[rhelv6-list] xfs and xfsconvertd issues

[Derek Yarnell]

On 3/1/12 2:55 PM, Stephen John Smoogen wrote:
> On 1 March 2012 12:39, Derek Yarnell <derek at umiacs.umd.edu> wrote:
>> Hi,
>>
>> Has anyone seen the following scenerio.  We have an XFS file system that
>> seems to be going a bit crazy.  Running a fully patched RHEL 6.2 system
>> we are running KVM instances with libvirt with a default pool on a XFS
>> file system.  I destroyed a domain at which point the qemu-kvm process
>> went into zombie (Z) state.  I tried to restart libvirtd and it also
>> went into zombie state.
>>
> 
> Does lsof show if you have anything still talking to the partition
> beyond your working kvm's?
> 

I had run 'file' on the bikeshare.img which has also fallen into device
wait state (D).

# lsof | grep 'libvirt/images'
qemu-kvm   2530      qemu    9u      REG              253,4  41943040000
       131 /var/lib/libvirt/images/itasca.img
qemu-kvm   2530      qemu   11u      REG              253,4 104857600000
       133 /var/lib/libvirt/images/itasca-svn.img
qemu-kvm   2569      qemu    9u      REG              253,4  42949672960
       132 /var/lib/libvirt/images/sketchpad.img
bash      20247      root  cwd       DIR              253,4          144
       128 /var/lib/libvirt/images
strace    34878      root  cwd       DIR              253,4          144
       128 /var/lib/libvirt/images
lsof      36716      root  cwd       DIR              253,4          144
       128 /var/lib/libvirt/images
grep      36717      root  cwd       DIR              253,4          144
       128 /var/lib/libvirt/images
lsof      36718      root  cwd       DIR              253,4          144
       128 /var/lib/libvirt/images
virsh     45921      root  cwd       DIR              253,4          144
       128 /var/lib/libvirt/images
file      50160      root  cwd       DIR              253,4          144
       128 /var/lib/libvirt/images
file      50160      root    3r      REG              253,4  42949672960
       134 /var/lib/libvirt/images/bikeshare.img

There is also a blocked kernel task when the qemu-kvm instance

INFO: task qemu-kvm:48005 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
qemu-kvm      D 0000000000000003     0 48005      1 0x00000080
 ffff880190c67a60 0000000000000082 ffff880000034f40 ffff8801f6360100
 00000000000280da ffff880190c67ae8 ffffffff81123741 ffff8800000241c0
 ffff880217de3078 ffff880190c67fd8 000000000000f4e8 ffff880217de3078
Call Trace:
 [<ffffffff81123741>] ? __alloc_pages_nodemask+0x111/0x940
 [<ffffffff814eef25>] rwsem_down_failed_common+0x95/0x1d0
 [<ffffffff810616f5>] ? dequeue_entity+0x105/0x2e0
 [<ffffffff814ef0b6>] rwsem_down_read_failed+0x26/0x30
 [<ffffffff81276d54>] call_rwsem_down_read_failed+0x14/0x30
 [<ffffffff814ee5b4>] ? down_read+0x24/0x30
 [<ffffffffa02bf77c>] xfs_ilock+0x8c/0xd0 [xfs]
 [<ffffffffa02e5d71>] xfs_rw_ilock+0x31/0x40 [xfs]
 [<ffffffffa02e6cda>] xfs_file_dio_aio_write+0xea/0x2b0 [xfs]
 [<ffffffff81110ac7>] ? unlock_page+0x27/0x30
 [<ffffffffa02e7566>] xfs_file_aio_write+0x2d6/0x2e0 [xfs]
 [<ffffffffa02e7290>] ? xfs_file_aio_write+0x0/0x2e0 [xfs]
 [<ffffffff8117603b>] do_sync_readv_writev+0xfb/0x140
 [<ffffffff810616f5>] ? dequeue_entity+0x105/0x2e0
 [<ffffffff81090a90>] ? autoremove_wake_function+0x0/0x40
 [<ffffffff8120c0d6>] ? security_file_permission+0x16/0x20
 [<ffffffff811770cf>] do_readv_writev+0xcf/0x1f0
 [<ffffffff8106cc0f>] ? release_task+0x36f/0x4e0
 [<ffffffff81177236>] vfs_writev+0x46/0x60
 [<ffffffff811772f2>] sys_pwritev+0xa2/0xc0
 [<ffffffff8100b0f2>] system_call_fastpath+0x16/0x1b

-- 
---
Derek T. Yarnell
University of Maryland
Institute for Advanced Computer Studies