Re: [squid-users] Performance tunning !

Disabling MTU discovery is not a optimization, it is a workaround to
some of the ugliness of most TCP interception methods.

Ling Hwa Hing wrote:
>
> Hi is there any optimization beside to disable the MTU discovery for the box?
>
> Thank you very much for your help!
>
> Best Regards,
> hwahing
>
> 10:34 AM 6/25/02 +0200, Henrik Nordstrom wrote:
> >If this is a transparent proxy then you are quite likely bitten by Path
> >MTU Discovery problems between the Squid server and your clients. Try
> >configuring the browser to use the cache. If that helps then you will
> >need to take a crash course in path MTU discovery and why most methods
> >for transparent redirection of port 80 breaks it to determine how to
> >best address it in your network.
> >
> >Regards
> >Henrik
> >
> >Ling Hwa Hing wrote:
> > >
> > > Hi,
> > > Me got a box with 1GB of ram and 2 scsi Harddisk.
> > > I've tweak the kernel to handle more TCP connection and increase the
> > > FD_size for my squid. Currently it just serve for 100 client
> > > if I route few more subnet to the box the client unable to receive the
> > > content of the web. I can only see the browser status is just "opening
> > page"
> > >
> > > My box looks like more on traffic saturation than transparent proxy.
> > > And in the cache manager I can't see ant problem for the client. and there
> > > only error message from cache.log are,
> > > 2002/06/24 02:44:30| WARNING: Closing client xxx.xxx.xxx.xxx connection due
> > > to lifetime timeout
> > > 2002/06/24 02:44:30| http://www.livescore.com/img/off.gif
> > > 2002/06/24 02:44:30| WARNING: Closing client xxx.xxx.xxx.xxx connection due
> > > to lifetime timeout
>
> -------------------------------------
> Ling Hwa Hing
> High Tech ASP Sdn Bhd
> Tel: 03-7803 7777
> Fax: 03-7805 3111
> We intend to destroy all dogmatic verbal systems.
Received on Wed Jun 26 2002 - 01:25:37 MDT