Jump to: navigation, search

Difference between revisions of "Meetings/NovaScheduler"

(Agenda for next meeting)
(Agenda for next meeting)
Line 17: Line 17:
 
** Reviews
 
** Reviews
 
*** Nested RPs:
 
*** Nested RPs:
**** ComputeDriver.update_provider_tree() series starting with: https://review.openstack.org/#/c/521685/ (efried) (See open discussion topic)
+
**** ComputeDriver.update_provider_tree() series starting with: https://review.openstack.org/#/c/521685/ (efried) (See open discussion topics)
**** Nested affordance in GET /allocation_candidates (jaypipes)
+
**** Nested affordance in GET /allocation_candidates series starting with: https://review.openstack.org/#/c/531443/1 (jaypipes)
 
**** Granular resource requests (efried): https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/granular-resource-requests (waiting for ^^)
 
**** Granular resource requests (efried): https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:bp/granular-resource-requests (waiting for ^^)
 
*** Alternate hosts (edleafe):
 
*** Alternate hosts (edleafe):

Revision as of 13:11, 8 January 2018

Weekly Nova Scheduler team meeting

MEETING TIME: Mondays 14:00 UTC (#openstack-meeting-alt)

This meeting is a weekly gathering of developers working on the Nova Scheduler subteam. We cover topics such as development focus, status, bugs, reviews, and other current topics worthy of real-time discussion.

NOTE: this wiki page should be 'emptied' at the end of each meeting.

Next meetings scheduled for:

Agenda for next meeting

  • Open discussion
    • Using /allocation_candidates?limit= from nova-side, should we? how?
    • sharing providers in ProviderTree: https://review.openstack.org/#/c/526539/
      • If sharing providers are managed by compute nodes, how does coordination work? Answer: careful handling of conflicts
        • Generation support needed for PUT /rp/{uuid}/aggregates
        • Sensible reGET/retry on 409s
      • Can we centralize all that logic in the report client and have it work for all cases, or do virt drivers need more control? (Related to below.)
    • Why can't (in-tree) virt drivers talk to placement? See conversation seeds in patch-level comments on https://review.openstack.org/#/c/526539/7

Previous meetings