running out of i-nodes

[utzoo!decvax!duke!unc!smb]

Mkfs normally calculates the number of i-nodes to reserve for a file
system based on its size.  This, of course, assumes a particular average
size for files, to ensure that there are enough i-nodes for each file.
Systems with a lot of netnews and uucp traffic may run into trouble with
this scheme, because a lot of small files are created; hence, you run
out of blocks before you run out of i-nodes.  This is especially true if
/usr/spool is a mounted file system.  The bypass is to use the "proto"
file option of mkfs, which permits explicit specification of the i-list
size.