)]}'
{"specs/mitaka/approved/resource-providers-allocations.rst":[{"author":{"_account_id":7,"name":"Jay Pipes","email":"jaypipes@gmail.com","username":"jaypipes"},"change_message_id":"8788f64a8c61eb73745dfa00ef4eed0460b690e1","unresolved":false,"context_lines":[{"line_number":8,"context_line":"Resource Providers - Compute Node Allocations"},{"line_number":9,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":10,"context_line":""},{"line_number":11,"context_line":"https://blueprints.launchpad.net/nova/+spec/compute-node-allocations"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"This blueprint specification explains the process of migrating the data that"},{"line_number":14,"context_line":"stores allocated/assigned resource amounts from the older schema to the new"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7af24918_dfedf8f0","line":11,"updated":"2016-03-06 17:01:00.000000000","message":"Done.","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":6873,"name":"Matt Riedemann","email":"mriedem.os@gmail.com","username":"mriedem"},"change_message_id":"13d81e159daed94f455341d7c690bf2af8bbd80e","unresolved":false,"context_lines":[{"line_number":8,"context_line":"Resource Providers - Compute Node Allocations"},{"line_number":9,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":10,"context_line":""},{"line_number":11,"context_line":"https://blueprints.launchpad.net/nova/+spec/compute-node-allocations"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"This blueprint specification explains the process of migrating the data that"},{"line_number":14,"context_line":"stores allocated/assigned resource amounts from the older schema to the new"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3a57f1b5_98826c93","line":11,"range":{"start_line":11,"start_character":44,"end_line":11,"end_character":68},"updated":"2016-02-11 15:56:46.000000000","message":"This doesn\u0027t exist, but you probably want to name it resource-providers-allocations anyway to be consistent with this spec name.","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":7,"name":"Jay Pipes","email":"jaypipes@gmail.com","username":"jaypipes"},"change_message_id":"8788f64a8c61eb73745dfa00ef4eed0460b690e1","unresolved":false,"context_lines":[{"line_number":146,"context_line":"    compute node, but won\u0027t tell you which instance is pinned to which CPU on a"},{"line_number":147,"context_line":"    compute node. Likewise, the `allocations` table isn\u0027t intended to tell you"},{"line_number":148,"context_line":"    what fraction of disk resources is swap, what is ephemeral and what is used"},{"line_number":149,"context_line":"    for root disks. Disk resources are disk resources."},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"In online object migrations, we will transform fields in the `compute_nodes`"},{"line_number":152,"context_line":"and `instance_extra` tables that store resource allocation amounts to instead"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7af24918_3fb97ced","line":149,"updated":"2016-03-06 17:01:00.000000000","message":"Done.","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":782,"name":"John Garbutt","email":"john@johngarbutt.com","username":"johngarbutt"},"change_message_id":"a53f82bd581e9ec8fa14361005058713a5c5ea17","unresolved":false,"context_lines":[{"line_number":146,"context_line":"    compute node, but won\u0027t tell you which instance is pinned to which CPU on a"},{"line_number":147,"context_line":"    compute node. Likewise, the `allocations` table isn\u0027t intended to tell you"},{"line_number":148,"context_line":"    what fraction of disk resources is swap, what is ephemeral and what is used"},{"line_number":149,"context_line":"    for root disks. Disk resources are disk resources."},{"line_number":150,"context_line":""},{"line_number":151,"context_line":"In online object migrations, we will transform fields in the `compute_nodes`"},{"line_number":152,"context_line":"and `instance_extra` tables that store resource allocation amounts to instead"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7a5de9d1_d051f8d9","line":149,"updated":"2016-02-02 18:33:16.000000000","message":"we probably need to define where those specific are persisted, probably the stay in instance_extra?","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":7,"name":"Jay Pipes","email":"jaypipes@gmail.com","username":"jaypipes"},"change_message_id":"8788f64a8c61eb73745dfa00ef4eed0460b690e1","unresolved":false,"context_lines":[{"line_number":171,"context_line":"   value shall be the instance\u0027s `uuid` field value. The `resource_provider_id`"},{"line_number":172,"context_line":"   field value shall be