You are not logged in.
Hi
I have home server PC running pure, console Archlinux on it 24/7. It works fine for about 3 years and still do, but after systemd transition I noticed that my server periodically runs systemd-tmpfile daemon which causes intensive and continuous hard disk activity.
My /etc/tmpfiles.d is empty, but here is the content of /usr/lib/tmpfiles.d folder:
-rw-r--r-- 1 root root 30 Sep 20 2012 apache.conf
-rw-r--r-- 1 root root 30 Jan 13 17:58 console.conf
-rw-r--r-- 1 root root 19 Jan 27 06:41 lastlog.conf
-rw-r--r-- 1 root root 1146 Jan 13 17:58 legacy.conf
-rw-r--r-- 1 root root 29 Feb 5 11:09 lirc.conf
-rw-r--r-- 1 root root 24 Mar 5 12:57 mpd.conf
-rw-r--r-- 1 root root 33 Feb 25 19:41 mysqld.conf
-rw-r--r-- 1 root root 27 Jan 27 09:09 nscd.conf
-rw-r--r-- 1 root root 51 Feb 25 17:32 samba.conf
-rw-r--r-- 1 root root 36 Oct 13 19:30 saslauthd.conf
-rw-r--r-- 1 root root 33 Dec 21 16:39 svnserve.conf
-rw-r--r-- 1 root root 736 Jan 13 17:58 systemd.conf
-rw-r--r-- 1 root root 449 Jan 13 17:58 tmp.conf
-rw-r--r-- 1 root root 30 Dec 13 15:24 uuidd.conf
-rw-r--r-- 1 root root 622 Jan 13 17:58 x11.conf
My question is how to know which *.conf file causes such a high HDD activity or how to resolve the problem at all?
Last edited by clovenhoof (2013-03-21 18:20:26)
Offline
Perhaps "lsof -c systemd-tmpfiles" run during that activity would show what files it is working on.
Offline
Ok here is what get with "lsof -c systemd-tmpfile" during this activity:
COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME
systemd-t 1069 root cwd DIR 8,3 4096 2 /
systemd-t 1069 root rtd DIR 8,3 4096 2 /
systemd-t 1069 root txt REG 8,3 59504 290840 /usr/bin/systemd-tmpfiles
systemd-t 1069 root mem REG 8,3 52144 262523 /usr/lib/libnss_files-2.17.so
systemd-t 1069 root mem REG 8,3 138206 262495 /usr/lib/libpthread-2.17.so
systemd-t 1069 root mem REG 8,3 18760 277530 /usr/lib/libattr.so.1.1.0
systemd-t 1069 root mem REG 8,3 2035539 262449 /usr/lib/libc-2.17.so
systemd-t 1069 root mem REG 8,3 31744 262453 /usr/lib/librt-2.17.so
systemd-t 1069 root mem REG 8,3 16776 299945 /usr/lib/libcap.so.2.22
systemd-t 1069 root mem REG 8,3 165436 262497 /usr/lib/ld-2.17.so
systemd-t 1069 root 0r CHR 1,3 0t0 6461 /dev/null
systemd-t 1069 root 1u unix 0xffff8801393df1c0 0t0 23160 socket
systemd-t 1069 root 2u unix 0xffff8801393df1c0 0t0 23160 socket
systemd-t 1069 root 3u unix 0xffff8801393dba80 0t0 23161 socket
systemd-t 1069 root 4r DIR 8,3 9482240 1970431 /var/tmp
systemd-t 1069 root 5r DIR 8,3 4096 793638 /var/tmp/systemd-private-G25Rag
Offline
Doesn't really reveal anything.
Offline
Uhaa, I have just executed " ls -1 /var/tmp/ | wc -l", the result is:
206926 (directories) !!
The names are kind of - "systemd-private-zzoypq"
I cannot understand what happened here?
PS: I cannot even delete them at once:
$ sudo rm -rf systemd-private-*
-bash: /usr/bin/sudo: Argument list too long
Last edited by clovenhoof (2013-03-21 20:40:39)
Offline
I just deleted all these directories, but they are slowly appearing again.
It is funny for me?!
Offline
I just checked size of my "journal" 4.1G with limit of 100M, so then I found that my problem is the same as https://bbs.archlinux.org/viewtopic.php?id=150052.
Offline