fre 2006-06-30 klockan 10:36 -0400 skrev Steve Snyder:
> Regarding configuration of Squid v2.6:
>
> --enable-kqueue / --disable-kqueue
>
> Given that this is an OS-specific option, shouldn't the name reflect that?
> I'm thinking --enable-bsd-kqueue.
Note: In most situations you don't need to specify any of these as the
configure script picks whatever it finds on your platform. The options
is just there if you want to force certain conditions.
> --enable-linux-tproxy
>
> Why use this instead of --enable-linux-netfilter on Linux 2.6.x systems?
It's not an instead. It's in addition to.
--enable-linut-netfilter is required for interception.
--enable-linux-tproxy is required for spoofing as the client on outgoing
requests.
> --enable-truncate
>
> The configure scripts says: "This uses truncate() instead of unlink() when
> removing cache files. Truncate gives a little performance improvement,
> but may cause problems when used with async I/O."
>
> Is that really universally true? I would have thought that relative
> truncate/unlink performance would be on a filesystem-specific basis.
In either case performance difference is pretty marginal, and
--enable-truncate may lead into problems with the cache so there is not
much reason to use this..
> Yet http://devel.squid-cache.org/collapsed_forwarding/ says:
>
> "This sacrifices general proxy latency in favor for accelerator
> performance and thus should not be enabled unless you are running an
> accelerator."
>
> So... who is right?
Both. It's a tradeoff, and depending on the traffic pattern it can be
either better of worse.. there is no universal answer.
Having it off makes behavior more predictable however.
Regards
Henrik
This archive was generated by hypermail pre-2.1.9 : Sat Jul 01 2006 - 12:00:02 MDT