[
https://issues.apache.org/jira/browse/HTTPCORE-571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16766162#comment-16766162
]
Oleg Kalnichevski commented on HTTPCORE-571:
--------------------------------------------
[~jaymode] HttpClient does not have full control over socket created by
pluggable socket factories. This ensures proper graceful connection termination
regardless of socket type and implementation. Why should we throw this away?
This is another case when some garbage from the server side gets pushed down
onto the client side.
Oleg
> SSLSocket input / output closing causes issues with TLSv1.3 SSLEngine
> ---------------------------------------------------------------------
>
> Key: HTTPCORE-571
> URL: https://issues.apache.org/jira/browse/HTTPCORE-571
> Project: HttpComponents HttpCore
> Issue Type: Improvement
> Components: HttpCore
> Affects Versions: 4.4.11
> Reporter: Jay Modi
> Priority: Major
>
> I've been doing some testing with TLSv1.3 and came across some interesting
> behavior in the OpenJDK SSLEngine running in server mode and being accessed
> with HttpClient. The server side winds up in a loop trying to wrap data to
> send the close. I tracked down the trigger for the issue to the shutdown of
> the input and output on the socket manually before closing. I'll also be
> reporting this on the OpenJDK mailing list as that's where the real issue is.
>
> However, I'm wondering if as an improvement the close() method of
> BHttpConnectionBase could special case SSLSocket? I came across HTTPCORE-22,
> which mentions that this would throw an UnsupportedOperationException in some
> JDKs when used with a SSLSocket, so maybe just doing an instanceof check and
> calling close() if it is an SSLSocket might be best?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
