"Tais M. Hansen" wrote:
>
> On Sunday 27 October 2002 17:08, Marc Elsen wrote:
> > Could there be induced bottlenecks as a result of the lan perimeter
> > parameters ?
>
> I doubt it.
>
> > What kind of connection is being realized to the Intranet.
>
> One server/gateway connected to a switch with 4 clients and a switch, which
> holds 3 clients.
>
> > Check port status on network hub, watch for possible errors over there.
>
> There's no collision light, I'm afraid. Only link and activity.
>
> > If the same file size is being transferred from the SQUID machine
> > to a SQUID client, without SQUID, I mean , just ftp a file
> > from the machine to the Intranet, are then the same bottlenecks
> > observed ?
>
> Nope. Just as with the request from the a webserver on the same box, it
> transfers very quickly. FTP showed about 11 MB/s. Same file from same server.
>
> I'm thinking if the bottleneck could be in the communication between squid and
> the diskio (aufs) thread?
Is there any difference if aufs is disabled ?
I myself use Redhat 6.2/Squid 2.5S1 (+aufs) , it works very well for
me.
M.
>
> --
> Regards,
> Tais M. Hansen
> OSD
-- 'Time is a consequence of Matter thus General Relativity is a direct consequence of QM (M.E. Mar 2002)Received on Sun Oct 27 2002 - 09:26:47 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:10:55 MST