[BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

ViruSzZ · 2012-03-07 22:46:50

Suddenly, today I needed to know what's the size of the backup tars from my linux-vserver based vps running fully up-to-date arch linux and I can't seem to find out why 'du' is acting like this.

┌─[23:39:09]-[root@stavrovski]-[~]
└─› du -csxh /tmp/*
6.8M    /tmp/CI
11M     /tmp/CI_BEFORE_SED
6.8M    /tmp/CI_w-o_mce
4.0K    /tmp/cron.fAq4LB
4.0K    /tmp/cron.ppna3r
4.0K    /tmp/cron.R3GV1t
4.0K    /tmp/cron.xDo4yg
4.0K    /tmp/hsperfdata_root
1.3M    /tmp/knobs
56M     /tmp/packerbuild-0
20K     /tmp/packertmp-0
8.0K    /tmp/screens
12K     /tmp/test
4.0K    /tmp/yaourt-tmp-d
4.0K    /tmp/yaourt-tmp-root
81M     total

┌─[23:39:23]-[root@stavrovski]-[~]
└─› du -csxh /media/tars/
1.7G    /media/tars/
1.7G    total

┌─[23:39:44]-[root@stavrovski]-[~]
└─› du -csxh /media/tars/*
0       total

┌─[23:39:48]-[root@stavrovski]-[~]
└─› ls -1 /media/tars/|wc -l
40

So, there are not that much files for du to return total zero. And this only happens to be happening to the '/media/tars' directory. Really strange.

┌─[23:43:26]-[root@stavrovski]-[~]
└─› du -csxh /media/*
37M     /media/backups

4.6G    /media/gentoo
1.7G    /media/tars
6.3G    total

Can somebody tell what may be going on out here?

Thanks

EDIT: I ran 'strace' and got lines for each file within that directory that looks like this:

fstatat64(AT_FDCWD, "/media/tars/root-archlinux_29_Feb_2012.tar.gz", {st_mode=S_IFREG|0644, st_size=11314753, ...}, AT_SYMLINK_NOFOLLOW) = 0

Is something wrong on my end or what exactly that means? Google did not clarify this enough for me.

EDIT: Edited the subject to be more relevant to the issue.

Last edited by ViruSzZ (2012-03-09 01:24:30)

ViruSzZ · 2012-03-07 23:20:48

Further investigation showed that this is not happening with coreutils-8.14-1. I have installed 8.15-1 by running Syu on 2012-02-25 and after downgrading it to 8.14 'du' is working as expected.

┌─[00:14:41]-[root@stavrovski]-[~]
└─› du -csxh /media/tars/*
1.1M    /media/tars/etc-archlinux_01_Mar_2012.tar.gz
1.1M    /media/tars/etc-archlinux_02_Mar_2012.tar.gz
1.1M    /media/tars/etc-archlinux_03_Mar_2012.tar.gz
1.2M    /media/tars/etc-archlinux_04_Mar_2012.tar.gz
1.2M    /media/tars/etc-archlinux_05_Mar_2012.tar.gz
1.2M    /media/tars/etc-archlinux_06_Mar_2012.tar.gz
1.2M    /media/tars/etc-archlinux_07_Mar_2012.tar.gz
992K    /media/tars/etc-archlinux_27_Feb_2012.tar.gz
1008K   /media/tars/etc-archlinux_28_Feb_2012.tar.gz
1.0M    /media/tars/etc-archlinux_29_Feb_2012.tar.gz
3.1M    /media/tars/mysqlbackup_01_Mar_2012.tar.gz
3.3M    /media/tars/mysqlbackup_02_Mar_2012.tar.gz
3.5M    /media/tars/mysqlbackup_03_Mar_2012.tar.gz
3.7M    /media/tars/mysqlbackup_04_Mar_2012.tar.gz
3.9M    /media/tars/mysqlbackup_05_Mar_2012.tar.gz
4.2M    /media/tars/mysqlbackup_06_Mar_2012.tar.gz
4.4M    /media/tars/mysqlbackup_07_Mar_2012.tar.gz
2.5M    /media/tars/mysqlbackup_27_Feb_2012.tar.gz
2.7M    /media/tars/mysqlbackup_28_Feb_2012.tar.gz
2.9M    /media/tars/mysqlbackup_29_Feb_2012.tar.gz
155M    /media/tars/public_html_01_Mar_2012.tar.gz
155M    /media/tars/public_html_02_Mar_2012.tar.gz
155M    /media/tars/public_html_03_Mar_2012.tar.gz
155M    /media/tars/public_html_04_Mar_2012.tar.gz
155M    /media/tars/public_html_05_Mar_2012.tar.gz
155M    /media/tars/public_html_06_Mar_2012.tar.gz
155M    /media/tars/public_html_07_Mar_2012.tar.gz
148M    /media/tars/public_html_27_Feb_2012.tar.gz
155M    /media/tars/public_html_28_Feb_2012.tar.gz
155M    /media/tars/public_html_29_Feb_2012.tar.gz
11M     /media/tars/root-archlinux_01_Mar_2012.tar.gz
11M     /media/tars/root-archlinux_02_Mar_2012.tar.gz
11M     /media/tars/root-archlinux_03_Mar_2012.tar.gz
11M     /media/tars/root-archlinux_04_Mar_2012.tar.gz
11M     /media/tars/root-archlinux_05_Mar_2012.tar.gz
11M     /media/tars/root-archlinux_06_Mar_2012.tar.gz
11M     /media/tars/root-archlinux_07_Mar_2012.tar.gz
11M     /media/tars/root-archlinux_27_Feb_2012.tar.gz
11M     /media/tars/root-archlinux_28_Feb_2012.tar.gz
11M     /media/tars/root-archlinux_29_Feb_2012.tar.gz
1.7G    total

