whm@arizona.edu (Bill Mitchell) (12/04/87)
We're considering the purchase of a couple of MicroVAX systems to do some
kernel hacking and such and I'm trying to get clear on what runs on what.
Here's the situation as I understand it; any corrections are welcome.
There are two basic MicroVAX systems: the MicroVAX II and the MicroVAX 2000.
Both systems have the basic CPU, but the II has a Qbus while the 2000 has
disk and tape directly connected to the CPU, thus saving the cost and
complexity of the Qbus. MicroVAX packages with a display tube are called
VAXstations.
We're currently running 4.3+NFS from Mt. Xinu and they say that their system
will run on a MicroVAX II, but not a 2000, due to the new interconnect
technology. It is my understanding that 4.3 direct from UCB does not run
on a MicroVAX II. Ultrix support is available for both the II and the 2000,
but it would be much easier for us to stay with 4.3+NFS and just run that
on all our VAXs.
It is my understanding that only Ultrix supports the display tube driver.
Therefore, if we're planning on running 4.3+NFS, the tube wouldn't be of
much use. (Or does anyone know of tube driver available for 4.3+NFS?)
I'm also unclear on the initial bootstrap situation. It appears that if you've
got an Ultrix machine around you don't need a tape drive to do the initial
loading and get an Ultrix up and running. Once you've done this, I'd assume
that you could replace Ultrix with 4.3+NFS. In our case, this would require
us to rely on someone else's Ultrix and I'm not sure that I like that. We
could get one TK50 and move it from one machine to the other on the (hopefully)
rare occasions when a machine needs a tape, but that seems a little hokey,
so maybe the best best is to just put a TK50 on each machine.
If you can offer knowledgeable corrections and/or suggestions on any of the
above, I would certainly appreciate it.
Bill Mitchell
whm@arizona.edu
{allegra,cmcl2,ihnp4,noao}!arizona!whmchris@mimsy.UUCP (Chris Torek) (12/04/87)
In article <3076@megaron.arizona.edu> whm@arizona.edu (Bill Mitchell) writes: >We're currently running 4.3+NFS from Mt. Xinu and they say that their system >will run on a MicroVAX II, but not a 2000, due to the new interconnect >technology. Correct. I believe someone at MIT has the necessary changes to either 4.3 or Mt Xinu to make it run on a Vaxstation 2000. >It is my understanding that 4.3 direct from UCB does not run on a >MicroVAX II. The kernel runs, but there is no boot code, and it may support fewer devices than Mt Xinu. >It is my understanding that only Ultrix supports the display tube driver. >... does anyone know of tube driver available for 4.3+NFS? We (actually CSC) have (has) one for 4.3BSD, but I am not sure of its history; it may be derived from the Ultrix driver. The only tricky parts are the virtual console stuff (we dropped it) and getting the Qbus mapped (some hacky code in autoconf.c does it, but this is the wrong way). >... It appears that if you've got an Ultrix machine around you don't >need a tape drive to do the initial loading and get an Ultrix up and >running. Right. There are some fancy boot floppies (using the VMS boot loader and a `compress'ed vmunix!) that let you boot via Ethernet from a server. Personally, I have never minded tape boots, but then I have never had to use TK50s either :-) . >Once you've done this, I'd assume that you could replace Ultrix >with 4.3+NFS. Or with 4.3BSD. (Disclaimer: I have not done any of this myself; most of this is secondhand information, except for the stuff about getting the QV driver to work, where I, ah, `provided suggestions' :-) .) -- In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7690) Domain: chris@mimsy.umd.edu Path: uunet!mimsy!chris
wyatt@cfa.harvard.EDU (Bill Wyatt) (12/04/87)
in article <3076@megaron.arizona.edu>, whm@arizona.edu (Bill Mitchell) says: > > We're considering the purchase of a couple of MicroVAX systems to do some > kernel hacking and such and I'm trying to get clear on what runs on what. [...] [ Chris Torek responded the other stuff much as I understand it, too] > I'm also unclear on the initial bootstrap situation. It appears that if you've > got an Ultrix machine around you don't need a tape drive to do the initial > loading and get an Ultrix up and running. Yes, but you have to have more than just the other machine running Ultrix. It has to be configured with the software packages in the correct form for downloading to a target machine, which may mean up to ~12 Mbytes of disk storage (less of course if you prune what you don't want). We have this all set up on our Ultrix server node, and it worked just fine. It's not really faster than installing from a TK, but it's convienient. -- Bill UUCP: {husc6,ihnp4,cmcl2,mit-eddie}!harvard!cfa!wyatt Wyatt ARPA: wyatt@cfa.harvard.edu (or) wyatt%cfa@harvard.harvard.edu BITNET: wyatt@cfa2 SPAN: cfa2::wyatt
wesommer@athena.mit.edu (William Sommerfeld) (12/05/87)
In article <9611@mimsy.UUCP> chris@mimsy.UUCP (Chris Torek) writes: >Correct. I believe someone at MIT has the necessary changes to either >4.3 or Mt Xinu to make it run on a Vaxstation 2000. Yes; we run it on a large (~100) number of VS2000's around here. We derived the device drivers from Ultrix 2.0; it was derived from the UW port of 4.3+NFS, which incidentally does come with support for the MicroVAX II. Win Treese <treese@athena.mit.edu> should be able to give you authoritative information on this, but here goes what I believe to be truth: For those of you with "early" test sources to Ultrix 2.0 who are trying to recreate this: be warned that the bad-block replacement code in the disk driver for the VS2000's is just plain broken; this has reportedly been fixed in the latest version. >Right. There are some fancy boot floppies (using the VMS boot >loader and a `compress'ed vmunix!) that let you boot via Ethernet >from a server. Personally, I have never minded tape boots, but >then I have never had to use TK50s either :-) . In the version I've seen, you don't need floppies; you just type 'B XQA0' on the console and it sends out a "help me, I'm a loser" packet over the ethernet using the MOP protocol; assuming the server is listening, and knows about you, it rams a kernel back at you. The Ultrix 2.x load image contains a ramdisk miniroot filesystem containing everything you need, so you can come up and nfs-mount the rest of the world, and run diskless. Tim Shepard (<shep@ptt.lcs.mit.edu>.. he'll probably kill me for mentioning his name) of MIT's Lab for Computer Science successfully reverse-engineered MOP, and has a "network install" of 4.3 for the VAXstation II; this basically uses the same techniques as the miniroot install, except that it uses MOP rather than a tape drive. - Bill Sommerfeld wesommer@athena.mit.edu