[one-users] Thin provisioning and image size
Steven C Timm
timm at fnal.gov
Wed Oct 8 07:03:03 PDT 2014
Interestingly, I found that I was able to do some overcommitting after all, somewhat to my surprise,
as my earlier tests of open nebula 4.4 and 4.6 didn't seem to allow that. This is what I think is happening:
Beginning setup: Image datastore is on NFS, system datastore is on local disk, launching
non-persistent image via CLONE operation in opennebula.
DATASTORE_CAPACITY_CHECK = NO for both.
Initial size of system data store, 450 GB.
Max size of qcow2 image--262 GB
Actuall size of qcow2 image on disk, unexpanded-2.6 GB
Launch first VM--SCHED asks--does system datastore have 262 GB available? Yes.
Launch second VM--SCHED asks--does system datastore have 262 GB available? Yes.. because first one
hasn't expanded out. and so forth.
That amount of overprovisioning is enough for our purposes--we can get 8-9 images (actually a lot more than
that) into the datastore and will only get in trouble if our users all run out the file to full size, which most
of them do not. The system runs out of RAM and CPU before it runs out of disk in this scenario.
Steve Timm
________________________________
From: Ruben S. Montero [rubensm at dacya.ucm.es]
Sent: Wednesday, October 08, 2014 3:50 AM
To: Steven C Timm
Cc: Javier Fontan; users at lists.opennebula.org
Subject: Re: [one-users] Thin provisioning and image size
Hi
Yes, the value for the size of the image is computed in libfs.sh (fs_size) for different file types. For qcow2 images:
SIZE=$($QEMU_IMG info "$1" | sed -n 's/.*(\([0-9]*\) bytes).*/\1/p')
I.e. the virtual size in bytes is used. As this is the max. value of the image it is not updated.
Some thoughts:
1.- Using the max size, enables a proper enforcement of disk quotas. And so prevents a kind of DoS attack , when a user starts a VM and grows the disk over the size of the Datastore, effectively disabling the hypervisor.
2.- Initially we thought of updating the image size every time it is copied back to the image datastore, but this may conflict with the quota system, as mentioned above.
3.- If needed we could implement datastore over-commitment as we do with the hosts, i.e. extend the LIMIT_MB attribute to allow OpenNebula use more datastore storage. (See,
http://docs.opennebula.org/4.8/administration/references/ds_conf.html)
Cheers
Ruben
On Mon, Oct 6, 2014 at 7:45 PM, Steven Timm <timm at fnal.gov<mailto:timm at fnal.gov>> wrote:
I am seeing the opposite problem in One 4.x
and have been ever since we started testing it.
When I do oneimage create using a qcow2 image,
opennebula always reports the size as the absolute
maximum to which the qcow2 file system could expand.
This keeps us from being able to over-provision our
disk on the VM hosts as we've done under Opennebula 3.2 for a long time.
