On Tue, 26 Oct 2004 19:48:39 +0200 (CEST), Henrik Nordstrom
<[EMAIL PROTECTED]> 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.

Reply via email to