the compute node\u0027s `uuid` field value. The"},{"line_number":173,"context_line":"   `resource_class_id` field shall contain the identifier for the appropriate"},{"line_number":174,"context_line":"   resource class (either RAM or vCPU)"},{"line_number":175,"context_line":""},{"line_number":176,"context_line":"4) Insert **one** record for the DISK resource class. The amount used should be"},{"line_number":177,"context_line":"   the sum of the `root_gb`, `ephemeral_gb`, and `swap` field values of the"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7af24918_1fbe40d4","line":174,"updated":"2016-03-06 17:01:00.000000000","message":"@johnthetubaguy: Yeah, I\u0027m *really* not keen to add overhead here. The XenAPI driver is the only driver that calculates overhead for each instance, and it\u0027s really wonky how it\u0027s done. I think having a reserved and an allocation_ratio for each resource class in the inventories table provides all the necessary knobs and gizmos needed to effectively track resource constraints. Adding custom overhead fields for a single virt driver isn\u0027t something I\u0027m keen to do.","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":782,"name":"John Garbutt","email":"john@johngarbutt.com","username":"johngarbutt"},"change_message_id":"a53f82bd581e9ec8fa14361005058713a5c5ea17","unresolved":false,"context_lines":[{"line_number":171,"context_line":"   value shall be the instance\u0027s `uuid` field value. The `resource_provider_id`"},{"line_number":172,"context_line":"   field value shall be the compute node\u0027s `uuid` field value. The"},{"line_number":173,"context_line":"   `resource_class_id` field shall contain the identifier for the appropriate"},{"line_number":174,"context_line":"   resource class (either RAM or vCPU)"},{"line_number":175,"context_line":""},{"line_number":176,"context_line":"4) Insert **one** record for the DISK resource class. The amount used should be"},{"line_number":177,"context_line":"   the sum of the `root_gb`, `ephemeral_gb`, and `swap` field values of the"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7a5de9d1_702b6434","line":174,"updated":"2016-02-02 18:33:16.000000000","message":"maybe should also include the estimated overhead, possibly. Similar issue with disk overhead (I think hyper-v is adding that one).","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":782,"name":"John Garbutt","email":"john@johngarbutt.com","username":"johngarbutt"},"change_message_id":"a53f82bd581e9ec8fa14361005058713a5c5ea17","unresolved":false,"context_lines":[{"line_number":274,"context_line":"---------------------"},{"line_number":275,"context_line":""},{"line_number":276,"context_line":"We could add a `nova-manage compute-node-allocations-migrate` command that"},{"line_number":277,"context_line":"could be run by deployers to batch-process the compute node migrations."},{"line_number":278,"context_line":""},{"line_number":279,"context_line":"Developer impact"},{"line_number":280,"context_line":"----------------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7a5de9d1_10cb1074","line":277,"updated":"2016-02-02 18:33:16.000000000","message":"actually, I think we should create a new generic \"nova-manage complete-online-data-migrations --batch-size\u003d10\" sort of thing.","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":7,"name":"Jay Pipes","email":"jaypipes@gmail.com","username":"jaypipes"},"change_message_id":"8788f64a8c61eb73745dfa00ef4eed0460b690e1","unresolved":false,"context_lines":[{"line_number":308,"context_line":"  `instance_extra` table, and read the resource allocation information from the"},{"line_number":309,"context_line":"  `allocations` table instead of the `instance_extra` table"},{"line_number":310,"context_line":"* Full functional integration tests added which stress the migration paths and"},{"line_number":311,"context_line":"  ensure backwards compatibility"},{"line_number":312,"context_line":""},{"line_number":313,"context_line":"Dependencies"},{"line_number":314,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7af24918_7fb3040b","line":311,"updated":"2016-03-06 17:01:00.000000000","message":"@johnthetubaguy: Nothing about the claim process is proposed to change in this blueprint. This is simply changing the target field storage for allocation amounts on the Instance object.","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"},{"author":{"_account_id":782,"name":"John Garbutt","email":"john@johngarbutt.com","username":"johngarbutt"},"change_message_id":"a53f82bd581e9ec8fa14361005058713a5c5ea17","unresolved":false,"context_lines":[{"line_number":308,"context_line":"  `instance_extra` table, and read the resource