hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Christopher Brown (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HTTPCLIENT-1495) Finalizer in PoolingHttpClientConnectionManager should not log "shutdown" messages if already shutdown normally
Date Wed, 16 Apr 2014 15:46:16 GMT

    [ https://issues.apache.org/jira/browse/HTTPCLIENT-1495?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13971565#comment-13971565

Christopher Brown commented on HTTPCLIENT-1495:

I've reviewed the changes by looking at the diffs linked to from this ticket.  I won't be
able to test in the near future.

Nevertheless, the changes look fine to me, I implemented something similar locally using a
subclass of the pool the used a volatile boolean "_isShutdown" just to workaround the logging
output.  Your use of an AtomicBoolean is better (I particularly like the use of "compareAndSet"
as a safeguard against concurrent shutdowns).

> Finalizer in PoolingHttpClientConnectionManager should not log "shutdown" messages if
already shutdown normally
> ---------------------------------------------------------------------------------------------------------------
>                 Key: HTTPCLIENT-1495
>                 URL: https://issues.apache.org/jira/browse/HTTPCLIENT-1495
>             Project: HttpComponents HttpClient
>          Issue Type: Improvement
>          Components: HttpConn
>    Affects Versions: 4.3.1
>            Reporter: Christopher Brown
>            Priority: Minor
>             Fix For: 4.3.4, 4.4 Alpha1
> Despite shutting down the CloseableHttpClient instance using "close()", which in turn
calls shutdown() on the PoolingHttpClientConnectionManager, messages are still logged such
as the following in all cases after a JVM garbage collection:
> [Finalizer] DEBUG o.a.h.i.c.PoolingHttpClientConnectionManager - Connection manager is
shutting down
> The finalizer thread always calls "shutdown", which always logs the "shutdown" messages.
 This led me to believe (when tracking memory leaks) that there was a problem in my usage
of HTTP client, when in fact the message was just misleading (everything was fine).
> It would be preferable, perhaps via a volatile boolean instance field on the connection
manager, to suppress the shutdown messages (and maybe even suppress redundant resource cleanup
attempts).  This would help users of the library to concentrate on other issues.
> Thanks.

This message was sent by Atlassian JIRA

To unsubscribe, e-mail: dev-unsubscribe@hc.apache.org
For additional commands, e-mail: dev-help@hc.apache.org

View raw message