For one thing, putting a normal filesystem on partition 'b' is best avoided,
unless you have no other free partitions on that disk. If you accidentally
enable anything that assumes 'b' is swap (and there's stuff lurking around
that will because it is such a common use of 'b'), it'll trash the filesystem
on that partition.

> /dev/sd0b  230340 189237    18069    91%   30270       0   100%   /usr/pkgsrc

> It looks like there is still 18069K of space, but no inodes for 
> /dev/sd0b. How can I get more inodes?

Backup the contents of /usr/pkgsrc and redo the filesystem with newfs,
only this time give it the extra args '-i 6144' and see how many inodes
you end up with.

The default -i value is 8192, and that's pretty close to what your df -i
printout suggests was used. The meaning of '-i' is a little weird, it's
supposed to be the average file size per inode, so reducing the number
should get you more inodes. It is not exact, however; newfs does some
rounding trick because of cylinder groups and such, you may have to play
with the -i value a few times to get something that works.

Another thing you can do is use '-m 0' with newfs also. That eliminates
the blocks-reserved-for-root-only setting that is usually put into each
filesystem by newfs. That should buy you another 2% or so, which will go
towards both inodes and data blocks.

Todd Whitesel
toddpw @ best.com