January 2010 - logback-dev - mailman3.qos.ch

[JIRA] Created: (LBCORE-131) SizeAndTimeBasedFNATP will overwrite previous logging files when File property is set at RollingFileAppender
by tomliliu (JIRA) 30 Jun '10

30 Jun '10

SizeAndTimeBasedFNATP will overwrite previous logging files when File property is set at RollingFileAppender ------------------------------------------------------------------------------------------------------------ Key: LBCORE-131 URL: http://jira.qos.ch/browse/LBCORE-131 Project: logback-core Issue Type: Bug Components: Rolling Affects Versions: 0.9.18 Reporter: tomliliu Assignee: Logback dev list Here's the failure case: Configuration: <appender name="appender" class="ch.qos.logback.core.rolling.RollingFileAppender"> <File>c:/var/tmp/base.log</File> <rollingPolicy class="ch.qos.logback.core.rolling.TimeBasedRollingPolicy"> <FileNamePattern> c:/var/tmp/%d{yyyy-MM-dd_HH}.%i.log </FileNamePattern> <TimeBasedFileNamingAndTriggeringPolicy class="ch.qos.logback.core.rolling.SizeAndTimeBasedFNATP"> <MaxFileSize>1KB</MaxFileSize> </TimeBasedFileNamingAndTriggeringPolicy> </rollingPolicy> <layout class="ch.qos.logback.classic.PatternLayout"> <Pattern> %d{HH:mm:ss.SSS} [%thread] %-5level %logger{36} - %msg%n </Pattern> </layout> </appender> Steps to reproduce: 1. Start the application 2. base.log was created. 3. yyyy-MM-dd_HH.0.log was created on rollover 4. rebounce the application 5. log to base.log 6. yyyy-MM-dd_HH.0.log was created on rollover, overwrite the previous yyyy-MM-dd_HH.0.log. Logging messages are lost. It looks like the root cause is that SizeAndtimeBasedFNATP does not update the currentPeriodsCounter when parentsRawFileProperty is set. if (tbrp.getParentsRawFileProperty() == null) { String regex = tbrp.fileNamePattern.toRegex(dateInCurrentPeriod); String stemRegex = FileFilterUtil.afterLastSlash(regex); computeCurrentPeriodsHighestCounterValue(stemRegex); } SizeAndTimeBasedFNATP should update currentPeriodsCounter regardless whether parentsRawFileProperty is set or not. Here's a fix works for me. SizeAndTimeBasedFNATP.java @Override public void start() { // we depend on certain fields having been initialized // in super.start() super.start(); archiveRemover = new SizeAndTimeBasedArchiveRemover(tbrp.fileNamePattern, rc); archiveRemover.setContext(context); // we need to get the correct value of currentPeriodsCounter. // usually the value is 0, unless the appender or the application // is stopped and restarted within the same period //if (tbrp.getParentsRawFileProperty() == null) { String regex = tbrp.fileNamePattern.toRegex(dateInCurrentPeriod); String stemRegex = FileFilterUtil.afterLastSlash(regex); computeCurrentPeriodsHighestCounterValue(stemRegex); //} started = true; } void computeCurrentPeriodsHighestCounterValue(final String stemRegex) { File file = new File(getCurrentPeriodsFileNameWithoutCompressionSuffix()); File parentDir = file.getParentFile(); File[] matchingFileArray = FileFilterUtil .filesInFolderMatchingStemRegex(parentDir, stemRegex); if (matchingFileArray == null || matchingFileArray.length == 0) { return; } FileFilterUtil.reverseSortFileArrayByName(matchingFileArray); currentPeriodsCounter = FileFilterUtil.extractCounter(matchingFileArray[0], stemRegex); //If parentsRawFileProperty is set, we should increment currentPeriodsCounter by one in order to avoid overwrite the last archive file. if(tbrp.getParentsRawFileProperty() != null) { currentPeriodsCounter ++; } } Thanks, Tom -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

