hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Oleg Kalnichevski (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HTTPCORE-261) Improve Socket Timeout Handling support when connections are being kept alive and re-used
Date Mon, 27 Jun 2011 09:06:47 GMT

     [ https://issues.apache.org/jira/browse/HTTPCORE-261?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Oleg Kalnichevski updated HTTPCORE-261:

    Issue Type: Bug  (was: Improvement)


I came up with a solution that does not require API changes and straightens the way timeouts
are handled in general. Please review the fix in the trunk. If that resolves the issue for
you, I'll merge the changes up to the 4.1.x branch.


> Improve Socket Timeout Handling support when connections are being kept alive and re-used
> -----------------------------------------------------------------------------------------
>                 Key: HTTPCORE-261
>                 URL: https://issues.apache.org/jira/browse/HTTPCORE-261
>             Project: HttpComponents HttpCore
>          Issue Type: Bug
>          Components: HttpCore NIO
>    Affects Versions: 4.1.1
>            Reporter: Asankha C. Perera
>            Assignee: Asankha C. Perera
>            Priority: Minor
>             Fix For: 4.1.2
>         Attachments: HTTPCORE-261-v4.1.x.patch, HTTPCORE-261-v4.2-alpha1.patch
> This is an enhancement request to improve the support to keep connections alive, and
re-use them effectively without letting them timeout unnecessarily
> The problem can be described against v4.1.1 as follows:
> 1. We establish an outgoing connection and set the socket timeout to 5 seconds (Time
= t0 )
> 2. After receiving the response, we want to keep the socket open for reuse for 30 seconds,
and we update the timeout (Time = t1)
> 3. After 6 seconds since #2 above, we get an opportunity to reuse this connection, and
again we set the socket timeout to 5 seconds
> 4. Almost immediately, the BaseIOReactor timeout check most definitely notices that the
time the socket was last used (t1) is over 5 seconds from the current time, and times out
the socket 
> See http://markmail.org/message/zgnrquosjahqog6f for more details

This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira


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

View raw message