[one-users] occasional error in deleting LVM images

[Shi Jin]

Hi there,

When I issue "onevm delete" for an image running on an LVM snapshot,
sometimes I will get the following error:
Thu Nov  5 16:52:25 2009 [TM][W]: Ignored: LOG - 45 tm_delete.sh:
ERROR: Command "ssh node5 sudo lvremove -f $(echo cloud/$(sudo lvs
--noheadings cloud|awk '{print $1}'|grep lv-one-45))" failed.
Thu Nov  5 16:52:25 2009 [TM][W]: Ignored: LOG - 45 tm_delete.sh:
ERROR:   Can't remove open logical volume "lv-one-45-0"

This error does not happen all the time. Sometimes it works perfectly.

I think the problem might be that this clean up script runs too fast
after the "virsh destroy" and the image might still be in use at the
time of running the script.
Would it be possible to add a delay in the script?
In the ideal case, there should be a checking loop to make sure the VM
is completely destroyed before any clean up job.

I think we have the similar issue for the ebtables-done/delete hooks.

Thanks.

-- 
Shi Jin, Ph.D.