)]}'
{"playbooks/roles/os_nova/tasks/nova_post_install.yml":[{"author":{"_account_id":10343,"name":"Jim Rollenhagen","email":"jim@jimrollenhagen.com","username":"jimrollenhagen"},"change_message_id":"b70cb56f79871d8caa68a827961ceae662fb8de1","unresolved":false,"context_lines":[{"line_number":20,"context_line":"    nova_scheduler_host_manager: nova.scheduler.ironic_host_manager.IronicHostManager"},{"line_number":21,"context_line":"    nova_ram_allocation_ratio: 1.0"},{"line_number":22,"context_line":"    nova_reserved_host_memory_mb: 0"},{"line_number":23,"context_line":"    nova_compute_manager: ironic.nova.compute.manager.ClusteredComputeManager"},{"line_number":24,"context_line":"    nova_scheduler_use_baremetal_filters: True"},{"line_number":25,"context_line":"    nova_scheduler_tracks_instance_changes: False"},{"line_number":26,"context_line":"  when: nova_virt_type \u003d\u003d \u0027ironic\u0027"}],"source_content_type":"text/x-yaml","patch_set":1,"id":"bae84128_7a408f07","line":23,"updated":"2016-02-24 03:41:06.000000000","message":"I\u0027m not necessarily sure that we should recommend this; it isn\u0027t great. I\u0027m not sure what tradeoffs OSA is typically willing to make, but this tradeoff is a higher ops cost and higher chance of failure in return for HA. It also only really makes if we deploy multiple nova-compute services, with the same hostname on all of them.\n\nMaybe we make this optional?\n\nAlso, does this set these configs for all nova-compute hosts? That seems like it would preclude a kvm+ironic cloud, maybe that\u0027s okay to start with but ideally OSA could deploy a cloud with both.","commit_id":"c697d3f78b35443588df687a8be0542023bf1793"},{"author":{"_account_id":8125,"name":"Michael Davies","email":"michael@the-davies.net","username":"mrda"},"change_message_id":"c33b5adac28cdd1dd85ee4e81afe02ead755dcbb","unresolved":false,"context_lines":[{"line_number":20,"context_line":"    nova_scheduler_host_manager: nova.scheduler.ironic_host_manager.IronicHostManager"},{"line_number":21,"context_line":"    nova_ram_allocation_ratio: 1.0"},{"line_number":22,"context_line":"    nova_reserved_host_memory_mb: 0"},{"line_number":23,"context_line":"    nova_compute_manager: ironic.nova.compute.manager.ClusteredComputeManager"},{"line_number":24,"context_line":"    nova_scheduler_use_baremetal_filters: True"},{"line_number":25,"context_line":"    nova_scheduler_tracks_instance_changes: False"},{"line_number":26,"context_line":"  when: nova_virt_type \u003d\u003d \u0027ironic\u0027"}],"source_content_type":"text/x-yaml","patch_set":1,"id":"bae84128_da439bad","line":23,"in_reply_to":"bae84128_7a408f07","updated":"2016-02-24 03:54:27.000000000","message":"Thanks for that jroll.  I\u0027ll switch it around so the default will be ComputeManager for now, perhaps having a comment that ClusteredComputeManager is an option.\n\nAs for a potential kvm+ironic cloud, there\u0027s no way of specifying a multi nova-hypervisor cloud today, so it\u0027s all or nothing for now.  Potentially in the future (probably post Cells V2 availability) we could look into supporting multiple hypervisor drivers, once we have an architecture in place to better support this.","commit_id":"c697d3f78b35443588df687a8be0542023bf1793"}]}