Ask Your Question
0

openstack assign bigger resources than exist in compute nodes to instances

asked 2015-06-23 02:09:41 -0500

Ams70 gravatar image

updated 2015-06-23 02:23:27 -0500

Hi,

I run OpenStack in 4 nodes: a controller node, a network node and two compute node. My first compute node have 7GB hard and 2GB ram and 1 core, my second compute node have 34GB hard and 5GB ram and 6 core. I want to create a instance that openstack allocate it the whole volume of both compute nodes, so I create a custom Flavors and define:

VCPUs:2 , RAM (MB):6048 , Root Disk (GB):2 , Ephemeral Disk (GB):39 and Swap Disk (MB):100.

When I‌ create instance and chose this flavor, after build it I see that the whole configure part was allocated from second compute node. It means that openstack assign amount of resources that bigger than the really resource that exist in second compute node.

I also create 4 instance that perched in first compute node and have totally bigger resources from really resource of first compute node.

Does this thing was mistake? if yes, How can I mange and solve this problem.

Thanks a lot in advance.

edit retag flag offensive close merge delete

1 answer

Sort by » oldest newest most voted
0

answered 2015-06-23 11:05:21 -0500

mpetason gravatar image

Instances will not span multiple hypervisors. You will be able to launch an instance to one hypervisor. This means that you are limited on what you can do with the compute node that has 2GB ram, you won't be able to do much. You'll probably be able to launch 1 or 2 micro VMs. On the other node you'll be able to launch two small VMs.

Regarding the amount of resources you are able to provision it will be based on your oversubscription. By Default you should be able to use:

VCPU : 16:1 Ram : 1:1.5 HD : 1:1

This means you can allocate up to 1.5 times the amount of memory you currently have. Memory is usually the limiting factor, so this is where you'll want to look for scheduling.

Search for ram_allocation_ratio:

http://docs.openstack.org/kilo/config...

edit flag offensive delete link more

Comments

Thanks for your explanation, in my first compute node that have 2GB ram, i lunch 3 VMs that have in total 2.5 GB ram, 9GB hard and 3 cores while this compute node have 2GB ram, 7GB hard and one core. all the nodes work fine without any error. according to your explanation this is a counterexample.

Ams70 gravatar imageAms70 ( 2015-06-23 23:41:03 -0500 )edit

Read the documentation and attempt the math and you'll see what your limits are.

mpetason gravatar imagempetason ( 2015-06-24 09:23:30 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2015-06-23 02:09:41 -0500

Seen: 9,770 times

Last updated: Jun 23 '15