[one-users] Ceph and thin provision

Campbell, Bill bcampbell at axcess-financial.com
Thu Dec 12 05:55:08 PST 2013


Kenneth, 
The allocation of images consuming the total image size looks to be a bug in Ceph: 

http://tracker.ceph.com/issues/6257 

They've identified, but doesn't look like there's been any movement on it since the bug was opened. 

----- Original Message -----

From: "Kenneth" <kenneth at apolloglobal.net> 
To: "Mario Giammarco" <mgiammarco at gmail.com> 
Cc: users at lists.opennebula.org 
Sent: Thursday, December 12, 2013 4:11:17 AM 
Subject: Re: [one-users] Ceph and thin provision 



I haven't tested much on non-persistent image as I have no use on them unless on experiments. Also, I haven't tried any volatile image, sorry. 

A not persistent image is writeable? 

Short answer: NO 

Long answer: Yes, sort of. When you instantiate a non persistent image, nebula create a "another disk" in the background temporarily. You can check that on when you issue "rbd ls -p one". You'll see something like this. 

one-34 -------> this is the non persistent image disk 
one-34-73-0 --------> this is the "temporary clone" of the disk when you instantiate a VM 
one-34-80-0 ---------> another VM which uses the non persistent image one-34 

This is why you can instantiate two or more VMs using a non-persistent image. If I'm not mistaken, the temporary disk will be destoyed once you shutdown the VM from nebula sunstone. But as long as the VM is running, the data is there. You can even reboot the VM with non-persistent disk and still have data. You lose the data once Nebula destroys VM disk, that is, when you SHUTDOWN or DELETE the VM from nebula sunstone. 

As for thick and thin provision, all of my images in ceph are thick, because my base image is 25 GB disk from a KVM template and then I imported it in ceph (it was converted from qcow2 to rbd). It consumes whole 25GB on my ceph storage. I just clone that "template image" every time I deploy a new VM. 

I haven't tried creating a thin or thick provision in ceph rbd from scratch. So basically, I can say that a 100GB disk will consume 100GB RBD in ceph (of course it will be 200GB in ceph storage since ceph duplicates the disks by default). 
--- 
Thanks,
Kenneth 
Apollo Global Corp. 


On 12/12/2013 04:52 PM, Mario Giammarco wrote: 


In several virtualization systems you can have a virtual disk drive: 

-thick, so a thick disk of 100gb uses 100gb of space; 
-thin, so a thin disk of 100gb uses 0gb when empty and starts using space when the virtual machine fills it. 

So I can have a real hdd of 250gb with inside ten virtual thin disks of 1000gb each, if they are almost empty. 
I have checked again and ceph rbd are "thin". 

BTW: I thank you for you explanation of persistent/not persistent, I was not able to find it in docs. Can you explain me also what a "volatile disk" is? 
A not persistent image is writeable? 
When you reboot a vm with a not persistent image you lose all datda written to it? 

Thanks again, 
Mario 


2013/12/12 Kenneth < kenneth at apolloglobal.net > 

<blockquote>



Hi, 

Can you elaborate more on what you want to achieve? 

If you have a 100GB image and it is set to persistent, you can instantiate that image immediately and deploy/live migrate it to any nebula node. Only one running instance of VM of this image is allowed. 

If it is a 100GB non persistent image, you'll have to wait for ceph to "create a copy" of it once you deploy it. But you can use this image multiple times simutaneously. 
--- 
Thanks,
Kenneth 
Apollo Global Corp. 


On 12/11/2013 07:28 PM, Mario Giammarco wrote: 

<blockquote>

Hello, 
I am using ceph with opennebula. 
I have created a 100gb disk image and I do not understand if it is thin or thick. 

I hope I can have thin provision. 

Thanks, 
Mario 
_______________________________________________
Users mailing list Users at lists.opennebula.org http://lists.opennebula.org/listinfo.cgi/users-opennebula.org 




</blockquote>


</blockquote>

_______________________________________________ 
Users mailing list 
Users at lists.opennebula.org 
http://lists.opennebula.org/listinfo.cgi/users-opennebula.org 


NOTICE: Protect the information in this message in accordance with the company's security policies. If you received this message in error, immediately notify the sender and destroy all copies.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opennebula.org/pipermail/users-opennebula.org/attachments/20131212/352027f2/attachment-0002.htm>


More information about the Users mailing list