dave@lsuc.UUCP (David Sherman) (03/06/85)
We have an MSM-80 disk, set up by mkconf with two large partitions
(28800 blocks each) and three small ones (9600 each). The small
ones are root, tmp and swap.
The other day we ran out of inodes on one of the large partitions.
(Lots of small files.) I dumped the file system, and ran mkfs to
increase the inodes from 6,400 to 10,000. Not knowing what number
to use for mkfs, I tried 28,800 and got a "write error". Playing
around by binary search found that 25,721 was the largest number
which would work, so I used that. All seems to be well.
Later on, I discovered in other places in the manual (not
in man mkfs, of course!) that the "correct" number is 25,600.
QUESTION 1: will I come to any harm for using 25,721 instead of
25,600, or am I just getting use out of 121 blocks that
all you guys don't use?
QUESTION 2: why doesn't 28,800 work? Are 3,000+ blocks taken up
by the indexing of the blocks on the disk?
QUESTION 3: do we really need 9,500 blocks of swap? What's the
minimum anyone's managed to cut it down to and live?
(If you guessed we're low on disk space and wouldn't
mind a few spare Mbytes, you're right.)
Dave Sherman
--
{utzoo pesnta nrcaero utcs hcr}!lsuc!dave
{allegra decvax ihnp4 linus}!utcsri!lsuc!dave