1. Proposed title of this feature request Enhance nova-scheduler to take into account total vcpus of instances each in nova database 2. Who is the customer behind the request? Account: ALCATEL-LUCENT 1554147 TAM customer: yes SRM customer: yes Strategic: yes 3. What is the nature and description of the request? Nova scheduler should take into account no. of instances & its cpu in a availability zone while starting an instance . This behaviour should avoid a scenario where addition of free cpus in all nodes in availability zone is equal to or larger than an instance to be started, but is not able to as no single node has the number of vcpu required by instance . 4. Why does the customer need this? (List the business requirements here) Customer has a set number of instances to be started in a zone , but if all instances with lower cpu count is started 1st , while spawning the instance with higher cpu , this issue is noticed . 4 nodes 5. How would the customer like to achieve this? (List the functional requirements here) nova scheduler should check the number of instance & cpus are already defined in db for the zone & avoid this scenarion . 6. For each functional requirement listed, specify how Red Hat and the customer can test to confirm the requirement is successfully implemented. In a zone of 4 16cpu nodes , spawn 5 4cpu instance & 5 8cpu instances . Any combination of spawn should not fail . 7. Is there already an existing RFE upstream or in Red Hat Bugzilla? 8. Does the customer have any specific timeline dependencies and which release would they like to target (i.e. RHEL5, RHEL6)? 9. Is the sales team involved in this request and do they have any additional input? 10. List any affected packages or components. nova 11. Would the customer be able to assist in testing this functionality if implemented?
Thanks for reporting the bug, but I need more clarification because you mix availability zones, database and vCPUs. By reading to your description, you'd like to see a scheduler behaviour where it would ensure that the total capacity would be fulfilled whatever the number of instances are ? Taking your example of 4 nodes with 16 pCPUs each, you would see the hosts stacked up by their number of free pCPUs for each ?
(In reply to Sylvain Bauza from comment #3) > Thanks for reporting the bug, but I need more clarification because you mix > availability zones, database and vCPUs. > > By reading to your description, you'd like to see a scheduler behaviour > where it would ensure that the total capacity would be fulfilled whatever > the number of instances are ? Yes . > > Taking your example of 4 nodes with 16 pCPUs each, you would see the hosts > stacked up by their number of free pCPUs for each ? Yes . Thanks .
This got classified as a bug for some reason, adjusting to RFE.