2 1

[JIRA] Created: (LBCLASSIC-167) RollingFileAppender does not auto create subdirectories in some circumstances
by Rick Janda (JIRA) 08 May '10

08 May '10

RollingFileAppender does not auto create subdirectories in some circumstances ----------------------------------------------------------------------------- Key: LBCLASSIC-167 URL: http://jira.qos.ch/browse/LBCLASSIC-167 Project: logback-classic Issue Type: Bug Components: appender Affects Versions: 0.917 Reporter: Rick Janda Assignee: Logback dev list If the FileNamePattern of the rollingPolicy contains another directory than the File element of the appender, than the directory in the FileNamePattern is not created automatically. The OnConsoleErrorListener writes out the following error in this case: -ERROR in c.q.l.core.rolling.helper.Compress - Error occurred while compressing [/opt/baa-logserver/logs/logback.log33794048136194907.tmp] into [/opt/baa-logserver/logs/archive/logback_2009-10-26.log.zip]. java.io.FileNotFoundException: /opt/baa-logserver/logs/archive/logback_2009-10-26.log.zip (No such file or directory) at java.io.FileNotFoundException: /opt/baa-logserver/logs/archive/logback_2009-10-26.log.zip (No such file or directory) The appender is configured as followed (files are rolled into a archive directory on daily basis): <appender name="LOGBACK" class="ch.qos.logback.core.rolling.RollingFileAppender"> <File>${LOG_DIR}/logback.log</File> <rollingPolicy class="ch.qos.logback.core.rolling.TimeBasedRollingPolicy"> <FileNamePattern>${LOG_DIR}/archive/logback_%d{yyyy-MM-dd}.log.zip</FileNamePattern> <MaxHistory>${MAX_HISTORY_DAYS}</MaxHistory> </rollingPolicy> <layout class="ch.qos.logback.classic.PatternLayout"> <Pattern>%date{dd-MM-yyyy HH:mm:ss.SSS} [%contextName] $%mdc{refId}$ %-5level %logger{0} - %msg%n%xEx{full}</Pattern> </layout> </appender> -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

3 3

[JIRA] Created: (LBGENERAL-28) JMX configurator should be able to use MBeanServers with a domain name
by C Z (JIRA) 09 Apr '10

09 Apr '10

JMX configurator should be able to use MBeanServers with a domain name ---------------------------------------------------------------------- Key: LBGENERAL-28 URL: http://jira.qos.ch/browse/LBGENERAL-28 Project: logback-general Issue Type: New Feature Affects Versions: 0.9.16 Reporter: C Z Assignee: Logback dev list Currently, logback exposes an MBean to provide logging related controls via the inclusion of the following configuration: <jmxConfigurator /> However, this *always* uses the platform mbean server: MBeanServer mbs = ManagementFactory.getPlatformMBeanServer(); (can be found in class JMXConfiguratorAction, line# 54). I think it should be able to provide an option to use an MBeanServer with a specific domain, which is provided in the JMX spec. The code needs to use the following: List servers = MBeanServerFactory.findMBeanServer(agentId); and then use the first one, possibly giving a warning if there are multiple. This is crucial for our use-case, especially because we have separate MBeanServers in our web-app environment, and we need each application to use its own MBeanServer with a specific domain. So, all in all, can we add an attribute of some sort, so that we can do the following: <jmxConfigurator agentid="someMbeanServerId"/> and if this is set, the code uses: MBeanServer mbs = MBeanServerFactory.findMBeanServer(agentId).get(0); and otherwise, it does whatever it used to do. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

2 2

[JIRA] Created: (LBCLASSIC-166) Static logger references in shared classes doesn't work with JNDIContextSelector
by Gabriele Contini (JIRA) 26 Mar '10

