[
https://issues.apache.org/jira/browse/TS-3469?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14387338#comment-14387338
]
Leif Hedstrom commented on TS-3469:
-----------------------------------
Ah, yeah, that makes sense. If you let ATS follow redirects, then the txn goes
back in the HTTP state machine. This is a very particular case, and a hack
added at Yahoo back when. The state machine, and documentations, were
definitely not intended to behave like that :).
So yeah, it sounds like if you let ATS follow redirects, this is expected
behavior. You are basically restarting the TXN, and you want it to go through
the state machine hooks again. This definitely would not be the normal case
though, i.e. the config above is (I'm assuming) necessary to trigger this
behavior.
> Update plugin docs to reflect multiple calls to READ_RESPONSE_HDR
> -----------------------------------------------------------------
>
> Key: TS-3469
> URL: https://issues.apache.org/jira/browse/TS-3469
> Project: Traffic Server
> Issue Type: Improvement
> Components: Documentation
> Reporter: Scott Beardsley
> Assignee: Alan M. Carroll
> Fix For: sometime
>
>
> We ran into a problem where SEND_RESPONSE_HDR was being called multiple times
> leading to a crash. The docs are unclear on this point so we should update
> them to reflect this intermittent case.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
