Unless you have retuned the filesystem all the normal utilities will show
that the filesystem is 100% full at 95%...
Anyway, it's isn't THAT hard to get high amount of fragmentation even
without changing this, the worst I have seen was something like *85%* if I
remember it correctly!
And this wasn't some contrieved test, I just happened to glance at the
line it printed when it had fsck'd that RAID-0 array.. Yes, I was a bit
surprised, to say the least... (the filesystem is fairly full, but it has
never been into the super-user only 5%)
I think it's time to retrieve a new version of defrag, and let it at the
25 GB ext2 filesystem (for extfs/ext2fs/minix/xiafs)... I have to wait to
a time when it can be offline for a LONG time first though, which is why I
haven't done something about it yet.
That, and the fact that it looks like it'll need something like *450* MB
of memory for tablespaces alone for a 25 GB filesystem if one believe the
comments in the README file (defrag-0.73, for extfs/ext2fs/minix/xiafs).
The machine doesn't have that much memory+SWAP! at the moment, fortunately
the 32 GB filesystem on the same machine isn't badly fragmented (different
usage patterns)...
Should probably add memory (up to 512MB) before doing THAT exercise. Yeah,
a backup before starting is probably a good idea too...
Hmmm... With 3 GB max per-process VM (i386 default, right?) defrag
would max out somewhere in the 170 GB area if this is correct (8B/block +
8B/inode), for filesystems with 512-byte blocks and one inode per 4096
bytes.... In this case most of it are for the block map (8/9 of the
memory).
Larger blocks would cut this down somewhat (~1.5TB for 8k block and one
inode per 8192 bytes, with 50/50 usage), but 64-bit hardware really looks
inviting..
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/