Date: Sat, 2 Oct 1999 17:10:51 +0100 (BST) From: Andrew Gordon <arg@arg1.demon.co.uk> To: Jeff Harris <jeff@dcnv.com> Cc: freebsd-small@freebsd.org Subject: Re: Help... heh. Message-ID: <Pine.BSF.3.96.991002164306.28624A-100000@server.arg.sj.co.uk> In-Reply-To: <19991002002552.B15170@dcnv.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, 2 Oct 1999, Jeff Harris wrote: > Well, so I've been charged (well, okay, this is the way I want to do it), > with making 100+ BSD-based machines all play server on our network, and it > seems to me that installing freebsd onto all 100+ machines (We could easily > scale to 500 or more) would make administrative duties a nightmare. 500+ PCs is probably a nightmare whichever way you do it... > The machines in question are big pIII/500's with 1gig+ of ram each, 100mbit > ethernet, etc. They're fast. My question is, where do I start looking for > info. I've seen lots of diskless stuff (well, okay, a few things) but these > machines aren't totally diskless. They at least have a floppy. Documentation is a bit confusing, as there are several different ways to do diskless booting, and even more ways to arrange the configuration after the initial boot stage. The state of the art on first-stage booting appears to be ports/net/etherboot (pending future developments in /sys/boot). Or you can probably still use the old /sys/i386/boot/netboot stuff if you build an a.out kernel - haven't tried it lately. Both of these are designed to boot from ROM, but if you really want to use those floppy drives, netboot or etherboot will compile a version that will run off a DOS floppy. Or you can boot a kernel of the floppy as for a normal local-disk setup but then mount an NFS root. The /etc/rc.diskless[12] stuff (and associated notes in /usr/share/examples/diskless) gives you one way to go for configuration, though it required a small amount of hacking for my requirements. > Anyone have any tips or pointers on where to look at? I think this is the > direction I should move, as opposed to loading fbsd on each machine, running > cvsup's on all of them and ssh'ing into each one to do a make world (or > at least make installworld), and rsyncing our software (which is only > apache-based). Another variation is CDROM boot/CDROM root. While you don't want CDROM drives to be constantly active (for reliability reasons), in applications which just load one application and keep on running it this can work well - the application and its files get cached in RAM and stay there, with the CD drive idle. I use this in a manufacturing test application - this way, there's nothing in the test stations that can possibly go corrupt even with the operators maltreating them (cutting the power etc). Test results get saved over the network to a machine away from the factory floor. And it's good for version control too: if I issue the right number of CDs, and demand return of the old ones, there's no chance of old versions of software getting used by accident. Drop me a line if you want sample config files or suchlike for any of this. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-small" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.96.991002164306.28624A-100000>