hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ortwin Glück (JIRA) <j...@apache.org>
Subject [jira] Reopened: (HTTPCLIENT-623) MultiThreadedHttpConnectionManager exceeds MaxTotalConnections
Date Fri, 19 Jan 2007 09:53:30 GMT

     [ https://issues.apache.org/jira/browse/HTTPCLIENT-623?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Ortwin Glück reopened HTTPCLIENT-623:

      Assignee: Ortwin Glück

I guess kernel developers will disagree about the necessity of TIME_WAIT *g*. The actual time
the kernel waits may be configured (default in Solaris seems to be 4 minutes): http://www.sean.de/Solaris/soltune.html#tcp_time_wait_interval
Socket.setReuseAddress is a JDK 1.4 feature, while HttpClient code is 1.3 compatible. We could
use reflection however. I am trying to come up with a patch. In the mean time you could write
your own ProtocolSocketFactory implementation that sets all kind of fancy flags on the socket.

>  MultiThreadedHttpConnectionManager exceeds MaxTotalConnections
> ---------------------------------------------------------------
>                 Key: HTTPCLIENT-623
>                 URL: https://issues.apache.org/jira/browse/HTTPCLIENT-623
>             Project: HttpComponents HttpClient
>          Issue Type: Bug
>          Components: HttpClient
>    Affects Versions: 3.1 Beta 1
>         Environment: Solaris 10 SPARC, JDK 1.5.0_10
>            Reporter: Anthony Bigbee
>         Assigned To: Ortwin Glück
>            Priority: Critical
> Multithreaded app that uses MultiThreadedHttpConnectionManager and GetMethod is exceeding
MaxTotalConnections specified in MultiThreadedHttpConnectionManager.
> Connections may be unncessarily created instead of reused, but I do not know if this
is the case or whether the MaxTotalConnections is not being honored as the cause of this behavior.
> netstat reveals that the number of http TIME_WAIT state  sockets exceeds MaxTotalConnections
after a short time period for a small number of threads / GetMethod objects.   The threads
with the GetMethod objects *are* invoking GetMethod.releaseConnection().  
> This behavior is reproducible every time.
> Relevant code snippets below:
> httpConnectionManagerParams.setMaxTotalConnections(MAXCONNECTIONS);  //MAXCONNECTIONS
= 50 for testing
> this.connectionManager = new MultiThreadedHttpConnectionManager();
> this.connectionManager.setParams(httpConnectionManagerParams);
> this.httpClient = new HttpClient(this.connectionManager);
> this.httpClient.getParams().setParameter("http.useragent", "Internet Explorer 6.x");
> this.httpClient.getParams().setParameter("http.protocol.version", HttpVersion.HTTP_1_1);
> this.httpClient.getParams().setParameter("http.connection-manager.max-total", MAXCONNECTIONS);
> this.httpClient.getParams().setParameter("http.connection.stalecheck", true);
> this.httpClient.getParams().setParameter("http.connection-manager.max-per-host", MAXCONNECTIONSPERHOST);
> This bug is reproducible every time.

This message is automatically generated by JIRA.
If you think it was sent incorrectly contact one of the administrators: https://issues.apache.org/jira/secure/Administrators.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira


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

View raw message