I threw the source on a spare Pentium Pro with plenty of disk space and 
compiled the package.  I had to make only one modification to get it to 
compile.  On my box (fairly standard) I don't have a 
/usr/src/linux/include/linux/modversions.h.  I commented that out of the 
makefile for the kernel module and it compiled fine.  After I got the 
pcfgen utility and re-ran the vice-server-setup script, I was able to 
start the server and create the volume.  I ran the client setup script 
(by the way, I had to do sh /usr/sbin/venus-setup - add a #!/bin/sh to 
the top ;)  After I ran that, I started venus and it all worked!

It swapped a lot - but it worked.

I mean - it swapped A LOT.  Anyway, I only have 32 megs of ram and was 
running X at the time, etc.  I threw some source code onto the disk and 
compiled it all - the load shot to about 2.3 or so for most of it.  My 
box usually lives around 1.0 when compiling.  There was definatly some 
overhead.  Oddly, even after I had a cached copy there was a lot of disk 
crunching when I went to access the files.

I shutdown the server with cfs shutdown and saw what a Good Thing coda is 
- I still worked with the local copies of the file.  I didn't know how to 
stop venus, so I just killed it.  Just to see what would happen, I 
started the server again.  When I tried to start venus again, it complains:

[root_at_cheetah bin]# venus
Coda Venus, version 4.0 (27)
coda_psdev_open, line: 323 init queue: vcp->vc_requests at 42294656
coda_psdev_write, line: 172 handling downcall
coda_psdev_write, line: 172 handling downcall
coda_psdev_write, line: 172 handling downcall
coda_upcall, line: 180 process 3723 woken up by Venus.
coda_getvattr, line: 370 result: 6
coda_cinode_make: coda_getvattr returned -6
coda_read_super, line: 202 iget root inode failed
Killed
[root_at_cheetah bin]#

Overall - this is all very neat and exciting to me.

Sorry if I post long messages, but I love neat new toys - and this is a 
nice one.

Matt Wilson
mswilso2_at_eos.ncsu.edu