yahoo-eng-team team mailing list archive

[Sylvain Bauza <sbauza@xxxxxxx>]

I feel that due to the design of the FilterScheduler that makes every filter independent of the others, it's necessarly up to the operator to make sure that if they need to use both AggregateCoreFilter and/or AggregateRamFilter *and* NUMATopologyFilter for NUMA placement, they necessarly need to make sure that both values for CPU and RAM are consistent between the aggregate and each compute node. 
That surely defeats the purpose of the AggregateCore/RAMFilter to prevent defining ratios for every compute, but since NUMATopologyFilter requires that, I don't see how we could improve that, except creating a chain of dependency which I dislike.

** Changed in: nova
       Status: New => Won't Fix

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1748858

Title:
  allocation_ratio set by aggregate metadata don't work in
  NUMATopologyFilter

Status in OpenStack Compute (nova):
  Won't Fix

Bug description:
  Description
  ===========
  Allocation_ratio setted by aggregate metadata don't work in NUMATopologyFilter. NUMATopologyFilter always takes the allocation_ratio from the compute node's configuration file. 
  We do not provide a consistent way to set the allocation_ratio when user use the AggregateCoreFilter, AggregateRamFilter and NUMATopologyFilter at same time.

  
  Environment
  ===========

  nova16.0.4

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1748858/+subscriptions