and strace with codeutils-8.14-1 gives:

fstatat64(AT_FDCWD, "/media/tars/root-archlinux_29_Feb_2012.tar.gz", {st_mode=S_IFREG|0644, st_size=11314753, ...}, AT_SYMLINK_NOFOLLOW) = 0
write(1, "11M\t/media/tars/root-archlinux_2"..., 5011M  /media/tars/root-archlinux_29_Feb_2012.tar.gz

I don't know why is this happening though and 'stat /media/tars' shows that it's a simple directory with tarballs in it.

┌─[00:19:16]-[root@stavrovski]-[~]
└─› stat /media/tars/
  File: `/media/tars/'
  Size: 4096            Blocks: 8          IO Block: 4096   directory
Device: 807h/2055d      Inode: 6668599     Links: 2
Access: (0755/drwxr-xr-x)  Uid: (    0/    root)   Gid: (    0/    root)

is this a bug in coreutils-8.15-1 or my vps is buggy?

Thanks

djgera · 2012-03-07 23:50:33

Ask upstream.

A more simple case:

$ du -x REGULAR_FILE
$

Anyway -x is supposed to be used in directory arguments.

ViruSzZ · 2012-03-08 00:24:47

Thanks for your input.

Anyhow, I can reproduce this and it __looks__ like a bug and the point is that I have never ever filled a bug report so this is going to be my first time I think o.O. (there always have to have a first time)

upgraded again to the latest version of coreutils and du is not playing nicely again.

┌─[01:17:15]-[root@stavrovski]-[~]
└─› mkdir /tmp/another_test/ && for i in $(seq 1 20); do dd if=/dev/zero of=/tmp/another_test/file-$i  bs=1024  count=10240 ; done

┌─[01:17:17]-[root@stavrovski]-[~]
└─› ls -1 /tmp/another_test/* | wc -l
20

┌─[01:17:36]-[root@stavrovski]-[~]
└─› du -csxh /tmp/another_test/*
0       total

┌─[01:23:13]-[root@stavrovski]-[~]
└─› ls -lah /tmp/another_test/*
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-1
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-10
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-11
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-12
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-13
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-14
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-15
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-16
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-17
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-18
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-19
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-2
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-20
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-3
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-4
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-5
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-6
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-7
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-8
-rw-r--r-- 1 root root 10M Mar  8 01:22 /tmp/another_test/file-9

djgera · 2012-03-08 00:34:34

As I said, just one file is the minimal case with just -x as parameter. There is no need to specify multiple files.

You can send a simple mail asking if such thing is a bug to coreutils user ML.

ViruSzZ · 2012-03-08 00:48:41

Ok, it's a habit of using du that way don't know why . Anyhow, I've just sent an email and will update here as soon as they'll update me .

ViruSzZ · 2012-03-08 22:12:21

To update on this, (i'm bit late as yesterday I engaged with my girlfriend and was //little?// drunk so needed to have a good sleep today.) it seems it is a bug in coreutils-8.15.

See => http://debbugs.gnu.org/cgi/bugreport.cgi?bug=10967

Last edited by ViruSzZ (2012-03-08 22:27:33)

djgera · 2012-03-08 23:58:51

Nice

ViruSzZ · 2012-03-09 01:17:58

What needs to be done now? Should I implement the patch or the devs/tu should apply the patch and push an update of coreutils ?

Thanks

EDIT: The PKGBUILD says maintainer of coreutils is Allan so should I contact him to apply the patch or .....

Last edited by ViruSzZ (2012-03-09 01:20:26)

djgera · 2012-03-09 01:23:15

File a bug report in our bug tracker

ViruSzZ · 2012-03-09 01:25:33

djgera wrote:

File a bug report in our bug tracker

Thanks, Will do it immediately

ViruSzZ · 2012-03-09 01:53:54

Bug report triggered at https://bugs.archlinux.org/task/28833

djgera · 2012-03-09 01:57:21

Assigned Thanks

Arch Linux

#1 2012-03-07 22:46:50

[BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#2 2012-03-07 23:20:48

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#3 2012-03-07 23:50:33

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#4 2012-03-08 00:24:47

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#5 2012-03-08 00:34:34

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#6 2012-03-08 00:48:41

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#7 2012-03-08 22:12:21

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#8 2012-03-08 23:58:51

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#9 2012-03-09 01:17:58

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#10 2012-03-09 01:23:15

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#11 2012-03-09 01:25:33

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#12 2012-03-09 01:53:54

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

#13 2012-03-09 01:57:21

Re: [BUG in coreutils-8.15] - Wierd 'du' behavior on my server ...

Board footer