[Openstack] oversubscribe node

Gangur, Hrushikesh (HP Converged Cloud - R&D - Sunnyvale) hrushikesh.gangur at hp.com
Wed Aug 21 22:45:15 UTC 2013

The Filter Scheduler (nova.scheduler.filter_scheduler.FilterScheduler) is the default scheduler for scheduling virtual machine instances. It supports filtering and weighting to make informed decisions on where a new instance should be created. This Scheduler can only be used for scheduling compute requests, not volume requests, i.e. it can only be used with the compute_scheduler_driver configuration option


Here is how it picks up the node (in the exact order):

1.            Check the disk space - in your it is tiny flavor so no size check is done

2.            Check the memory - it does allow you provision beyond 1.5x (default) of physical memory. For example, if you have 72 GB physical memory, and each tiny flavor takes 512 MB (0.5 GB), the no. of instances that can be provisioned must not be more than (72*1.5)/0.5 =216 instances

3.            Check the CPU - it does not allow you provision vCPUs beyond 16x (default) of the physical core no. For example, if you have 24 cores, the scheduler will not consider the host that has more than 16x24= 384 vCPUs. In case of tiny flavor, it maps to 384 instances (1 vcpu each).

For further reading: http://docs.openstack.org/developer/nova/devref/filter_scheduler.html

From: Diego Parrilla SantamarĂ­a [mailto:diego.parrilla.santamaria at gmail.com]
Sent: Wednesday, August 21, 2013 3:43 PM
To: Steve Heistand
Cc: openstack at lists.openstack.org
Subject: Re: [Openstack] oversubscribe node

Try to put those options in the nova.conf file that the nova-scheduler process reads, not in the nova-compute servers.


Diego Parrilla
www.stackops.com<http://www.stackops.com/> |  diego.parrilla at stackops.com<mailto:diego.parrilla at stackops.com> | +34 649 94 43 29 | skype:diegoparrilla


On Wed, Aug 21, 2013 at 11:56 PM, Steve Heistand <steve.heistand at nasa.gov<mailto:steve.heistand at nasa.gov>> wrote:
Hi folks,

so  Im a little confused, on my compute nodes I have the allocation ratio set:

root at node004:~# cat /etc/nova/nova-compute.conf

but I got into a state where the node is oversubscribed.  From nova hypervisor-show node004:

vcpus_used              36
memory_mb_used  72837
memory_mb            64379
vcpus                       32

I have zones/aggregates set up and of the instances, one was started in just launch it anywhere mode and another was told to start on
a zone that is this node.

In looking back I would have expected the zone specified instance be rejected.

is this to be expected or a bug?



 Steve Heistand                           NASA Ames Research Center
 SciCon Group                             Mail Stop 258-6
 steve.heistand at nasa.gov<mailto:steve.heistand at nasa.gov>  (650) 604-4369  Moffett Field, CA 94035-1000
 "Any opinions expressed are those of our alien overlords, not my own."

# For Remedy                        #
#Action: Resolve                    #
#Resolution: Resolved               #
#Reason: No Further Action Required #
#Tier1: User Code                   #
#Tier2: Other                       #
#Tier3: Assistance                  #
#Notification: None                 #

Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
Post to     : openstack at lists.openstack.org<mailto:openstack at lists.openstack.org>
Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130821/9d8ecddb/attachment.html>

More information about the Openstack mailing list