Having a simple cron script to run xfs_fsr would be VERY
nice!

I have a huge (? :) filesystem (2.6Tb) and a couple of days ago I got
the 'No space left on device' error even though I had 768Gb free. I
immediately realized that it was a INode problem. Looking through the
'Net, a defrag should help with this...

This filesystem is (was) very fragmented (85.86%). The defrag have now
been running for about 24 hours, and the frag is now down to 68.35% which
is STILL to much! This is going to take _a lot of time_!


This filesystem have grown over time (LVM over MD or is it the other
way around? - physical disk, in MD arrays and then the MD's in a LVM).
It whould have been nice if the defrag whould have been run from the
start... And since it's no risk in running xfs_fsr on a system that
doesn't have XFS, a default cron script whould be nice... ?

If you don't want to directly enable this, just ship the cronscript
with the xfs_fsr command commented out. Or just put a couple of notes
in the README. Not that it whould have helped me (or most people) -
who reads documentations these days!? :) If it works, why bother :).
----- s n i p -----
[celia.pts/3]$ cat /etc/cron.daily/xfs
#!/bin/sh

# Run a XFS defrag for two hours (default) every day...
xfs_fsr
----- s n i p -----

I currently have the xfs_fsr commented out because I'm running it
manually and verbosely but will enable it as soon as that is done.
Whenever that will be...

... so in other words, please include a cron script!

I agree, adding a cron script, which can be enabled/disable through
dpkg-reconfigure would be nice.  Nathan, can you look into that for the
next xfsdump release?