Dave Lorand <[EMAIL PROTECTED]> writes:
> Do you install each package in a separate volume per @sys, or do you
> have a common volume containing all architechtures? This seems to be
> the major issue about which folks here have waflled in the past. Some
> packages, like Matlab, seem to go better if there is a separate volume
> for each architechture because the authors wrote with single-platform
> installations in mind. Others, like Perl and Sendmail, expect to mingle
> architechtures within their trees, and they know how to separate
> architechture-specific files from architechture-independant ones. I'd
> appreciate anyone's comments on this issue.
It depends on who you talk to :) Most of the software I install, I put it
all in the same volume, unless it's an excessively large (+1 gig) package.
The only benefit I see from making separate volumes for different
architectures deals with releasing the whole package versus releasing the
individual architectures. If I make a change to the solaris version of
IMSL, but not not the aix version, then I only need to release the
architecture that was modified. There may be other benefits of doing
per-architecture volumes, but this is the only one that sticks out in my
head.
Regarding cell layout, I had nothing to do with our current layout, but it
seems to make sense to me. We have 3 main cells, and I see us eventually
merging 2 of the cells and leaving us with a cell mainly for users, and a
cell for software. This is basically how the cells look:
/afs/<cell>/adm - admin databases, logs, tools.
/sadm
/admin
/afs/<cell>/contrib - open source software
/emacs -> emacs1934 - symlink to default version
/emacs1929 - cont.emacs1929
/emacs1934 - cont.emacs1934
/afs/<cell>/dist - licensed software
/sas -> sas610 - symlink to default version
/sas609 - dist.sas609
/sas610 - dist.sas610
/afs/<cell>/lockers - Course, user-purchasable, and departmental lockers
/class/bus590 - lck.cls.bus590
/users/w/wilson - lck.usr.wilson
/dept/pr - lck.dpt.pr
/afs/<cell>/project - Various software and research projects
/linux - project.linux
(I think I would have used prj.linux or proj.linux)
/afs/<cell>/source - Source for in-house programs, kerberos, afs
/krb5 - src.krb5
/afs/<cell>/system - Installed software symlinked on client
machines per architecture. (/usr/local, /usr, etc)
/sun4x_56/usr - sun4x_56.usr
(new naming scheme, was system.sun4x_56.usr, but
because of naming size limits, it was re-done)
/afs/<cell>/users - User directories
/w/wilson - users.wilson
(another cell of ours uses the naming scheme
users.w.wilson, so it just depends on what you
think looks more consistant with your mount-point
structure)
/afs/<cell>/www - Web related stuff
/verity - www.verity
This is our basic setup and like I said, it seems to work well and serve
its purpose. Our naming scheme is such that you know, *usually* without
looking, what the volume name will be for any given mount-point. Anyways,
I hope this helps.
Brian
[ Brian D. Wilson Systems Programmer ]
[ [EMAIL PROTECTED] Computing Services ]
[ 919.515.5498 NC State University ]
