Re: [squid-users] Raid0

On Tue, 26 Oct 2004 19:48:39 +0200 (CEST), Henrik Nordstrom
<hno@squid-cache.org> wrote:
> On Tue, 26 Oct 2004, Jim Nachlin wrote:
>
> >> Squid already balances the load on multiple drives.
> >
> > How so? Can you explain this?
>
> If using either the diskd or aufs cache_dir type Squid automatically
> selects the correct drive according to it's current load and in addition
> issues multiple concurrent I/O requests (with diskd one I/O request per
> cache_dir, with aufs a number of I/O requests in paralell to the number
> of I/O threads)
>

like I have two drives with mount points as

/cacheA --------> SCSI
/cacheB --------> SCSI

and I want to use them separately, What you suggest would be better
way to utlize these two SCSI's,, as currently a cache already setup in
this manner is creating problem with RAID(?). and I have to check it
at regular times manually for the squid program to work properly.

mostly getting error's like

clientReadRequest: FD 134 Invalid Request
clientReadRequest: FD 188 Invalid Request
clientReadRequest: FD 84 Invalid Request
clientReadRequest: FD 80 Invalid Request
clientReadRequest: FD 39 Invalid Request

and

Config 'request_header_max_size'= 10240 bytes.
Request header is too large (11680 bytes)
Config 'request_header_max_size'= 10240 bytes.
Request header is too large (11680 bytes)
Config 'request_header_max_size'= 10240 bytes.
Request header is too large (11680 bytes)

some times a WARNING : Your Cache is out of File descriptors

and squid stops accepting connections.

any Idea.

> Regards
> Henrik
>

-- 
Nasir Mahmood
Systems + Network Admin.
Asia Net.