Unable to view i-node values for Partition

msivadass · 02-11-2009, 03:22 AM

I am not able to view the i-node values for my Linux partition by using
df -i.The output of df -i is

linux-hqif:/ # df -i
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/sda5 0 0 0 - /
udev 224087 432 223655 1% /dev
/dev/sda1 0 0 0 - /boot

The Os Version: SUSE 10.2.

But i can able to view the I-node values of files by using ls -i ( The files are located in same partition(/)).

Is there any other command's to view the i-node information for partition?

Thanks

jschiwal · 02-11-2009, 05:46 AM

A partition doesn't have an i-node. df -i will give you the number of i-nodes used and the number available on the filesystem.

Looking at "df -i /dev/sda*" will list the i-node counts for each partiton on the first disk.

Are you instead looking for the i-node number of the special file itself?
e.g.
ls -i /dev/sda1

msivadass · 02-12-2009, 01:15 AM

Thanks For Ur Reply

But I am getting same Error.

inux-hqif:~ # df -i /dev/sda1
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/sda1 0 0 0 - /boot
linux-hqif:~ #

i92guboj · 02-12-2009, 01:24 AM

What file system are we talking about?

Not all fs's are based on i-nodes.

msivadass · 02-12-2009, 01:30 AM

i want to know how many i-nodes are free for particular partition.

jschiwal · 02-12-2009, 01:33 AM

The answer could depend on whether the device is mounted or not.

Code:

df -i /dev/sda2
Filesystem            Inodes   IUsed   IFree IUse% Mounted on
/dev/sda2             255552  190801   64751   75% /mnt/vista
sudo umount /dev/sda2
df -i /dev/sda2
Filesystem            Inodes   IUsed   IFree IUse% Mounted on
udev                  506053    1379  504674    1% /dev

The first is the number of inodes, which is from the virtual layer (VFS) of the mounted filesystem. These inodes are in memory or memory cache.
The second line after unmounting is the inodes of the root filesystem.

i92guboj · 02-12-2009, 01:42 AM

Quote:

Originally Posted by msivadass

i want to know how many i-nodes are free for particular partition.

I don't understand that phrase at all.

Anyway, what you don't seem to understand is that the i-nodes are not a partition thing. They are just internal structures of the filesystem, and only for some filesystems. There are not i-nodes on a partition, and there are not i-nodes on a disk. You only have i-nodes on a filesystem after you've created it, and only on some of them.

Some others organize the information using different base structures.

So, again: if that partition formated? And if it is: what filesystem are we talking about?

PS. What it seems to me is that you are trying to find how many i-nodes are free for a FAT32 volume, which isn't supported since FAT is not a filesystem based on i-nodes (hence, it truly contains zero i-nodes, and if my memory serves correctly, two File Allocation Tables).

theYinYeti · 02-12-2009, 02:02 AM

OP probably has the same “bug?” as I have:

Code:

[yves@localhost ~]$ df -i /dev/sda6
Sys. de fich.         Inodes   IUtil.  ILib. %IUti. Monté sur
/dev/sda6                  0       0       0    -  /

All numbers are zero, although the partition is definitely mounted (that's / !) and the reiserfs filesystem does use inodes!

Yves.

jschiwal · 02-12-2009, 02:12 AM

Quote:

Originally Posted by theYinYeti

OP probably has the same “bug?” as I have:

Code:

[yves@localhost ~]$ df -i /dev/sda6
Sys. de fich.         Inodes   IUtil.  ILib. %IUti. Monté sur
/dev/sda6                  0       0       0    -  /

All numbers are zero, although the partition is definitely mounted (that's / !) and the reiserfs filesystem does use inodes!

Yves.

What do
sudo df -i /dev/sda6
and
sudo df -i /

return.

If you look at the device itself, you need to have read access to the raw device.

i92guboj · 02-12-2009, 02:16 AM

Thanks for that info, theYinYeti. I wasn't aware of that reiserfs bug. One more -though minor- reason to stay away from reiserfs, I guess.

I googled and indeed it seems that there are people having this same issue. However every post that I can find is like 5 or 6 years old.

jschiwal · 02-12-2009, 02:25 AM

I think that reminds me of something I read, how reiserfs does things its own way instead of integrating into the way things are done by the kernel. It makes the reiserfs not to popular with the kernel authors.

I looked in the manpage and -i is not in the posix manpage. Also, be careful if the target is a link.

theYinYeti · 02-12-2009, 03:10 AM

Don't confuse ReiserFS 3, and ReiserFS 4. While the latter has a long history of clashing with the kernel team, and has quite a record of unstability, the former and older ReiserFS is nothing like that.

The ReiserFS 3 filesystem was the first journaling filesystem to be made available to the “non-tech public” (that was in Mandrake 7.2 if I remember correctly) and I have used it since day one of its availability in the distribution.
The disks went through many brutal power-off, I had electric outages due to storms and such, and there were the occasional game-related- or X-driver-related-crash which meant unclean power off. To this date, the filesystems were ALWAYS recovered.

More over, ReiserFS is still one of the fastest filesystem in usage (although slower to boot).

In contrast, I lost a filesystem once with Ext3, even though I use this one much less. So if the most obvious bug of ReiserFS is a bad output in “df -i” (and is ReiserFS the cause?), then it's OK with me.

Now some more information, before some more people jump on the ReiserFS-fud-bandwagon:

Code:

[root@localhost ~]# df -i
Sys. de fich.         Inodes   IUtil.  ILib. %IUti. Monté sur
/dev/sda6                  0       0       0    -  /
/dev/sda7                  0       0       0    -  /home
none                    111K      28    111K    1% /tmp
//remote_RHCE_2/share      0       0       0    -  /home/yves/.mnt/smb/remote_RHCE_2
//remote_RHCE_1/share   629K     13K    617K    3% /home/yves/.mnt/smb/remote_RHCE_1

[root@localhost ~]# mount
/dev/sda6 on / type reiserfs (rw,noatime,notail,user_xattr)
none on /proc type proc (rw)
/dev/sda7 on /home type reiserfs (rw,noatime,notail,user_xattr)
none on /tmp type tmpfs (rw)
none on /proc/sys/fs/binfmt_misc type binfmt_misc (rw)
none on /sys/fs/fuse/connections type fusectl (rw)
gvfs-fuse-daemon on /home/yves/.gvfs type fuse.gvfs-fuse-daemon (rw,nosuid,nodev,user=yves)
//remote_RHCE_2/share on /home/yves/.mnt/smb/remote_RHCE_2 type cifs (rw,mand)
//remote_RHCE_1/share on /home/yves/.mnt/smb/remote_RHCE_1 type cifs (rw,mand)

As you can see, the CIFS filesytem shows the “bug” for one remote server but not the other. I actually think that:
— this is not a bug, and could be explained somehow,
— it has nothing to do with the filesystem being used.

Yves.

i92guboj · 02-12-2009, 03:32 AM

Quote:

Originally Posted by theYinYeti

— this is not a bug, and could be explained somehow,

It depends on how do you interpret it. We could just say that df and reiserfs are not fully compatible. Just like with vfat and many others.

Quote:

— it has nothing to do with the filesystem being used.

I've never seen this on ext[234], can't comment on other filesystems. This behavior is only expected on fs's where there aren't inodes. However today I've learned that reiserfs can fail at this as well. So I think that the fs matters.

I know nothing about how will CIFS handle this but maybe it emulates the inode properties copying them from the original fs. So, just out of curiosity, what are the underlying fs's for those CIFS shares?

About the reiserfs vs ext3 thingie, better let's keep it out of the thread. There's already enough of that in the net, and neither of us is going to change his mind no matter the arguments.

theYinYeti · 02-12-2009, 06:48 AM

Quote:

Originally Posted by i92guboj

I know nothing about how will CIFS handle this but maybe it emulates the inode properties copying them from the original fs. So, just out of curiosity, what are the underlying fs's for those CIFS shares?

Both are Ext3 on LVM.

Yves.