🚀 go-pugleaf

In article <1993Jan8.191643.1@molbiol.ox.ac.uk>, jasper@molbiol.ox.ac.uk writes:
>In article <1993Jan8.123112.8651@tc.cornell.edu>, olin@vger.cheme.cornell.edu (Steve Thompson) writes:
>> I have a DEC 3000/400 alpha in a cluster with a VAX 4000/500 and a bunch of
>> other VAXes using ethernet for communications. The alpha has its own
>> system disk, but all other files are on the 4000/500, including all
>> user files, rightslist, mail database, batch/print queue database, UAF,
>> etc. 
>This sounds like you ran cluster_config, defined a few logicals and booted....
>Was it really this easy....?  (I wonder what supported will feel like :)

Yes, it was really that easy. I didn't even run cluster_config (I didn't
even look to see if cluster_config was there!); all I did was to edit
sys$system:modparams.dat on the alpha, insert appropriate definitions, run
autogen, copy my sylogicals.com over from the VAX (which defines SYSUAF,
RIGHTSLIST, etc, and mounts the disk that they are on), and booted. The
cluster formed, and that was that. I use all the same startup procedures as
I did on the VAX; no changes were necessary. In my sylogin.com I used
to run an .exe; I had to change this to run it from two different
directories depending on whether it was VAX or Alpha (using the result
of f$getsyi("hw_model").lt.1024). The .exe for the alpha was VESTed from
the VAX version (I have since compiled it natively for a bit more speed).

>> Everything seems to work perfectly!
>Magic!  Well, thats it, I'm off to the cash-n-carry to get one!

I have been using it for production work. It hasn't done anything untoward
yet! Now I am looking forward to seeing what will happen when we get our
4000/610 alpha and I plug in in to the other end of our DSSI bus! (yes, I
do backups).

>> No-one told me that I couldn't tell you, so I did :-)
>Thats 'cos noone thought we would try !

Certainly in our environment it has caused no problems at all. We have
had the alpha CPU pegged for several days now running production stuff.
I ran a large CPU-bound application previously on a set of DECsystem
5000/200's; it's 5.2 times faster on the Alpha (but maybe it could be
speeded up somewhat; all I did was compile and go for now). Incidentally,
the VESTed version of my CPU hog ran at about the same speed as the
4000/500.

Steve

---------------------------------------------------------------------------
Steve Thompson, System Mangler   Internet: thompson@cheme.cornell.edu
School of Chemical Engineering   Bitnet:   thompson@crnlchme
Olin Hall, Cornell University    Phone:    (607) 255 5573
Ithaca NY 14853
---------------------------------------------------------------------------