How Big Is That UNIX Directory?
Categories: Database Administration, Information Technology, System Administration, UNIXNeed to know how much space a directory and its contents are taking up on your UNIX system? Here’s what I use:
du -ks directory
The du command is used to summarize disk usage. Without any flags will show you the usage in blocks for every directory and subdirectory specified. Since the number of blocks varies by operating system we add the -k option to specify that we want the output in kilobytes. In many operating systems you could also use -h for a “human-readable” output with abbreviations like B for bytes, K for kilobytes, M for megabytes and so on.
The -s option lets us gather only the sum of the directory specified. Without the -s flag we would get output on every subdirectory as well as the specified directory.
$ du -k stuff
408560 stuff/patch
104 stuff/scripts
408688 stuff
One other thing that is useful for finding the biggest files and directories where there are a lot to sift through is to use a wildcard to size up multiple directories, then pipe the output of du to the sort command like this:
$ du -ks ./* | sort -n
0 ./sdtvolcheck727
8 ./mpztaWqc
8 ./speckeysd.lock
304 ./dtdbcache_:0
408688 ./stuff
With sort we use the -n option to order things by arithmetic value rather than alphabetic value (making 8 come before 304) so we see the largest things at the bottom.
Try it out. As always check the man pages for more info.
For more tips like this check out my book Easy Linux Commands, only $19.95 from Rampant TechPress.
6 Responses to “How Big Is That UNIX Directory?”
-
m Says:
July 15th, 2006 at 7:37 ama darn good way to put your server to knees is running
du -ks / | sort -n
-
Jon Says:
July 15th, 2006 at 11:36 amM brings up a good point. This is pretty IO intensive and will load down your system if you’re running it on a big directory. Use it carefully when you’re on production systems.
-
Jan Bernard Says:
December 2nd, 2006 at 1:11 pmSince you’ve emphasized the easy part in your booktitle, I’d also advice to use the -h parameter. This will decode the large numbers in readable blocks of kilo-, mega- or gigabytes.
-
Jon Emmons Says:
December 3rd, 2006 at 10:30 pmJan,
See the end of the second paragraph:
you could also use -h for a “human-readable†output with abbreviations like B for bytes, K for kilobytes, M for megabytes and so on.
But, if you’re going to use this with the
sort -nyou’ll want everything in the same size so it sorts properly. -
vijaysys Says:
September 7th, 2007 at 5:45 amfriends,
as you know, in solaris 8 systems, we dont have df -h (human readable) option available, can you please any way of using like df -h option (Any script would be helpful for me)sincerely
vijaysys -
Xeleema Says:
October 11th, 2007 at 3:12 pmWell, to answer your question on a “du -h” in Solaris 8…my answer’s not a script, but BlastWave.org.
As long as you don’t have to worry about a production environment, they provide all the GNUness you could ever want on Solaris (any version on either CPU architecture).