This error message will be printed after failure of openvz guest os and attempt to stop vm by vzctl stop <container_id>. As I find out on the web there is problem with some unkillable process which is i D state what can be discovered with ps -aux command.
This D state is probably waiting state to finish some I/O operations so there must be some problem with local storage or network. dmesg display some errors:
But dm-6 is not existed according /sys/block/ at moment of freezed process state so it must be some kind of virtual lvm2 file system used temporary by running vm.
But on my second hw node dmesg show me similar report.
Only simple way to get rid of D state process is reboot whole node.
Is there any another way to remove these unkillable processes?
What could cause this kind of failure?
I have installed Proxmox VE 1.1 which contain kernel 2.6.24-2-pve. Probem occured on diffrent hosts with slightly different hw configuration so probably there is not depend directly on hw configuration.
So far there were problem only on vm using template fedora-10-x86_64-default-20090318.tar.gz and installed with Xorg with GNOME environment. But it could be simply cause of more intense usage of these vms as remote desktop.
Error occured some times randomly by usage of different applications but in some situation can be repeated.
For example it is possible to connect to remote nxserver desktop and use some application but some combination as launching krusader and next launching avi movie with kmplayer cause krusader move into D state but nothing suspicious in any log.
Sometimes evan suffices to try to log with su command and process get stuck in D state.
Only way I can currently reproduce this is launchning kmplayer from krusader by double click on file.
There is another process which get to D state in same time
pstree display hiearchy as
So as krusader is KDE(currently version 2.0 works is KDE4 compatible) appliacation there can be issue with KDE. Launching kmplayer on same file from Nautilus working good but from krusader fails. Interesting.
Anyway, is there any way to stop D state processes or find out for which resource are waiting?
And second question what are dm-6 devices and what can cause errors like "lost page write due to I/O error on dm-6"?
Code:
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
500 18762 0.0 0.6 901540 50820 ? D 06:35 0:28 krusader -caption Krusader --icon=krusader_user.png
Code:
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204807 offset 0
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204807 offset 0
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204807 offset 0
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204807 offset 0
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204807 offset 0
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204807 offset 0
EXT3-fs error (device dm-6): ext3_get_inode_loc: unable to read inode block - inode=101209555, block=202408112
EXT3-fs error (device dm-6): ext3_get_inode_loc: unable to read inode block - inode=101209550, block=202408112
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204954 offset 0
EXT3-fs error (device dm-6): ext3_find_entry: reading directory #101204954 offset 0
Buffer I/O error on device dm-6, logical block 1545
lost page write due to I/O error on dm-6But on my second hw node dmesg show me similar report.
Code:
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20693312 offset 0
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20693312 offset 0
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20693312 offset 0
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20693312 offset 0
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20693312 offset 0
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20529154 offset 0
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20694234 offset 0
EXT3-fs error (device dm-8): ext3_find_entry: reading directory #20694234 offset 0
printk: 9 messages suppressed.
Buffer I/O error on device dm-8, logical block 1545
lost page write due to I/O error on dm-8Is there any another way to remove these unkillable processes?
What could cause this kind of failure?
I have installed Proxmox VE 1.1 which contain kernel 2.6.24-2-pve. Probem occured on diffrent hosts with slightly different hw configuration so probably there is not depend directly on hw configuration.
So far there were problem only on vm using template fedora-10-x86_64-default-20090318.tar.gz and installed with Xorg with GNOME environment. But it could be simply cause of more intense usage of these vms as remote desktop.
Error occured some times randomly by usage of different applications but in some situation can be repeated.
For example it is possible to connect to remote nxserver desktop and use some application but some combination as launching krusader and next launching avi movie with kmplayer cause krusader move into D state but nothing suspicious in any log.
Sometimes evan suffices to try to log with su command and process get stuck in D state.
Only way I can currently reproduce this is launchning kmplayer from krusader by double click on file.
There is another process which get to D state in same time
Code:
5055 ? D 0:00 iceauth remove netid=local/george-virt-old:/tmp/.ICE-unix/dcop1524-1240851203
Code:
│ ├─dcopserver,500
│ │ └─dcopserver_shut --nokill
│ │ └─iceauth remove netid=local/george-virt-old:/tmp/.ICE-unix/dcop1524-1240851203Anyway, is there any way to stop D state processes or find out for which resource are waiting?
And second question what are dm-6 devices and what can cause errors like "lost page write due to I/O error on dm-6"?
