Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 26 Aug 1996 15:02:50 -0500 (CDT)
From:      Joe Greco <jgreco@brasil.moneng.mei.com>
To:        nate@mt.sri.com (Nate Williams)
Cc:        jgreco@brasil.moneng.mei.com, michaelv@mindbender.serv.net, freebsd-isp@freebsd.org
Subject:   Re: Anyone using ccd (FreeBSD disk striper) for news
Message-ID:  <199608262002.PAA01517@brasil.moneng.mei.com>
In-Reply-To: <199608261910.NAA19448@rocky.mt.sri.com> from "Nate Williams" at Aug 26, 96 01:10:57 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> > What I want is a solution which postpones the writes until hell freezes
> > over.  Or, better yet, doesn't do them at all.
> 
> Use MFS for your news spool partition. :)
> 
> You *have* to write them down sometime unless you have gigabytes of
> battery backed memory.

Why?

Of what practical value or use is writing back data which will never be
looked at?

Think about it:  if you were to unmount your news spool and remount it -ro,
nnrpd would continue to work just fine because NOTHING ever looks at the
file atime value (which FFS can't/won't modify if you mount -ro)... and if
the only reason you are doing an update is to write back the modified atime,
what the hell is the value of doing the write?

I would agree that writing out the mtime is necessary (or better: "free") 
during article creation because you're writing out the other data at the 
same time, but I do not see the value in updating hundreds or thousands 
of atime's every 30 seconds just because somebody read those files.

Now if I had gigabytes of battery backed memory, I might just make the
entire news spool a MFS..  that could be fast... just gotta wait for RAM
prices to drop a bit more.

... JG



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199608262002.PAA01517>