In the last episode (Dec 07), Matt Dillon said:
> Well, too-large a C/G will result in greater file fragmentation,
> because FFS can't manage the file layouts in the cylinder groups as
> well. The default of 16 is definitely too little. 100+ is probably
> too much. Something in the middle will be about right.
Here's the output of "newfs -N -b 16384 -f 2048 -i 262144 -c 107" on a
100-gig RAID filesystem:
Warning: 468 sector(s) in last cylinder unallocated
/dev/da4s1e: 216612396 sectors in 52884 cylinders of 1 tracks, 4096 sectors
105767.8MB in 495 cyl groups (107 c/g, 214.00MB/g, 896 i/g)
and the fsck outputs for eight filesystems formatted like this:
195227 files, 30914874 used, 22947139 free (48723 frags, 2862302 blocks, 0.1%
fragmentation)
16258 files, 33527019 used, 20587979 free (4283 frags, 2572962 blocks, 0.0%
fragmentation)
747 files, 15748237 used, 38366761 free (657 frags, 4795763 blocks, 0.0% fragmentation)
9517 files, 23425935 used, 30689063 free (3039 frags, 3835753 blocks, 0.0%
fragmentation)
1220 files, 16551150 used, 37563848 free (736 frags, 4695389 blocks, 0.0%
fragmentation)
325 files, 25023763 used, 29091235 free (243 frags, 3636374 blocks, 0.0% fragmentation)
99 files, 18106183 used, 36008815 free (127 frags, 4501086 blocks, 0.0% fragmentation)
295 files, 22555093 used, 31559905 free (129 frags, 3944972 blocks, 0.0% fragmentation)
The first filesystem has a full CVS repo, plus a checked-out copy of
-stable, and took 90 seconds to check. All the others took under 20
seconds.
--
Dan Nelson
[EMAIL PROTECTED]
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-hackers" in the body of the message
