Tim,
A file consists of a set of blocks which start at an inode. When the
set of blocks is exhausted (full), a pointer to the next set of blocks
is established, which is another inode with its own set of blocks. Thus
a file that is bigger than the number of blocks per inode has a daisy
chain of inodes linking the bits together. Hence it is rather easy to
use 620k inodes for 50k files. When you do ls -i on the file, you get
the 'head' inode at the beginning of the file. Of course, you can
change the number of blocks per inode but only when you create the
filesystem in the first place.
Martin.
--
Martin Bly
RAL Tier1 Fabric Team
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On
> Behalf Of Tim Edwards
> Sent: 29 September 2009 10:19
> To: scientific-linux-users
> Subject: More than one inode per file?
>
> The root partition on one of our machines filled up today
> because it ran
> out of indoes. We've cleared out some files but can't see how it could
> possibly use so many inodes. Find reports ~50,000 files while df -i
> claims over 620,000 are used out of 640,000. The fs is ext3.
>
> [r...@localhost /]# find / -xdev | wc -l
> 47636
> [r...@localhost /]# df -i
> Filesystem Inodes IUsed IFree IUse% Mounted on
> /dev/sda2 640000 623423 16577 98% /
> /dev/sda1 30120 35 30085 1% /boot
> none 210437 1 210436 1% /dev/shm
> /dev/sda5 8077312 11 8077301 1% /diska
> /dev/mapper/vg0-diskb
> 2101255552 95868443 2005387109 5% /diskc
> /dev/sdc1 4294967295 30501632 4264465663 1% /diskb
>
>
> Anyone know why this inconsistency could be?
>
> Tim Edwards
>
--
Scanned by iCritical.