allocation information from the"},{"line_number":309,"context_line":"  `allocations` table instead of the `instance_extra` table"},{"line_number":310,"context_line":"* Full functional integration tests added which stress the migration paths and"},{"line_number":311,"context_line":"  ensure backwards compatibility"},{"line_number":312,"context_line":""},{"line_number":313,"context_line":"Dependencies"},{"line_number":314,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":1,"id":"7a5de9d1_90a5c0b6","line":311,"updated":"2016-02-02 18:33:16.000000000","message":"But.. we still need to do the specific resource reservations in the resource tracker. I can\u0027t quite visualize how that changes in here.","commit_id":"60b9e4e778db1a802ff1e453eaafc674f649ab26"}],"specs/newton/approved/resource-providers-allocations.rst":[{"author":{"_account_id":1063,"name":"Ed Leafe","email":"ed@leafe.com","username":"ed-leafe"},"change_message_id":"2c6e817e934011c2755d737f82fc3aa0649d2ebf","unresolved":false,"context_lines":[{"line_number":179,"context_line":"   flavor. The `resource_provider_id` value of the record shall be the compute"},{"line_number":180,"context_line":"   node\u0027s resource provider ID *or* the generic resource pool\u0027s provider ID if"},{"line_number":181,"context_line":"   the compute node is associated to an aggregate that is tied to a resource"},{"line_number":182,"context_line":"   pool"},{"line_number":183,"context_line":""},{"line_number":184,"context_line":"5) If the `instance_extra.numa_topology` is not None, insert one record in the"},{"line_number":185,"context_line":"   `allocations` table for the `NUMA_SOCKETS`, `NUMA_CORES`, and `NUMA_THREADS`"}],"source_content_type":"text/x-rst","patch_set":4,"id":"fa0719c6_b813b5b7","line":182,"updated":"2016-03-22 16:38:45.000000000","message":"Don\u0027t we also need to specify the `consumer_id` value for this record (as well as the NUMA and PCI records) as the instance UUID?","commit_id":"84c58abd1659db0c7ea9d436df2d5d9cbd6cea9f"},{"author":{"_account_id":7,"name":"Jay Pipes","email":"jaypipes@gmail.com","username":"jaypipes"},"change_message_id":"f0051ff58d33d2f4e15ae17da9bf5c6acfd2e427","unresolved":false,"context_lines":[{"line_number":310,"context_line":"  `allocations` table instead of the `instance_extra` table"},{"line_number":311,"context_line":"* Full functional integration tests added which stress the migration paths and"},{"line_number":312,"context_line":"  ensure backwards compatibility"},{"line_number":313,"context_line":""},{"line_number":314,"context_line":"Dependencies"},{"line_number":315,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":316,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"1af94dfe_ac982f6b","line":313,"updated":"2016-03-18 13:04:59.000000000","message":"Hi Yingxin! Sorry for the delayed response here!\n\nThe code in the RT that handles orphaned instances and adjusts inventory/capacity counts would not really change in this blueprint. All we are doing in this blueprint is changing the backend storage locations of various allocation amounts. So instead of storing the allocated amount of RAM in instance_extra.flavor[old].memory_mb we are storing that in the allocations table as a record with key of (compute_node.uuid, RAM_MB, instance.uuid).\n\nThe ComputeNode object is a facade that will allow the RT to continue setting fields like ComputeNode.memory_mb, but when the RT calls ComputeNode.save(), instead of saving that value into compute_nodes.memory_mb, it saves the record in the inventories table with a key (compute_node.uuid, RAM_MB).","commit_id":"84c58abd1659db0c7ea9d436df2d5d9cbd6cea9f"},{"author":{"_account_id":15648,"name":"Yingxin","email":"yingxin.cheng@intel.com","username":"Yingxin"},"change_message_id":"103e2203a51b161158a7086ae732ec4c436d10f4","unresolved":false,"context_lines":[{"line_number":310,"context_line":"  `allocations` table instead of the `instance_extra` table"},{"line_number":311,"context_line":"* Full functional integration tests added which stress the migration paths and"},{"line_number":312,"context_line":"  ensure backwards compatibility"},{"line_number":313,"context_line":""},{"line_number":314,"context_line":"Dependencies"},{"line_number":315,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":316,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"3afc51ec_004846cb","line":313,"updated":"2016-03-15 08:27:29.000000000","message":"RT has an ability to detect orphan instances[1] and adjust related resource usage according to the untracked instance uuids. How to apply this logic to resource provider?