[
https://issues.apache.org/jira/browse/TS-1499?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13462066#comment-13462066
]
Kit Chan commented on TS-1499:
------------------------------
Agreed. I think that the keep-alive was handled properly between user and ATS,
and between ATS and origin server.
The root problem is that ESI plugin uses TSFetchUrl() with the request having
the keep-alive connection header and it looks like the success fetch event is
not sent till the keep-alive connection timed out. And so the plugin never got
involved again to handle the success fetch event until really late.
The patch is rather just a dumb approach (quick fix, too) to not use keep-alive
for these ESI include requests.
> ESI plugin does not work well with 'Connection: keep-alive'
> -----------------------------------------------------------
>
> Key: TS-1499
> URL: https://issues.apache.org/jira/browse/TS-1499
> Project: Traffic Server
> Issue Type: Bug
> Components: Plugins
> Affects Versions: 3.3.0
> Reporter: Kit Chan
> Attachments: ts-1499.diff
>
>
> ESI plugin will pass on almost all request headers from user request to all
> async requests that it tried to make for each of the ESI includes.
> We should not be passing the Connection keep-alive header because we don't
> want these requests to be "keep-alive" after they are done.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
