I don't think read_while_writer work on 3.2.x/3.0.x. https://issues.apache.org/jira/browse/TS-1433
On Fri, Jan 27, 2012 at 12:54 AM, Walsh, Peter <[email protected]>wrote: > Hello all,**** > > I wanted to follow up and see if anyone has had success with the > read_while_writer feature. We are currently using version 3.0.1. Any > thoughts or advice would be greatly appreciated.**** > > Thanks!**** > > ** ** > > Pete Walsh**** > > Software Engineer**** > > ** ** > > *From:* > users-return-1309-Peter.Walsh=disney....@trafficserver.apache.org[mailto: > users-return-1309-Peter.Walsh=disney....@trafficserver.apache.org] *On > Behalf Of *Walsh, Peter > *Sent:* Tuesday, January 24, 2012 10:03 AM > *To:* [email protected] > *Subject:* read_while_writer question**** > > ** ** > > Hello,**** > > I'm trying to get the "read while writer" feature working but am seeing > some behavior I was not expecting or perhaps I just don't understand how > its supposed to work. I saw a couple other threads regarding this but > didn't see any clear resolution or open bugs for this in JIRA.**** > > ** ** > > I configured ATS to use the following settings…**** > > ** ** > > CONFIG proxy.config.cache.max_doc_size INT 0**** > > CONFIG proxy.config.cache.enable_read_while_writer INT 1**** > > CONFIG proxy.config.http.background_fill_completed_threshold FLOAT 0**** > > ** ** > > According to this link I found…**** > > http://www.ogre.com/node/386**** > > ** ** > > To test this, I issued a request to a URI that I configured to wait 5 > seconds before responding. During the 5 seconds, I issued a second request > to the same URI. In my wireshark capture, I observed the first request > going to the origin server and the second is blocked until the 1st > request's response is returned, but rather than reading the response from > cache, the second request then is made to the origin server. My > expectations (or maybe misunderstanding) was that the 2nd request would > read the response from cache once the 1st request's response is returned. > At this point re-issuing the request returns a cached response, so I know > the response is cacheable.**** > > ** ** > > Any thoughts? Has anyone had luck getting this feature to work? **** >
