Open Stack

Nova has stayed out of doing this auto-migrations so far.  You are requesting a very specific type of overallocation of resources, as in don't count machines that are shutoff when scheduling, but then automatically move them if they are turned back on.  It is an interesting idea, but not something we have discussed implementing.

You can generally over-allocate quite a bit, assuming you have swap enabled, and if some of your machines are not fully allocating their ram internally (which is usually the case) you should be fine. In deployments I have done when performance is really important, we simply do not over-allocate at all.

Keep in mind that the main use case of OpenStack is to run a cloud system, not a virtualization system. In a cloud, you very rarely stop machines and start them later, you generally delete the machine completely when you are done and launch a new one when you need it.  For this reason the particular use-case you are looking for hasn't really been considered.

Vish

On Apr 12, 2012, at 9:50 PM, Văn Đình Phúc wrote:

> Ignoring issues of Ram. I have a number of questions as the following model:
>  
> My node-1's resource as the following:
> CPU : 4 core
> RAM: 4 GB.
> Now , My Node-1 is hosting 4 instances:
> 
> 3 instance: (each VM: 1 core CPU, 2 GB ram,..). 
> 2 instance(i-00000001, i-00000002) is turn of and 1 instance(i-00000003) is running.
> 1 instance : (1 core CPU, 1 GB ram). 
> It is running.(i-00000004)
> I request new ínstance by the euca-run-instance command. My new instance ( i-00000005) is located on the node 1. This is the problem for the nova-scheduler.
> 
> What is happen ? If I reboot 2 instances : i-00000001,i-00000002.
> 
> How do we manage this problem in Openstack ?
> 
> I think that: My 2 instances (i-00000001 , i-00000002) should be migrated to the rest available resources as node-2 and node-3.
> 
> If not, it's actually very difficult to manage and reboot  the instances which are turned off. And their resources-allocated will be reused for the new instances.
> 
> ----- Thông điệp gốc : -----
> Gửi từ   : Vishvananda Ishaya [vishvananda at gmail.com]
> Gửi lúc  : 12/04/2012 11:43 PM
> Gửi tới  : openstack at lists.launchpad.net;vishvananda at gmail.com
> Chủ đề : Re: [Openstack] The Scheduling feature when reboot the instance
>  
> No this does not exist.  If you are going to overprovision ram, you should probably make sure that you have swap enabled on your host.
> 
> Vish
> 
> On Apr 12, 2012, at 4:10 AM, Văn Đình Phúc wrote:
> 
>> Hi.
>> I'm using Openstack (2011.3 (2011.3-nova-milestone-tarball:tarmac-20110922115702-k9nkvxqzhj130av2) with 3 Compute nodes.
>> My scheduling option is default.
>> I have the problem as the following steps:
>> Compute node 1 is installed the modules: nova-api, nova-compute and nova-network.
>> The resource of node-1 max is fitfor 3 instances (m1.small): i-00000001,   i-00000002, i-00000003.
>> I turn off  i-00000001.
>> and then I request new ínstance by the euca-run-instance command , My new instance ( i-00000004) is located on the node 1.
>> The resources: RAM , CPU,.. of node-2 and node-3 is available.
>> finally, I reboot the i-00000001.What is happen? the i-00000001 is rebooted on the node-1 , and It's shutdown off after reboot.I checked the libvirt's log :
>> 
>> "Failed to allocate 2147483648 B: Cannot allocate memory
>> 2012-04-12 16:11:50.530: shutting down"
>> So , Do Openstack have the sheduling feature when rebooting the instance ?
>> 
>> Example: As above , My instance (i-00000001) will be migrated to node-2 or node-3 when the resource of node-1 is not available.
>> 
>> <phucvdbmail.gif>
>>  
>> Văn Đình Phúc
>> Manager - Cloud Team
>> Techno Department
>> Bkav Specialization Division
>>  
>> Office:  Bkav Building - Yen Hoa New Town, Cau Giay Dist, Hanoi
>> Tel:       (84-4) 3763.2552
>> Mobile: 0169 4702 388 
>> ____________________________________________
>> Do your best, the rest will come !  
>> Hãy làm việc hết mình, những điều tốt đẹp sẽ đến với bạn !
>>  
>> Disclaimer: This e-mail and any files transmitted with it are confidential and may contain privileged information. It is intended solely for the use of the individual to whom it is addressed and others authorized to receive it. If you are not the intended recipient you are notified that disclosing, copying, distributing or taking any action in reliance on the contents of this information is strictly prohibited. If you have received this message in error, please notify the sender immediately by reply e-mail and delete completely this e-mail from your system, without reproducing, distributing or retaining copies.
>>  
>> _______________________________________________
>> Mailing list: https://launchpad.net/~openstack
>> Post to     : openstack at lists.launchpad.net
>> Unsubscribe : https://launchpad.net/~openstack
>> More help   : https://help.launchpad.net/ListHelp
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20120412/d7684a78/attachment.html>