[
https://issues.apache.org/jira/browse/LOG4J2-1049?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14581666#comment-14581666
]
Robert Schaft commented on LOG4J2-1049:
---------------------------------------
Remko: In the comment to your patch you wrote:
??(Yes, this means there is a possibility the same event is logged twice.)??
I don't see how. If {{put()}} is successful, it will not throw an
{{InterruptedException}}. What am I missing?
> AsyncAppender eats my Interrupt
> -------------------------------
>
> Key: LOG4J2-1049
> URL: https://issues.apache.org/jira/browse/LOG4J2-1049
> Project: Log4j 2
> Issue Type: Bug
> Components: Appenders
> Affects Versions: 2.2, 2.3
> Environment: Java 7, Windows
> Reporter: Robert Schaft
> Assignee: Remko Popma
> Labels: Async
> Fix For: 2.4
>
> Original Estimate: 48h
> Remaining Estimate: 48h
>
> I wrote a LoggingThread that will be stopped by sending
> loggingThread.interrupt() from another Thread.
> {code:title=LoggingThread.java}
> class LoggingThread extends Thread {
> private final static Logger LOGGER = LoggerFactory.getLogger(LoggingThread
> .class);
> public void run() {
> while (!Thread.currentThread().isInterrupted()) {
> LOGGER.debug("{} is logging {}", this, "nothing");
> }
> }
> }
> {code}
> Unfortunately, loggingThread.interrupt() is sometimes set before
> AsyncAppender.append() or at the beginning of that.
> {code:title=AsyncAppender#append() line 153}
> try {
> // wait for free slots in the queue
> queue.put(Log4jLogEvent.serialize(coreEvent, includeLocation));
> appendSuccessful = true;
> } catch (final InterruptedException e) {
> LOGGER.warn("Interrupted while waiting for a free slot in the AsyncAppender
> LogEvent-queue {}",
> getName());
> }
> {code}
> This will make queue.put throw an InterruptedException and remove the
> interrupted flag of the current flag.
> I suggest a simple bugfix: Call Thread by Thread.currentThread().interrupt()
> in the catch clause. This will reset the interrupted flag of the current
> Thread.
> A more complex bugfix which doesn't loose a message:
> {code:title=AsyncAppender#append() line 153}
> Serializable serialized = Log4jLogEvent.serialize(coreEvent, includeLocation);
> try {
> // wait for free slots in the queue
> queue.put(serialized);
> appendSuccessful = true;
> } catch (final InterruptedException e) {
> try {
> // The interrupt is catched and the interrupted flag is unset.
> // Therefore offer() won't return with an InterruptedException
> // unless another Thread sends again an interrupt.
> // Use a timeout to handle the case where the Interrupt e
> // was really meant for a hanging put() and not just
> // coincidently sent at the same time.
> // If the put() was really hanging,
> // offer() would return with false after 10ms and no second
> // external interrupt() is required
> appendSuccessful = queue.offer(serialized, 10L, TimeUnit.MILLISECONDS);
> } catch (final InterruptedException e2) {
> // queue.put is really hanging and someone
> }
> if (!appendSuccessful) {
> LOGGER.warn("Interrupted while waiting for a free slot in the
> AsyncAppender LogEvent-queue {}",
> getName());
> }
> // set the interrupted flag again.
> Thread.currentThread().interrupt();
> }
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
