Hi, all. I was thinking about the parallel filter scheduler problem these days. Main problem is the selection conflict among multiple filter scheduler instances. Now nova-scheduler can retry when the conflict happened. Someone said it is enough in practise to solve this problem, moreover, the conflict only impacts when the selected host's resource is exhaused. So I'd like to have your opinion on this issue. Thank you! 发自网易邮箱手机版 -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160329/0e4fd584/attachment.html>