Solaris Tip of the Week: a better du experience

In my day job as a Systems Engineer I frequently find myself switching between different UNIX and Linux distributions.  While many of the commands exist on both sides of the aisle, I often find vast differences in the command line parameters that can be consumed by a given command when used in, for example, Linux vs Solaris.

Recently I came upon this again with the need to easily ferret out the majority consumer of drive space on a Solaris 10 system.  While we did have the xpg4 specification support available, the du command was still missing my favorite option “max-depth”.

In Linux I use this to limit the output to only the current directory level so that I don’t have to face to possibility of wading through a tremendously large listing of sub-directories to find the largest directory in the level I am in.  Unfortunately, in Solaris, even with xpg4, the du command doesn’t have this option, so my solution was to pipe the results through egrep and use that to filter out the sub-directories.

Here is some example output from a RedHat Linux 5.11 server:

[code language=”bash” gutter=”false”]
[root@atl4cmweb01 var]# du -h
8.0K ./games
8.0K ./run/saslauthd
8.0K ./run/lvm
8.0K ./run/setrans
8.0K ./run/ppp
8.0K ./run/snmpd
4.0K ./run/mysqld
8.0K ./run/pm
8.0K ./run/dbus
8.0K ./run/nscd
8.0K ./run/console
8.0K ./run/sudo
8.0K ./run/netreport
176K ./run
8.0K ./yp/binding
24K ./yp
8.0K ./lib/games
8.0K ./lib/mysql
4.0K ./lib/nfs/statd/sm.bak
8.0K ./lib/nfs/statd/sm
24K ./lib/nfs/statd
8.0K ./lib/nfs/v4recovery
0 ./lib/nfs/rpc_pipefs/statd
0 ./lib/nfs/rpc_pipefs/portmap
0 ./lib/nfs/rpc_pipefs/nfs/clntf
0 ./lib/nfs/rpc_pipefs/nfs/clnt5
0 ./lib/nfs/rpc_pipefs/nfs/clnt0
0 ./lib/nfs/rpc_pipefs/nfs
0 ./lib/nfs/rpc_pipefs/mount
0 ./lib/nfs/rpc_pipefs/lockd
0 ./lib/nfs/rpc_pipefs
40K ./lib/nfs
8.0K ./lib/dhclient
8.0K ./lib/iscsi/isns
[/code]

Here is the same example ouput from the RedHat server using the max-depth option:

[code language=”bash” gutter=”false”]
[root@atl4cmweb01 var]# du -h –max-depth=1
8.0K ./games
176K ./run
24K ./yp
22M ./lib
32K ./empty
1.5G ./log
12K ./account
236K ./opt
24K ./db
8.0K ./nis
2.9M ./tmp
8.0K ./tmp-webmanagement
40K ./lock
8.0K ./preserve
8.0K ./racoon
16K ./lost+found
1.4M ./spool
8.0K ./net-snmp
83M ./cache
8.0K ./local
1.6G .
[/code]

Here is the command example run without my egrep mod in Solaris 10:

[code language=”bash” gutter=”false”]
[root@atl4sfsbatchb log]# /usr/xpg4/bin/du -h
25K ./webconsole/console
26K ./webconsole
1K ./pool
1K ./swupas
2K ./ilomconfig
1K ./current/ras1_sfsuperbatchb
1K ./current/od1_atl4sfsuperbatchb
4.3G ./current/ras1_atl4sfsbatchb
2.1G ./current/od1_atl4sfsbatchb
560K ./current/avs
2K ./current/ebaps/output
9.3M ./current/ebaps
4.0M ./current/psh
3.1M ./current/autoresponder
5K ./current/fdms_download
29K ./current/fdms_server
109K ./current/fmt
5K ./current/paris/output
653K ./current/paris
1K ./current/od1_sfsuperbatchb
28K ./current/ccTemplateLoader
633K ./current/ccTemplateLoaderLegacy
15M ./current/whinvoices
1K ./current/appmonitor.prod.netsol.com
132M ./current/chase
6.6G ./current
160K ./archive/ccTemplateLoader
1K ./archive/od1_atl4sfsuperbatchb
4.9M ./archive/avs
1K ./archive/ebaps/output
26M ./archive/ebaps
881M ./archive/psh
1014M ./archive/autoresponder
1K ./archive/fdms_download
6.8M ./archive/fdms_server
21M ./archive/paris
1K ./archive/ccTemplateLoaderLegacy
4.1G ./archive/ras1_atl4sfsbatchb
3.1G ./archive/od1_atl4sfsbatchb
5.9G ./archive/chase
102M ./archive/whinvoices
15G ./archive
22G .
[/code]

And here is the improved command output using my egrep mod on the same Solaris server:

[code language=”bash” gutter=”false”]
[root@atl4sfsbatchb log]# /usr/xpg4/bin/du -hx | egrep -v ‘.*/.*/.*’
26K ./webconsole
1K ./pool
1K ./swupas
2K ./ilomconfig
6.6G ./current
15G ./archive
22G .
[/code]

1 thought on “Solaris Tip of the Week: a better du experience”

Leave a Reply to sneezypb Cancel reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: