From owner-freebsd-afs Thu Apr 29 17: 5:24 1999 Delivered-To: freebsd-afs@freebsd.org Received: from fledge.watson.org (fledge.watson.org [204.156.12.50]) by hub.freebsd.org (Postfix) with ESMTP id A62A314FB3 for ; Thu, 29 Apr 1999 17:05:18 -0700 (PDT) (envelope-from robert@cyrus.watson.org) Received: from fledge.watson.org (robert@fledge.pr.watson.org [192.0.2.3]) by fledge.watson.org (8.8.8/8.8.8) with SMTP id UAA16846; Thu, 29 Apr 1999 20:05:08 -0400 (EDT) (envelope-from robert@cyrus.watson.org) Date: Thu, 29 Apr 1999 20:05:07 -0400 (EDT) From: Robert Watson X-Sender: robert@fledge.watson.org Reply-To: Robert Watson To: "David E. Cross" Cc: freebsd-afs@FreeBSD.ORG Subject: Re: a new beginning In-Reply-To: <199904292137.RAA00355@cs.rpi.edu> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-afs@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.org On Thu, 29 Apr 1999, David E. Cross wrote: > > Given the availability of Arla, is there a reason we still want to port > > the Transarc AFS client? Wouldn't it be more productive to get really > > bored and write a BSD-licensed AFS server for FreeBSD, making FreeBSD the > > ultimate free network server platform? :) This wouldn't necessarily > > involve learning FreeBSD VM/VFS because you don't have to do the Transarc > > direct inode-open thing, and if you really wanted to the Arla module could > > easily be made to do that. They already have the initial framework for a > > server (all the backend RPC call names and some setup and examples). > > > > If there is interest in writing a free AFS server, I may have a month of > > time I could devote to it this summer. My suspicion is that this would be > > > an incredibly worthwhile project, given the availability of a free client > > already. No doubt the KTH folk are interested in doing this already, but > > last I spoke to them (Assar was here in Pittsburgh about a month ago) they > > hadn't started yet. We already have Quorum code that was originally > > intended for Coda, but hasn't gone in there yet (BSD-licensed), and there > > are some folk around here who have various components that could go into > > it, hopefully all also liberally licensed. > > Hmm... that is something that I had not considered. I do use Arla to a > very limited extent (just getting started with it). Last I had heard > arla sat on the host as a virtual NFS server, so access to /afs/... > would get mapped as a NFS request to arla, who would then make the call > to the AFS server itself. Is this still true? Does Arla do block > caching, or is it still whole file? Finally, how stable is Arla, how > likely am I to panic a Kernel running with arla? I ask since we wish to > make our department FreeBSD machines AFS-ized. Arla is now a fully functional AFS client. It has occasional bugs, and indeed caches the whole file, but works really well and is quite stable. And the Arla folk are very responsive to bug reports. And also have a BSD-style license. I've been using their kernel module for a caching file system I've been playing with running on top of a CDROM drive for performance (don't worry, it doesn't work :). I believe they also have a functional disconnected mode, although they don't have Coda-like reintegration. > I certainly think this is a very worthwhile project, as is the DCE/DFS > project. I am just trying to decide in my mind what is a better use of > my time, writing from scratch something that already exists and is > 'close' to working, or trying to understand someone elses code and bend > it to fit a model it doesn't presently. Now the only thing preventing AFS from being a free file system is the lack of a server that is freely available. And the nice thing about getting to the project first is the ability to dictate the BSD license :-). Or something. But it would certainly be a useful project. The latest version of Arla is 0.23 (the ports collection version is really obscure, btw) -- it comes with the basic RPC calls of a server and some hard-coded stuff. We'd need to get a working Ubik implementation for the supporting database servers, and then have volume servers. The ability to clone, etc, is pretty neat and we'd definitely want to reproduce all that stuff. AFS also makes a great platform to do high volume web hosting, needless to say. Robert N Watson robert@fledge.watson.org http://www.watson.org/~robert/ PGP key fingerprint: AF B5 5F FF A6 4A 79 37 ED 5F 55 E9 58 04 6A B1 Carnegie Mellon University http://www.cmu.edu/ TIS Labs at Network Associates, Inc. http://www.tis.com/ Safeport Network Services http://www.safeport.com/ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-afs" in the body of the message