|
From: | Assaf Gordon |
Subject: | Re: [Savannah-hackers-public] disk-usage monitoring on vcs.sv.gnu.org |
Date: | Tue, 03 Feb 2015 18:37:59 -0500 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:31.0) Gecko/20100101 Thunderbird/31.4.0 |
On 02/01/2015 06:13 PM, Karl Berry wrote:
On the other hand, if we're faced with a situation where the disk is more-or-less suddenly (say) 95+% full, my immediate question would be which directories have increased consumption the most lately, rather than which are, overall, consuming the mos. That's easy to find by comparing a previous du -S to a current du -S.
I see. So something like the following, once a week: $ time nice -n 19 \ ionice -c 3 \ du --one-file-system --human --bytes --separate-dirs / \ | bzip2 > /tmp/du-disk-scan.txt.bz2 real 44m25.142s user 0m8.333s sys 0m31.502s $ ls -lh /tmp/du-disk-scan.txt.bz2 -rw-r--r-- 1 root root 1.2M Feb 3 23:13 /tmp/du-disk-scan.txt.bz2 Then keep few previous runs in /var/cache/diskusage/[date].txt.bz2 , and if need be, compare them with your script?
[Prev in Thread] | Current Thread | [Next in Thread] |