I was having the same problem with a Centos 5.5 system, lots of free space, lots of free inodes, but "no space left on device' errors.

I had 30,000+ small files in one directory, deleting them seems to have solved the problem.

FWIW:

rm -rf /home/Movies/jpgs/*

doesn't work, 'argument list too long', so you have to do something like:

find /home/Movies/jpgs/ -name '*.jpg' -delete

Why would you do:
when you could just delete the entire directory (then recreate it if you choose):

To other posters, please be aware that this thread is over SEVEN YEARS OLD

I'm with Wim on this one, isn't it astounding that SEVEN years later, Linux still throws a "Disk Full" error when the failure condition is "To Many Files In Some Unspecified Directory"?

Oh, well, at least Google can still find this thread, and the fix still works.

LQ Wins again!

It is pretty much true of any filesystem that a disk is considered (or reported as) "full" when any critical data structure runs out of space ... and yes, that can include an individual sub-directory. There will always be some amount of disk space which is set-aside for administrative purposes.

And, yes, this kind of message is always confusing. Even seven years later.

Extracted posts to a new thread.

@OP As other members have pointed out, please do not resurrect old threads. Easier to create a new thread with your problem.

Use the command "df -iP".
It will show whether the inode threshold has been reached.

Then rm command has a limit of accepting number of files. That is the reason it has given too many inputs error.
As andrewthomas mentioned, the best to remove the directory itself and then recreate the directory.
If the entire directory can't be removed, some filtering mechanism can be used.

rm *.jpg .. etc

if that also too large, I usually use a for loop mechanism.. But that is a weird one.. there might be some other better options.

for i in `ls *.jpg`
do
rm $i
done

OK, after working for a while, the problem is back:

# df -iP
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/mapper/VolGroup00-LogVol00 37650432 82267 37568165 1% /
/dev/sda1 26104 80 26024 1% /boot
tmpfs 218890 1 218889 1% /dev/shm
/dev/md0 244203520 4023724 240179796 2% /mnt/raid

# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/VolGroup00-LogVol00
140G 139G 0 100% /
/dev/sda1 99M 25M 69M 27% /boot
tmpfs 2.0G 0 2.0G 0% /dev/shm
/dev/md0 1.8T 867G 875G 50% /mnt/raid

and I'm getting errors while running my Perl scripts like:

Exception 450: Output file write error --- out of disk space? `/home/Movies/jpgs/P11246.jpg' @ error/jpeg.c/EmitMessage/235 at pool01.pl line 247.


Where the 'offending' lines in pool01.jpg is where I use ImageMagick to write another output file:

$status = $Image->Write($OFN);
warn "$status" if "$status";

Centos 5.5, Linux 2.6.18-194.3.1.el5PAE

/dev/md0 is the source of the images, and they are being written to /dev/sda1 at /home/Movies/jpgs if that helps.

Any thoughts as to what might be going wrong, and how I could fix it? I've checked /dev/sda1 a couple of times on boot with tune2fs, but that didn't help...

I don't think I'm leaving temporary files somewhere else that might be clogging up a different directory, could there be something else going on? Is there another error message that I find (or increase a logging level of) so that I could get a better idea of what I'm running out of?


Many Thanks In Advance!

Well, don't I feel stupid?

For some reason I thought that /dev/mapper/VolGroup00-LogVol00 was some virtual thing that was always full, as I seem to remember always seeing it full.

I'm clearing out over 100G of space on it now, and will probably end up moving the movies to the raid array.

Urg. Sorry about that...