26 Mar '10

Static logger references in shared classes doesn't work with JNDIContextSelector -------------------------------------------------------------------------------- Key: LBCLASSIC-166 URL: http://jira.qos.ch/browse/LBCLASSIC-166 Project: logback-classic Issue Type: New Feature Components: Other Affects Versions: 0.917 Reporter: Gabriele Contini Assignee: Logback dev list The issue was already described in: LBCLASSIC-87 by Lars Ködderitzsch {quote} As Jens already pointed out the most common pattern to logging (even in common libaries as Jakarta Commons, Spring whatsoever) is using a static logger, eg. public class XY { private static [final] Logger log = LoggerFactory.getLogger(XY.class); ... } If such libraries are shared (either by being in tomcats shared libs, or by being directly in the ear) by multiple webapps, the logger context wins in which the original loading of the logging class (aka XY, see above) happens. To make an example. An ear contains two webapplications A and B, both have different logger context CtxA and CtxB. A library (say Spring for instance) is place in the ear and both webapps reference the library through their MANIFEST.MF. On application startup webapp A gets initialized first, during initialisation the classes of the shared libary (Spring) are loaded, therefor static loggers initialized with the context CtxA. Now webapp B gets initialized, classes of the shared libary are already loaded by the EARClassloader, the loggers continue to use CtxA. At runtime regardless of wich logger context will be set by context selectors of the webapps, all logging done by the shared classes will always go to CtxA. One can argue that common libaries should not use static loggers, but always obtain loggers freshly from the LoggerFactory. But that is a pipe dream, because the de-facto pattern in obtaining and using a logger is through a static field as depicted above. To achieve true per-webapp logging, already initialized loggers need to be able to switch logger contexts, for instance through TreadLocal or other mechanisms. {quote} And was closed with: {quote} If you need static references problem to be dealt with, then please file a *new* jira issue. {quote} I want to separate hibernate logs in my j2ee application (under jboss). Hibernate is included in server libs, and it uses static loggers: {code} public class XY { private static [final] Logger log = LoggerFactory.getLogger(XY.class); ... } {code} I also want to use separate configuration files for each j2ee application (so sifting appender is not enough), thus i'm using the JndiContextSelector. Logback is unable to switch context once hibernate is initialized, and the logger separation doesn't work for static loggers in shared libraries. --- I also tried to do my own ContextSelector in order to have a Logger that perform a jndi lookup when it's needed but: * ''LoggerContext'' has a method ''getLogger(String)'' that is final. I can't override LoggerContext and return my own instance of Logger. * In my opinion the ''contextSelector.getLoggerContext()'' could return a ''ILoggerFactory'' (there is no need it returns a full ''LoggerContext'' Best regards. Gabriele Contini -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

2 4

[JIRA] Created: (LBACCESS-11) Jetty 7.x support for logback-access
by Daniel Peters (JIRA) 25 Mar '10

25 Mar '10

Jetty 7.x support for logback-access ------------------------------------ Key: LBACCESS-11 URL: http://jira.qos.ch/browse/LBACCESS-11 Project: logback-access Issue Type: Improvement Affects Versions: 0.9.18 Environment: Jetty 7.0.1 - http://eclipse.org/jetty/ Reporter: Daniel Peters Assignee: Logback dev list Since Jetty moved to the Eclipse Foundation, all package names changed from "mortbay" to "eclipse", which is why logback-access 0.9.18 doesnt work with Jetty7. When configured like described in http://logback.qos.ch/access.html I get this: Caused by: java.lang.NoClassDefFoundError: org/mortbay/jetty/RequestLog Can you maybe create a 2nd Jetty RequestLogImpl that uses the new "org.eclipse.jetty"-imports? Thanks! -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

3 5

[JIRA] Created: (LBCORE-95) BurstFilter
by Ceki Gulcu (JIRA) 25 Mar '10

25 Mar '10

BurstFilter ----------- Key: LBCORE-95 URL: http://jira.qos.ch/browse/LBCORE-95 Project: logback-core Issue Type: New Feature Reporter: Ceki Gulcu Assignee: Ceki Gulcu A filter to regulate log message bursts, useful for things like throttling the number of logs sent to an e-mail appender when the database is unavailable. See http://www.qos.ch/pipermail/logback-dev/2009-March/003921.html -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

