Of course, this doesn't mean that Yann should wait for me... you seem to have a good grasp.
On Feb 25, 2014, at 10:01 AM, Jim Jagielski <j...@jagunet.com> wrote: > Now that I understand the issue and the intent, let > me take a look. There might be a non-obvious but > elegant solution... > > BTW: FWIW I'm traveling the next few days so (1) the T&R > will likely be pushed to next week and (2) I'll be mostly > unavail until next week anyway > > On Feb 25, 2014, at 9:20 AM, Yann Ylavic <ylavic....@gmail.com> wrote: > >> On Tue, Feb 25, 2014 at 2:56 PM, Yann Ylavic <ylavic....@gmail.com> wrote: >>> On Tue, Feb 25, 2014 at 2:46 PM, Yann Ylavic <ylavic....@gmail.com> wrote: >>>> If fact it works if I use the following rule (sorry to have not >>>> thought about it before) : >>>> >>>> RewriteRule "^/(.*)$" "http://localhost/$1"; [P] >>>> <Proxy "unix:/tmp/backend.sock|http://localhost"; disablereuse=off> >>>> </Proxy> >>>> >>>> Since the worker is registered without the UDS path, it's URL the http one. >>>> However this raises the question about different worker with the same >>>> (non-UDS) URL, one with UDS and the other without. >>>> Shouldn't this be checked at startup to avoid strange things to happen? >>> >>> My bad, httpd will define the first one and reuse it for the others. >>> >>> The <Proxy> section works well with UDS though... >>> So maybe there's no patch to apply, simply document about not using >>> the UDS part in RewriteRule. >> >> Only for defined workers, that could be difficult to know. >> So the UDS part should probably be handled in mod_rewrite (ie. not >> fully-qualified). >> It would then be ignored by ap_proxy_pre_request() when a defined >> worker is found (using the worker's one). >
