[RFC] Changes to pass mode

Poul-Henning Kamp phk at phk.freebsd.dk
Wed Feb 7 15:26:20 CET 2007

In message <ujrzm7qf59a.fsf at cat.linpro.no>, Dag-Erling =?iso-8859-1?Q?Sm=F8rgra
v?= writes:
>"Poul-Henning Kamp" <phk at phk.freebsd.dk> writes:
>> Dag-Erling Sm=F8rgrav <des at linpro.no> writes:
>> > There is one downside to PIPE: we can't insert a header telling the
>> > backend that we're acting on behalf of a client, which makes
>> > logging difficult.  We *can* do so with PASS, though I believe we
>> > currently don't.
>> Well, it should be (will be) possible to do that in vcl_recv{} but
>> it may make sense to have a vcl_pipe{} for that job, to avoid
>> repetitions in vcl_recv{}.
>Well, by definition we *can't* do it in PIPE since PIPE is supposed to
>pass traffic back and forth unaltered.  We *can* (and IMHO *should*)
>do it in PASS.


We can do it in the _first_ PIPE transaction, and that could maybe
be used to force the backend to close after that transaction, so
we avoid piping more than that one transaction.

Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

More information about the varnish-misc mailing list