1 4

[JIRA] Created: (LBGENERAL-38) Javadoc is missing in logback 0.9.18 distribution
by Jean-Luc Geering (JIRA) 24 Mar '10

24 Mar '10

Javadoc is missing in logback 0.9.18 distribution ------------------------------------------------- Key: LBGENERAL-38 URL: http://jira.qos.ch/browse/LBGENERAL-38 Project: logback-general Issue Type: Task Components: build-structure Affects Versions: 0.9.18 Reporter: Jean-Luc Geering Assignee: Logback dev list http://logback.qos.ch/dist/logback-0.9.17.tar.gz(or .zip) packaged the javadoc in docs/apidocs http://logback.qos.ch/dist/logback-0.9.18.tar.gz(or .zip) doesn't contain this folder -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

3 6

[JIRA] Created: (LBCLASSIC-148) Deadlock with DBAppender
by Mel T. (JIRA) 24 Mar '10

24 Mar '10

Deadlock with DBAppender ------------------------ Key: LBCLASSIC-148 URL: http://jira.qos.ch/browse/LBCLASSIC-148 Project: logback-classic Issue Type: Bug Affects Versions: 0.917 Environment: MacBook Pro 17" Intel Core 2 Duo 2.33 GHz OS X 10.5.7 (9J61) java version "1.6.0_13" slf4j 1.5.8 c3p0 0.9.1.2 Reporter: Mel T. Assignee: Logback dev list A deadlock occurs when running the attached application. For this lock to occur, we need to: - load a configuration other than the default in java app. - not have any default logback.xml in the classpath. - have these jars in the classpath c3p0-0.9.1.2.jar logback-core-0.9.17.jar logback-classic-0.9.17.jar log4j-over-slf4j-1.5.8.jar slf4j-api-1.5.8.jar mysql.jar - Place attached the logback2.xml file in the same package as the attached MainTest class. - adapt the logback2.xml file to your DB schema. (note that you do not need to have the DB tables created for the lock to occur, but the test is more conclusive if you have a valid DB connection and schema.) - Execute this java -cp ".:../c3p0-0.9.1.2.jar:../logback-core-0.9.17.jar:../logback-classic-0.9.17.jar:../log4j-over-slf4j-1.5.8.jar:../slf4j-api-1.5.8.jar:../mysql.jar" MainTest The application should output the logger status and hang. Workaround: - Do not include log4j-over-slf4j in the classpath, - or add a default logback.xml that will be loaded first (just copy and rename logback2.xml in same default package) - or use the attached logback2-works.xml instead of logback2.xml (modify MainTest to load that file instead.) In this last one, the change is that the DB appender is only used in a logger element instead of the root. I did not spend the time to look at the cause of this locking problem as the workarounds allow me to move on. So I hope this really belong in jira, and not a config/usage error on my part. -- Mel T. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

2 11

[JIRA] Created: (LBGENERAL-39) Add support for H2 in DBAppender
by Ceki Gulcu (JIRA) 23 Mar '10

23 Mar '10

Add support for H2 in DBAppender -------------------------------- Key: LBGENERAL-39 URL: http://jira.qos.ch/browse/LBGENERAL-39 Project: logback-general Issue Type: New Feature Reporter: Ceki Gulcu Assignee: Logback dev list -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

1 1

[JIRA] Created: (LBCORE-97) Starvation on AppenderBase.doAppend
by Joern Huxhorn (JIRA) 23 Mar '10

