[
https://issues.apache.org/jira/browse/TS-749?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13072386#comment-13072386
]
Leif Hedstrom commented on TS-749:
----------------------------------
That's difficult to do, without changing how ATS works. This would require that
you buffer the entire response in ATS, before you start sending anything to the
client. That's simply not how ATS works, and it's intentional for performance
reasons. It could perhaps be possible to make a feature request for this, and
make it optional.
That much said, I believe Varnish works like this (buffering the responses
before sending anything to the client). I think it's an incredibly price to pay
to deal with a hopefully rare error scenario.
Maybe William has some better ideas / thoughts on this?
> Connection hangs if origin server goes down in the middle of a response
> -----------------------------------------------------------------------
>
> Key: TS-749
> URL: https://issues.apache.org/jira/browse/TS-749
> Project: Traffic Server
> Issue Type: Bug
> Components: HTTP
> Affects Versions: 2.1.7, 2.1.6, 2.1.5, 2.1.4
> Environment: Any
> Reporter: William Bardwell
> Assignee: Leif Hedstrom
> Fix For: 2.1.8
>
> Attachments: svn.diff
>
>
> If you start a long HTTP response, and then kill the origin web server, the
> connection to the client goes idle. It looks like TS would accept further
> commands, but the client doesn't know that the response is finished. After a
> few minutes an idle timeout will close the connection. Instead the
> connection should be closed, allowing the client to notice that it only got a
> partial response in some cases, and letting it proceed in all cases.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira