[
https://issues.apache.org/jira/browse/HTTPCORE-604?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16940946#comment-16940946
]
Oleg Kalnichevski commented on HTTPCORE-604:
--------------------------------------------
[~rhernandez] I was able to reproduce the defect with your instructions. It
turned out the HTTP/1.1 i/o event handler was keeping sessions that could not
be kept alive open for a bit too long, so some queued up message exchanges got
assigned to connections that were about to get closed as non-reusable.
Please review the fix:
https://github.com/apache/httpcomponents-core/commit/e326a64971573cd9a42dc77d65174a906bad85c0
Oleg
> HttpClient5 double-closes IOSessions for HTTP/1.1 HEAD requests
> ---------------------------------------------------------------
>
> Key: HTTPCORE-604
> URL: https://issues.apache.org/jira/browse/HTTPCORE-604
> Project: HttpComponents HttpCore
> Issue Type: Bug
> Components: HttpCore NIO
> Reporter: Richard Hernandez
> Priority: Major
> Fix For: 5.0-beta9
>
>
> HttpClient5 consistently fails HEAD requests that have the "Connection:
> close" header on the response because the runtime always attempts to close
> the IOSession handling the connection twice. The second close triggers a
> ConnectionClosedException, causing the request to fail.
> For some reason, this only affects HEAD requests. All other HTTP methods work
> fine.
> I don't have a repro that I can share yet, but I'm working on a fix. I'm
> letting you know because I don't know if I'll need to make any changes in
> core (I suspect I will). Can you wait to release beta9 until I have a root
> cause?
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
