From owner-freebsd-stable@FreeBSD.ORG Thu Nov 15 22:31:23 2007 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id D7C9816A418 for ; Thu, 15 Nov 2007 22:31:23 +0000 (UTC) (envelope-from rwatson@FreeBSD.org) Received: from cyrus.watson.org (cyrus.watson.org [209.31.154.42]) by mx1.freebsd.org (Postfix) with ESMTP id A77F213C457 for ; Thu, 15 Nov 2007 22:31:23 +0000 (UTC) (envelope-from rwatson@FreeBSD.org) Received: from fledge.watson.org (fledge.watson.org [209.31.154.41]) by cyrus.watson.org (Postfix) with ESMTP id B07164765A; Thu, 15 Nov 2007 17:33:18 -0500 (EST) Date: Thu, 15 Nov 2007 22:31:19 +0000 (GMT) From: Robert Watson X-X-Sender: robert@fledge.watson.org To: Ivan Voras In-Reply-To: Message-ID: <20071115222521.T82897@fledge.watson.org> References: <86y7d1nk8t.fsf@zid.claresco.hr> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII; format=flowed Cc: freebsd-stable@freebsd.org Subject: Re: FreeBSD 6.3 or FreeBSD 7.0 X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 15 Nov 2007 22:31:23 -0000 On Wed, 14 Nov 2007, Ivan Voras wrote: > Marko Lerota wrote: > >> I see that 6.3 and 7.0 is comming. Now I'm using 6.2-RELEASE for my >> servers. To what should I upgrade? Which of them will be stable or >> production release? > > For low-loaded machines, 7.0 is stable enough even with all the new goodies > like ZFS, tmpfs, ULE - I have such a machine with almost 30 days uptime and > have done 24h+ stress testing on another machine before blessing it for > production - but I still wouldn't trust it for mission critical "heads will > roll" type of servers. > > If you can, try it on a spare or lightly loaded server, it's worth it. It's probably worth pointing out three more things: (1) Quite a bit of the work in 7.x has also had to do with improving stability; for example, I've spent a significant amount of time restructuring socket and netinet-layer code to reduce its vulnerability to TCP race conditions that earlier in the 6.x life cycle could lead to panics under high load with man TCP resets in flight. These are papered over in 6.x in a less clean way because the changes to address the underlying source of the problems required significant changes I felt too aggressive to merge to 6.x. So while 7.x comes with potentially destablizing new features, it also comes with a lot of cleanup relative to 6.x that should have net stability benefits (at least in the long run). (2) The areas I would particularly stay away from in sensitive production servers are things like new file systems (ZFS), where the chances and consequences of failure may be greater. (3) At least for now, a 7.x kernel works quite well with a 6.x user space, so if you want to try out the 7.x kernel to see how it runs for you, you can do that without disrupting your application installs. You won't be able to use features like ZFS that depend on new userland tools and libraies, but you will be able to evaluate stability and performance for most base OS features. 7.0 is a .0 release, but I think it's also a really strong .0 release. While I might hesitate to recommend ZFS in less experimental settings, I feel that the 7.0 kernel will prove to be one of our most stable, not to mention most performant, .0 releases to date. I would encourage people to try it out and see how it goes for them, but as with all new releases, to do it with adequate caution and a fallback plan in the event you run into something that hasn't been found or addressed in testing to date. Robert N M Watson Computer Laboratory University of Cambridge