Re: [PLUG] Reclaiming inodes

Usually means you have a ton of tiny files somewhere. Last time I saw that was helping someone who never deleted past kernel builds... If the system is in a working state track down the directory that contains the most inodes, in this past case it was src.

I doubt you actually have this problem, it is possible, but my guess is there are a lot of tiny files somewhere... Maybe spool because mail issue?? Find the directory and you will find your problem.

To give yourself some quick breathing room - run bleachbit to reclaim something and if the system is too full then do all this on a bootable USB. Good luck!

On Thu, Sep 28, 2017, 10:51 AM brent timothy saner <brent.saner@gmail.com> wrote:

On 09/28/2017 10:45 AM, K.S. Bhaskar wrote:
> Might fstransform to transform the filesystem in place to something else
> (xfs seems to never run out of inodes) be an option?
>
> Regards
> -- Bhaskar
>

well, sure, but you should be making a full backup at that point
anyways. ;) fstransform, while awesome, is still doing some pretty crazy
stuff and if this is irreplaceable data, well... i personally wouldn't
risk it without making a backup.

plus there's this:

"""
the device must have a little free space, typically at least 5%

WARNING: transforming an almost full device to 'xfs' file-system can be
tricky:

you need either slightly more free space, typically at least 10%,
or you must be VERY quick at suspending fstransform when either the
source or the target (or both) file-system is almost full and run
'xfs_fsr' on the source or target device (or both) before resuming
fstransform. A future fstransform version may automate this operation.
"""

and this:

"""
If the device contains a HUGE number of files with multiple hard links,
fstransform will be very slow and consume a LOT of memory. Devices with
more than one million files with multiple hard links can cause
fstransform to crash with "out of memory" errors.
"""

i'd say overall, it's faster and safer to just reformat the filesystem
after copying the files to a safe place, and then copying them back when
the mkfs is done.

(though you ARE on to something with XFS; it doesn't really have the
same sort of "oops, i hit the inode limit" issues you run into with
extents filesystems. i wouldn't want to run / on it, though.)

___________________________________________________________________________
Philadelphia Linux Users Group -- http://www.phillylinux.org
Announcements - http://lists.phillylinux.org/mailman/listinfo/plug-announce
General Discussion -- http://lists.phillylinux.org/mailman/listinfo/plug