[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[linux-lvm] LVM Snapshot create trouble



Hello,
 
This post relates to an issue posted by Steffen Plotner back in
December.  See
https://www.redhat.com/archives/linux-lvm/2005-December/msg00105.html
I am experiencing the same type of problem.
 
On 2.6.13-1.1532_FC4:
When a lot of write activity is going on my server goes into a dead
lock trying to create a snapshot.  The write activity is from creating
the snap itself, as disk i/o normally is not intense on this server.

lvm_snap_copy> source: /dev/vg_scsi_0/lv_os_root
lvm_snap_copy> target: /dev/vg_scsi_0/lv_os_root_alt
lvm_snap_create> source: /dev/vg_scsi_0/lv_os_root
lvm_snap_create> target: snap
lvm_snap_create> snap_space: 15%
lvm_snap_create> source uuid: Jm4sGV-fGt5-XRHl-ESds-AERY-pZtq-GOg8Tu
lvm_snap_create> short name: lv_os_root
lvm_snap_create> size: 8388608000 bytes
lvm_snap_create> snap_size: 1200 Mb
lvm_snap_create> lvcreate -s -L1200 -n lv_os_root_snap
/dev/vg_scsi_0/lv_os_root
  Rounding up size to full physical extent 1.19 GB

At this point no I/O was possible.  No ssh, no x terminal, no logging
into the console,  nothing at all!  Disk activity was null.  In VNC you
could see the load window all black filled.  I had to push the reset
button on the server.

This is on some very decent hardware- 3Ghz Xenon, 4Gb RAM, 7200RPM SCSI
Hardware raid controller drives.

When I run the same command on a different system, same hardware (but a
bit less loaded CPU wise), I did not yet have this problem, however I
have been able to reproduce this same type of issue, it seems when there
is continuous i/o, one can not remove the snapshot until all i/o
ceases.

Can anyone reproduce this problem?  Now there are at least two of us
with the same exact issue.

Tim


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]