0

I'm trying to configure a software RAID 10 on my CentOS server. During my test, I disabled a device to check that my filesystem was still working.

When I executed partprobe, the server wasn't responsive, I couldn't get the prompt, so I closed my ssh session and reopened a new one.

Now I have this :

# ps aux | grep partprobe 
root   6319  0.0  0.0  21704   916 ?        D07:40   0:00 partprobe

I tried running kill -9 6319, but that doesn't work !!

# lsof /dev/md10
COMMAND    PID USER   FD   TYPE DEVICE SIZE NODE NAME
partprobe 6319 root    4u   BLK   9,10      8708 /dev/md10

edit: 

root      6212  0.0  0.0  71800   752 ?        D    07:36   0:00 umount /var/lib/mysql/
root      6319  0.0  0.0  21704   916 ?        D    07:40   0:00 partprobe
root      6424  0.0  0.0  71812   856 ?        D    07:47   0:00 mount /dev/md10 /var/lib/mysql/
root      6552  0.0  0.0  71812   860 ?        D    07:56   0:00 mount /dev/md10 /var/lib/mysql/
root      6583  0.0  0.0   5888   668 ?        D    07:57   0:00 shutdown -r 0 w
root      6663  0.0  0.0  71812   856 ?        D    08:02   0:00 mount /dev/md10 /var/lib/mysql/
root      6779  0.0  0.0   5888   668 ?        D    08:08   0:00 shutdown -r 0 w
root      7285  0.0  0.0   5888   668 ?        D    08:54   0:00 shutdown -r 0 w
root      7325  0.0  0.0  90108  3364 ?        Ss   08:59   0:00 sshd: root@pts/8
root      7328  0.0  0.0  66088  1608 pts/8    Ss   09:00   0:00 -bash
root      8332  0.0  0.0  65612   964 pts/8    R+   10:49   0:00 ps aux

Can you help me please ?

Bastien974
  • 1,896
  • 12
  • 44
  • 62
  • Ok, try lazy umount and then stop soft raid. – Ency Jan 24 '11 at 22:02
  • No luck either, tells me device not mounted. I tried to delete, stop, kill everything :( Each time I reboot, my load increase, now it's around 12.... – Bastien974 Jan 24 '11 at 22:17
  • Each time you reboot? Well, ok, try to remount FS read only sending sysrq ALT+SysRq+u (enable by echo 1 > /proc/sys/kernel/sysrq) – Ency Jan 24 '11 at 22:32

2 Answers2

2

From WIKI: (Read last sentence)

Uninterruptible sleep

An uninterruptible sleep state is a sleep state that cannot handle a signal (such as waiting for disk or network IO (input/output)).    
When the process is sleeping uninterruptibly, the signal will be noticed when the process returns from the system call or trap.    
In Unix-like systems the command 'ps -l' uses code "D" for the uninterruptible sleep state of a process.    
Such processes cannot be killed even with 'kill -9' command, and the only nonsophisticated way to get rid of them is to reboot the system.
Ency
  • 1,231
  • 1
  • 19
  • 27
  • I just realized, when I try a "reboot", it close my putty, but not rebooting... Hard rebooting...worked – Bastien974 Jan 24 '11 at 22:34
  • 1
    Next time, it is better to reboot more safely through sysRq, it allows you reboot almost every-time and you can flush disk buffers first. – Ency Jan 24 '11 at 22:40
0

Where it says D07:40, that's actually two fields. The D means that it's in disk I/O wait. You can't kill a process in that state. It has to wait and finish the I/O operation first.

bahamat
  • 6,263
  • 24
  • 28
  • Oh ok, but I really think it's stuck somewhere, I'm trying to kill/stop/reset my software raid with all possible mdadm command, no success. – Bastien974 Jan 24 '11 at 21:34