From owner-freebsd-hackers  Sat Jan 11 13:14:57 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id NAA06367
          for hackers-outgoing; Sat, 11 Jan 1997 13:14:57 -0800 (PST)
Received: from whale.gu.kiev.ua (whale.gu.net [194.93.190.4])
          by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id NAA06346
          for <hackers@freebsd.org>; Sat, 11 Jan 1997 13:14:48 -0800 (PST)
Received: from trifork.gu.net (trifork.gu.net [194.93.190.194]) by whale.gu.kiev.ua (8.7.5/8.7.3) with SMTP id XAA24960; Sat, 11 Jan 1997 23:12:59 +0200
Date: Sun, 12 Jan 1997 02:17:05 +0200 (EET)
From: Andrew Stesin <stesin@gu.net>
To: Terry Lambert <terry@lambert.org>
cc: The Hermit Hacker <scrappy@hub.org>, hackers@freebsd.org
Subject: Re: mount -o async on a news servre
In-Reply-To: <199701102006.NAA20414@phaeton.artisoft.com>
Message-ID: <Pine.BSF.3.95.970112015625.4758B-100000@trifork.gu.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-hackers@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk


Dear Terry,

thanks a lot for your explicit analysis!  Every time reading
your messages I learn much, it's a great gift.

Now just a few minor additions, sorry.

On Fri, 10 Jan 1997, Terry Lambert wrote:

> > 	Exactly *how* dangerous is setting a file system to async?
> 
> Depends on how much you will miss the data if it disappears.  For
> passive news servers, not very.

	It depends. After such a failure you'll loose sync
	between your actual newsspool and all the databases
	carried by INN or CNews; =>

	rebuild of history and active will be needed to recover
	consistency of the entire news system =>

	terrific amount of time will be lost doing this each time,
	even on a small (1Gb) news spool.

	On the other hand, I assume that crash will result
	only in article lossage from spool, no other consequenses;
	so if you'll tolerate the fact that your history and
	active databases do carry information about
	several hundreds of lost articles -- your'e (probably) Ok.

[...]

> was there before) pointed to by a valid index record, and a valid (but
> orphaned) data record not pointed to by anyone.

	... in the case of news this means "an article in spool
	which isn't in the history+active", right?  It's also bad.

[...]

> The INN history file would probably need to be rebuilt after each crash

	God save me! as I told earlier. :-)

	Just for now I consider that "noatime" mount option
	is the best balance between reliability and performance
	on FreeBSD, right?

Best regards,
Andrew Stesin

nic-hdl: ST73-RIPE