23 Mar '10

Starvation on AppenderBase.doAppend ----------------------------------- Key: LBCORE-97 URL: http://jira.qos.ch/browse/LBCORE-97 Project: logback-core Issue Type: Bug Components: Appender Affects Versions: 0.9.15 Reporter: Joern Huxhorn Assignee: Logback dev list Priority: Critical Attachments: AppenderBaseLock.patch The problem we are facing here is that several threads are trying to obtain the object monitor of the exact same resource, namely the Appender. The assumption that multiple threads waiting for ownership of a monitor would receive the monitor in the order that they tried to acquire it is simply incorrect. This is documented behavior. See the last paragraph of [1]: "Likewise, no assumptions should be made about the order in which threads are granted ownership of a monitor or the order in which threads wake in response to the notify or notifyAll method. An excellent reference for these topics is Chapter 9, "Threads," in Joshua Bloch's book Effective Java Programming Language Guide. " The documentation of Object.notifyAll() [2] states the following: "[..] The awakened threads will compete in the usual manner with any other threads that might be actively competing to synchronize on this object; for example, the awakened threads enjoy no reliable privilege or disadvantage in being the next thread to lock this object." The documentation in [5] also lists the following as a weak-spot of Built-in Synchronization in J2SE 1.4.x: "No way to alter the semantics of a lock, for example, with respect to reentrancy, read versus write protection, or fairness." In LBCORE-96 Ceki stated the following: "Logback is using the standard synchronization mechanism found in the JDK. You are saying that that mechanism is unsuitable which, truth be told, sounds quite suspicious." Yes, it's unsuitable in this situation because obtaining the object monitor is not guaranteed to be fair. It's not working in a "first come, first serve" manner. I assumed that, too, but it isn't the case. I had the exact same problem in Lilith some time ago because I made exactly this very same wrong assumption. Fairness of object monitor lock acquisition seems to be logical and a "good thing" but it's not specified that way, and for good reasons. Without fairness in place, the VM can optimize the execution of an application much better. A context switch is a costly operation for a CPU so performance is increased significantly if such a switch can be omitted. Concerning a test-case, this would be pretty hardcore to implement since it's 100% implementation dependent. One implementation *may* handle the locking of object monitors fairly while others don't. Therefore I'll try the following first: I assume I could convince you that object monitor acquisition (OMA) is not fair. If we take that for granted the following scenario should show my point: There are 4 Threads (e.g. 4 Threads handling concurrent webapp-requests) that have one single chokepoint, the appender. Since OMA isn't fair, it's possible that only some of those threads can work at all (see [3]). Exactly that is happening right now in our webapp running on a 4-core Solaris on JDK6. My next assumption is that we both agree that this isn't acceptable behavior. Logging should be handled "first come, first served" so that if Thread A is waiting to append an event and Thread B is waiting to append an event subsequently, the actual appending order should be A, then B. This is currently not the case. It *may* be the case but there is no guarantee of it. One could even argue that the logging system is working incorrectly (aside from the starvation problem) because events are appended in a different order than the actual execution of the logger call. The only way to prevent this is the introduction of fairness into the locking process. The way to do this is to use ReentrantLock [4]. >From the ReentrantLock javadoc: "The constructor for this class accepts an optional fairness parameter. When set true, under contention, locks favor granting access to the longest-waiting thread. Otherwise this lock does not guarantee any particular access order. Programs using fair locks accessed by many threads may display lower overall throughput (i.e., are slower; often much slower) than those using the default setting, but have smaller variances in times to obtain locks and guarantee lack of starvation." This is exactly what is absolutely necessary for a logging framework. Logging must *NOT* be able to introduce a starvation problem into a multi-threaded application! And it does exactly that right now. I doubt that the performance will decrease in a significant manner due to the use of a fair ReentrantLock but even if this would be the case it would be necessary anyway. Otherwise, appending simply isn't working correctly. [1] http://java.sun.com/j2se/1.5.0/docs/guide/vm/thread-priorities.html [2] http://java.sun.com/javase/6/docs/api/java/lang/Object.html#notifyAll() [3] http://en.wikipedia.org/wiki/Starvation_(computing) [4] http://java.sun.com/javase/6/docs/api/java/util/concurrent/locks/ReentrantL… [5] http://java.sun.com/developer/technicalArticles/J2SE/concurrency/ -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://jira.qos.ch/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira

13 75