://github.com/isaacs/npm
http://github.com/isaacs/npm/blob/master/doc/publish.md

I would like to be able to define the level of logging on a per appender basis. Would it be possible to do something like:

var log4js = require('./log4js-node/lib/log4js-node');
log4js.addAppender(consoleAppender('INFO'));
log4js.addAppender(fileAppender('ERROR', 'logs/my.log'), 'my');
var logger = log4js.getLogger('my');

In my mind that would put all log messages below INFO and the log file would have all messages below ERROR. Is that possible now? Is that something that could be added?

I have a lot categories, such as a, b, c, d, e... I want to configure the console appender to log all the categories but category c and e.
I check the source code and seems no exclude operation. I have to specify appender to log for category a, b, d, ... one by one to achieve it.
It would be better if I can exclude or specify categories for an appender by regular expression:)

I have a use case in which I would like log statements to go to a file, but use of console.log() to still go to the screen. For instance, when using nodeunit, I would like the results of each test to still echo to the screen, but all the minutiae of the tests to be logged to a file. Unfortunately, I'm unable to configure log4js such that these aren't commingled. I've tried clearAppenders() like so:

log4js.clearAppenders();
log4js.addAppender(log4js.fileAppender( '/my/fie.log'), 'main');

bu then I lose my nodeunit status messages entirely. How does one suppress consoleAppender?

Most of the time it would be convenient if you could control what relative paths in configuration file are resolved to. The most intuitive would be to resolve all the relative paths against the config file location. It could be easily done by adding something like basePath config property that is intialized to a loaded file name and than all the rest work with config requires to do path resolution against that path.

Roll the log over when the day changes... for example, at 12:00 AM, roll the log file :-)

At each log file roll time, an fs.statSync call is made to check for the existence of each possible log file, based on the value of the "backups" config parameter in log4js.json. So, e.g., if "backups" is set to 100000, then 100K statSync calls are made, even if only a few log files actually exist. With high "backups" settings, this can cause a noticeable delay when log file rolling is triggered.

I think it would be better to check what files actually exist on the filesystem, via fs.readdirSync, for example, then do the rename operation on those.

Hi, I am trying to use log4js on node 0.5.10. When I use a file appender with backup, it bails out with:

Error: Problem reading log4js config { appenders:
   [ { type: 'file',
       filename: 'node-consumer.log',
       maxLogSize: 1024,
       makers: [Object] } ] }. Error was "use fs.watch api instead" (Error: use fs.watch api instead
    at Object.watchFile (fs.js:696:11)

If you have reload interval for config file the appenders will not be reassigned to new loggers and thus output is surpressed.

Hi ,

xml2js configure fails when used with the load.
I created a patch, I am happy to help people change.
https://gist.github.com/854019

※ xml2js is, proto.js in String.prototype.forEach created.