Date: Tue, 9 Jul 1996 14:17:52 -0700 (MST) From: Terry Lambert <terry@lambert.org> To: gpalmer@freebsd.org (Gary Palmer) Cc: robmel@nadt.org.uk, questions@freebsd.org Subject: Re: 2.1.5 Stable and rpc.lockd Message-ID: <199607092117.OAA25001@phaeton.artisoft.com> In-Reply-To: <28413.836936774@palmer.demon.co.uk> from "Gary Palmer" at Jul 9, 96 07:26:14 pm
next in thread | previous in thread | raw e-mail | index | archive | help
> > We understand from several correspondents that a working version of > > rpc.lockd is under development on the -current branch. Obviously > > this is of importance to us given our NFS requirement. > > > Does the -stable release contain the rpc.lockd implementation, or should we > > wait for 2.2? > > If you want rpc.lockd (and I'm not sure WHY you want it, nothing you > said above about what you use FreeBSD for, apart from e-mail, really > needs rpc.lockd, and if you are NFS mounting /var/mail, you're also > asking for trouble :-( ) then youy will likely have to wait for > 2.2-RELEASE, and even then I can't guarentee that a fully functional > rpc.lockd will be present. Currently, I believe, a stub rpc.lockd is > in the code, which just blindly accepts all lock requests without > actually doing anything about checking to see if it just granted > multiple clients with locks on a single file. It's mainly to quieten > DOS based NFS clients which insist on talking to a lockd server. Actually, it was to quiet Sun machines, according to the author. > Considering that people (in the know) say that Sun never got rpc.lockd > right either, I'd be surprised if we get WORKING NFS locking which > also inter-operates with Sun code properly. The people who say this are the same people who rail against NFS for being stateless, but I yet to see any of them implement RFS to replace it, so mostly it's just griping. I posted the patches to the kernel for proxy locking support via the fcntl() interface to the -current list. In combination with the rpc.lockd code in the -current tree, it should be relatively trivial to put to gether a server-side locking soloution that actually works and interoperates with Sun's for an NFS server running on FreeBSD. It's about 200 or so lines of code (a trivial hack). Jordan was going to do the code for his BSD internals class at UCB. The client locking is more difficult, and real support will require pre-testing local locking before going to the wire. This means an order change for the VOP layering for VOP_ADVLOCK, something I have been pushing on for about 14 months now. In any case, if you are running a pure FreeBSD environment, you will need both client and server if you need either, since there's nothing else to talk to. 8-). Regards, Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199607092117.OAA25001>