For instance:
[oneadmin at fermicloud198 ~]$ oneimage show 5
IMAGE 5 INFORMATION
ID : 5
NAME : SLF6Vanilla
USER : oneadmin
GROUP : oneadmin
DATASTORE : cloud_images
TYPE : OS
REGISTER TIME : 10/03 16:31:31
PERSISTENT : No
SOURCE : /var/lib/one/datastores/102/180caf99a13146dbd1b60593378d4479
PATH : /tmp/55c42a4cc7f87ea3390bc2bef14212c5
SIZE : 256G
STATE : used
RUNNING_VMS : 1
PERMISSIONS
OWNER : um-
GROUP : u--
OTHER : u--
IMAGE TEMPLATE
DESCRIPTION="SLF6 Vanilla"
DEV_PREFIX="vd"
DRIVER="qcow2"
EC2_AMI="YES
------------------------
[oneadmin at fermicloud198 ~]$ onedatastore show 102
DATASTORE 102 INFORMATION
ID : 102
NAME : cloud_images
USER : njp
GROUP : oneadmin
CLUSTER : cloudworker
TYPE : IMAGE
DS_MAD : fs
TM_MAD : shared
BASE PATH : /var/lib/one/datastores/102
DISK_TYPE : FILE
DATASTORE CAPACITY
TOTAL: : 7T
FREE: : 1.6T
USED: : 4.1G
LIMIT: : -
PERMISSIONS
OWNER : um-
GROUP : u--
OTHER : ---
DATASTORE TEMPLATE
BASE_PATH="/var/lib/one/datastores/"
CLONE_TARGET="SYSTEM"
DATASTORE_CAPACITY_CHECK="NO"
DISK_TYPE="FILE"
DS_MAD="fs"
LN_TARGET="NONE"
TM_MAD="shared"
TYPE="IMAGE_DS"
IMAGES
5
6
[oneadmin at fermicloud198 ~]$ onedatastore show 100
DATASTORE 100 INFORMATION
ID : 100
NAME : localnode
USER : oneadmin
GROUP : oneadmin
CLUSTER : cloudworker
TYPE : SYSTEM
DS_MAD : -
TM_MAD : ssh
BASE PATH : /var/lib/one/datastores/100/100
DISK_TYPE : FILE
DATASTORE CAPACITY
TOTAL: : -
FREE: : -
USED: : -
LIMIT: : -
PERMISSIONS
OWNER : um-
GROUP : u--
OTHER : ---
DATASTORE TEMPLATE
BASE_PATH="/var/lib/one/datastores/100/"
DATASTORE_CAPACITY_CHECK="no"
SHARED="NO"
TM_MAD="ssh"
TYPE="SYSTEM_DS"
IMAGES
[oneadmin at fermicloud198 ~]$
---------------------------------
Any suggestions?
Steve
On Mon, 8 Sep 2014, Javier Fontan wrote:
Then the value makes sense as the units stored are Megabytes.
On Fri, Sep 5, 2014 at 3:34 PM, Daniel Dehennin
<daniel.dehennin at baby-gnu.org<mailto:daniel.dehennin at baby-gnu.org>> wrote:
Javier Fontan <jfontan at opennebula.org<mailto:jfontan at opennebula.org>> writes:
Which was the size of the original image? I think that when you do a
save_as (deferred disk snapshot) it just copies the size of the
original image to the new one.
I started with empty qcow2 disk of several virtual sizes, but on disk
they are all 196KB.
Regards.
--
Daniel Dehennin
Récupérer ma clef GPG: gpg --recv-keys 0xCC1E9E5B7A6FE2DF
Fingerprint: 3E69 014E 5C23 50E8 9ED6 2AAD CC1E 9E5B 7A6F E2DF
_______________________________________________
Users mailing list
Users at lists.opennebula.org<mailto:Users at lists.opennebula.org>
http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
--
Javier Fontán Muiños
Developer
OpenNebula - Flexible Enterprise Cloud Made Simple
www.OpenNebula.org<http://www.OpenNebula.org> | @OpenNebula | github.com/jfontan<http://github.com/jfontan>
_______________________________________________
Users mailing list
Users at lists.opennebula.org<mailto:Users at lists.opennebula.org>
http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
------------------------------------------------------------------
Steven C. Timm, Ph.D (630) 840-8525<tel:%28630%29%20840-8525>
timm at fnal.gov<mailto:timm at fnal.gov> http://home.fnal.gov/~timm/
Fermilab Scientific Computing Division, Scientific Computing Services Quad.
Grid and Cloud Services Dept., Associate Dept. Head for Cloud Computing
_______________________________________________
Users mailing list
Users at lists.opennebula.org<mailto:Users at lists.opennebula.org>
http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
--
Dr. Ruben Santiago Montero
Associate Professor (Profesor Titular), Complutense University of Madrid
URL: http://dsa-research.org/doku.php?id=people:ruben
Weblog: http://blog.dsa-research.org/?author=7
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20141008/71c18f32/attachment-0001.htm>
More information about the Users
mailing list