\n\nA doable solution is to create allocations to those orphan instances and update them periodically. An alternative way(maybe better) is to update the reserved field of compute node inventories periodically to reflect the resource consumption of those instances.\n\n\n[1] https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L572-L573","commit_id":"84c58abd1659db0c7ea9d436df2d5d9cbd6cea9f"},{"author":{"_account_id":15648,"name":"Yingxin","email":"yingxin.cheng@intel.com","username":"Yingxin"},"change_message_id":"282ff52ac4c374b47a28491bac7888507ea76f02","unresolved":false,"context_lines":[{"line_number":310,"context_line":"  `allocations` table instead of the `instance_extra` table"},{"line_number":311,"context_line":"* Full functional integration tests added which stress the migration paths and"},{"line_number":312,"context_line":"  ensure backwards compatibility"},{"line_number":313,"context_line":""},{"line_number":314,"context_line":"Dependencies"},{"line_number":315,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":316,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"fa0719c6_4259eafd","line":313,"in_reply_to":"1af94dfe_ac982f6b","updated":"2016-03-22 01:39:53.000000000","message":"Yes, I mean in current implementation[1], orphan usage will be calculated into `compute_nodes.memory_mb_used`, which is considered untrusted in this bp(@69). So I think the \"orphaned instances handling\" and \"overhead handling\" should be changed in this bp(or maybe in the previous bp compute-node-inventory), or the accurate \"available resources\" cannot be calculated simply from inventories and allocations from db.\n\n[1] https://github.com/openstack/nova/blob/master/nova/compute/resource_tracker.py#L961","commit_id":"84c58abd1659db0c7ea9d436df2d5d9cbd6cea9f"},{"author":{"_account_id":15648,"name":"Yingxin","email":"yingxin.cheng@intel.com","username":"Yingxin"},"change_message_id":"d9e82fc0b09527d632f19110e77bc1deca7cee4d","unresolved":false,"context_lines":[{"line_number":310,"context_line":"  `allocations` table instead of the `instance_extra` table"},{"line_number":311,"context_line":"* Full functional integration tests added which stress the migration paths and"},{"line_number":312,"context_line":"  ensure backwards compatibility"},{"line_number":313,"context_line":""},{"line_number":314,"context_line":"Dependencies"},{"line_number":315,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":316,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"fa0719c6_a7f53bcc","line":313,"in_reply_to":"fa0719c6_25c66a49","updated":"2016-03-23 01:54:01.000000000","message":"reasonable","commit_id":"84c58abd1659db0c7ea9d436df2d5d9cbd6cea9f"},{"author":{"_account_id":7,"name":"Jay Pipes","email":"jaypipes@gmail.com","username":"jaypipes"},"change_message_id":"185b65065cccb63abb0b17dbf68a036e14feedfa","unresolved":false,"context_lines":[{"line_number":310,"context_line":"  `allocations` table instead of the `instance_extra` table"},{"line_number":311,"context_line":"* Full functional integration tests added which stress the migration paths and"},{"line_number":312,"context_line":"  ensure backwards compatibility"},{"line_number":313,"context_line":""},{"line_number":314,"context_line":"Dependencies"},{"line_number":315,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":316,"context_line":""}],"source_content_type":"text/x-rst","patch_set":4,"id":"fa0719c6_25c66a49","line":313,"in_reply_to":"fa0719c6_4259eafd","updated":"2016-03-22 13:12:21.000000000","message":"@Yingxin, that field won\u0027t be set any more by the resource tracker. It will only be *read* and then compared to what the hypervisor reports is used on the node. There is an inventories.reserved field that is used to handle any mismatch of the hypervisor\u0027s view of resources and the scheduler\u0027s.","commit_id":"84c58abd1659db0c7ea9d436df2d5d9cbd6cea9f"},{"author":{"_account_id":11564,"name":"Chris Dent","email":"cdent@anticdent.org","username":"chdent"},"change_message_id":"6ff60ed923cbf9e552f73f2074b055353f7b4ec9","unresolved":false,"context_lines":[{"line_number":357,"context_line":"updated. Useful to let reader understand what\u0027s happened along the"},{"line_number":358,"context_line":"time."},{"line_number":359,"context_line":""},{"line_number":360,"context_line":".. list-table:: Revisions"},{"line_number":361,"context_line":"   :header-rows: 1"},{"line_number":362,"context_line":""},{"line_number":363,"context_line":"   * - Release Name"}],"source_content_type":"text/x-rst","patch_set":4,"id":"fa0719c6_60251faa","line":360,"updated":"2016-03-21 15:41:34.000000000","message":"I\u0027ve never understood this section. Is this now out of date with the move under newton?","commit_id":"84c58abd1659db0c7ea9d436df2d5d9cbd6cea9f"}]}
