[
https://issues.apache.org/jira/browse/RIVER-31?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15042762#comment-15042762
]
Peter Firmstone commented on RIVER-31:
--------------------------------------
This is still a problem, although relatively easy to fix, will be fixed in
River 3.0.1
> HttpsEndpoint.EndpointInfo.connect should not establishCallContext while
> synchronized
> -------------------------------------------------------------------------------------
>
> Key: RIVER-31
> URL: https://issues.apache.org/jira/browse/RIVER-31
> Project: River
> Issue Type: Bug
> Components: net_jini_jeri
> Reporter: Bob Scheifler
>
> In HttpsEndpoint.EndpointInfo.connect, if an existing connection can't be
> reused, a new HttpsConnection is created and establishCallContext is called
> on it. Both of the latter are done while still holding the EndpointInfo
> lock, and establishCallContext does connection establishment and SSL
> handshake, which can take a long time. As a result, concurrent calls are
> unable to proceed concurrently. I believe these can and should be pulled out
> of the synchronized block (similar to ConnectionManager.connect).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
