is there a simple explanation for how symlinks are treated under
NFS? here's what i'm after.
with each release of red hat, i immediately copy the CD directory
trees under a main directory structure, but keep each CD tree separate
to distinguish between them, so i'll end up with a top level directory
structure of
/7.3
/inst1 (contents of installation CD 1)
/images
/dosutils
/RedHat ... etc etc ...
/inst2
/inst3
note that each of these subdirectories will have, inside it, the
RedHat/RPMS directory containing their respective RPMS.
for convenience, i also typically create another directory called
"rpms" or something like that contains nothing but symlinks to *all*
of the actual RPM files in the other three RPMS directories. in the
case of 7.3, i end up with a directory containing 1439 symlinks to
all of the RPMs, so i can pretend all of the rpms are in one place
to make perusing them easier.
/7.3
/rpms (contains 1439 symlinks to other RPM files)
/inst1
/inst2 ... etc etc ...
it occurred to me that i can do this, and simultaneously make
available, via NFS, a red hat installation tree, if i create the
directory /7.3/RedHat/RPMS containing all the symlinks. then
a network-based kickstart client could refer to this host and
to the location "/7.3", under which it would find the installation
tree "RedHat/RPMS" to install from.
does the fact that the red hat install tree RedHat/RPMS contain
pose any possible problems? i'm assuming i'd have to export the
entire /7.3 directory since i have to export the locations that
contain the real files as well as the symlinks.
i'd test this but i have only one host at the moment, and
while i can play by exporting and mounting to myself, i'm not
that would constitute a valid test.
comments?
rday
_______________________________________________
Redhat-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/redhat-list
