nnrpd distributed

[Leo.Wong at Datacraft-Asia.com]

Thanks...

Then how should we scale the architecture in high load ? this was based on
an Architecture guide i found on internet....

LEo

-----Original Message-----
From: Russ Allbery [mailto:rra at stanford.edu]
Sent: Monday, July 26, 2004 10:05 AM
To: inn-workers at isc.org
Subject: Re: nnrpd distributed


Leo Wong <Leo.Wong at Datacraft-Asia.com> writes:

> I implemented 2 server , one with innd and innfeed, the other for reader
> only, thus, nnrpd -D.

> They share a NFS volume in between where i put the /var/lib/news and
> /var/spool/news there...

> I found that the reader machine always lag behind the feeding machine for
> like 4-8 hours, i can verify by directly doing a news read in innfeed.

> Can anyone help ? i assume this is pretty standard setup.. ...

Actually, it's recommended against in general, although setting nfsreader
and nfswriter as appropriate may help at least on systems other than
Solaris.  INN makes extensive use of memory-mapped files, which frequently
don't cooperate that well with NFS.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>

    Please send questions to the list rather than mailing me directly.
     <http://www.eyrie.org/~eagle/faqs/questions.html> explains why.


This email and all contents are subject to the following disclaimer:

http://www.datacraft-asia.com/html/001about_us/disclaimer.asp