)]}'
{"/COMMIT_MSG":[{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":6,"context_line":""},{"line_number":7,"context_line":"Libvirt: add vPMU spec for train"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This spec adds a way for tenants to contol"},{"line_number":10,"context_line":"the emulation of a virtual performacne monitoring"},{"line_number":11,"context_line":"unit when using the libvirt virt dirver."},{"line_number":12,"context_line":""}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"9fb8cfa7_e6c7c011","line":9,"range":{"start_line":9,"start_character":36,"end_line":9,"end_character":42},"updated":"2019-06-24 21:59:28.000000000","message":"control","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":7,"context_line":"Libvirt: add vPMU spec for train"},{"line_number":8,"context_line":""},{"line_number":9,"context_line":"This spec adds a way for tenants to contol"},{"line_number":10,"context_line":"the emulation of a virtual performacne monitoring"},{"line_number":11,"context_line":"unit when using the libvirt virt dirver."},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"Change-Id: Id759c7fb42b878bc0bb237558aa2e9548d995320"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"9fb8cfa7_06cd3430","line":10,"range":{"start_line":10,"start_character":27,"end_line":10,"end_character":38},"updated":"2019-06-24 21:59:28.000000000","message":"performance","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":9,"context_line":"This spec adds a way for tenants to contol"},{"line_number":10,"context_line":"the emulation of a virtual performacne monitoring"},{"line_number":11,"context_line":"unit when using the libvirt virt dirver."},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"Change-Id: Id759c7fb42b878bc0bb237558aa2e9548d995320"}],"source_content_type":"text/x-gerrit-commit-message","patch_set":2,"id":"9fb8cfa7_f13778af","line":12,"updated":"2019-06-24 21:59:28.000000000","message":"add blueprint tag","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"}],"specs/train/approved/libvirt-pmu-configuration.rst":[{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":8,"context_line":"Libvirt: tenant contol of qemu performance monitoring unit (vPMU)"},{"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\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":"Include the URL of your launchpad blueprint:"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"https://blueprints.launchpad.net/nova/+spec/libvirt-disable-pmu-for-realtime-guests"},{"line_number":14,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_a6cd4831","line":11,"range":{"start_line":11,"start_character":0,"end_line":11,"end_character":44},"updated":"2019-06-24 21:59:28.000000000","message":"remove boilerplate","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":10,"context_line":""},{"line_number":11,"context_line":"Include the URL of your launchpad blueprint:"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"https://blueprints.launchpad.net/nova/+spec/libvirt-disable-pmu-for-realtime-guests"},{"line_number":14,"context_line":""},{"line_number":15,"context_line":"qemu/kvm supports emulation of a vPMU to enable standard performance monitoring"},{"line_number":16,"context_line":"tools such as perf to be used within a virtualisation environment. The vPMU"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_714b882a","line":13,"range":{"start_line":13,"start_character":0,"end_line":13,"end_character":83},"updated":"2019-06-24 21:59:28.000000000","message":"this doesn\u0027t match the file name","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":21,"context_line":"see https://libvirt.org/formatdomain.html#elementsFeatures"},{"line_number":22,"context_line":"so this feature is available in nova minimum supported libvirt of 1.3.1."},{"line_number":23,"context_line":""},{"line_number":24,"context_line":"This spec aims to allow tenants to contol when the vPMU is enabled."},{"line_number":25,"context_line":""},{"line_number":26,"context_line":""},{"line_number":27,"context_line":"Problem description"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_9195dc89","line":24,"range":{"start_line":24,"start_character":35,"end_line":24,"end_character":41},"updated":"2019-06-24 21:59:28.000000000","message":"control","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":30,"context_line":"While kvm/qemu support for a vPMU is generally a useful feature the requirement"},{"line_number":31,"context_line":"to collect and maintain virtual performance counter introduces additional"},{"line_number":32,"context_line":"latency of ~10us which is about 1% of the total budget for 5G  end to end"},{"line_number":33,"context_line":"traffic processing latency. While this might seem small its an appreciable"},{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_518be466","line":33,"range":{"start_line":33,"start_character":56,"end_line":33,"end_character":59},"updated":"2019-06-24 21:59:28.000000000","message":"it is","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":31,"context_line":"to collect and maintain virtual performance counter introduces additional"},{"line_number":32,"context_line":"latency of ~10us which is about 1% of the total budget for 5G  end to end"},{"line_number":33,"context_line":"traffic processing latency. While this might seem small its an appreciable"},{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"As the provision of a vPMU is not currently controllable by an operator"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_f1e9b819","line":34,"range":{"start_line":34,"start_character":56,"end_line":34,"end_character":58},"updated":"2019-06-24 21:59:28.000000000","message":"and","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":32,"context_line":"latency of ~10us which is about 1% of the total budget for 5G  end to end"},{"line_number":33,"context_line":"traffic processing latency. While this might seem small its an appreciable"},{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"As the provision of a vPMU is not currently controllable by an operator"},{"line_number":38,"context_line":"or tenant directly this creates a problem for those that want to enable"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_11efac1e","line":35,"range":{"start_line":35,"start_character":23,"end_line":35,"end_character":30},"updated":"2019-06-24 21:59:28.000000000","message":"achieving","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":36,"context_line":""},{"line_number":37,"context_line":"As the provision of a vPMU is not currently controllable by an operator"},{"line_number":38,"context_line":"or tenant directly this creates a problem for those that want to enable"},{"line_number":39,"context_line":"or disable the vPMU to either avoid the latency overhaed or rely on it"},{"line_number":40,"context_line":"to monitor the performace of their workload."},{"line_number":41,"context_line":""},{"line_number":42,"context_line":"Use Cases"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_71ddc86e","line":39,"range":{"start_line":39,"start_character":48,"end_line":39,"end_character":56},"updated":"2019-06-24 21:59:28.000000000","message":"overhead","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":43,"context_line":"---------"},{"line_number":44,"context_line":""},{"line_number":45,"context_line":"As a telecoms operator building a 5G network I wish to be able to deploy"},{"line_number":46,"context_line":"a virtualias Radio access network (vRAN) applicance with minimal latency"},{"line_number":47,"context_line":"impact form my virtualiation stack."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"As a tenant I wish to be able to monitor the performace of my application"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_91dabc75","line":46,"range":{"start_line":46,"start_character":2,"end_line":46,"end_character":12},"updated":"2019-06-24 21:59:28.000000000","message":"virtualized?","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":44,"context_line":""},{"line_number":45,"context_line":"As a telecoms operator building a 5G network I wish to be able to deploy"},{"line_number":46,"context_line":"a virtualias Radio access network (vRAN) applicance with minimal latency"},{"line_number":47,"context_line":"impact form my virtualiation stack."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"As a tenant I wish to be able to monitor the performace of my application"},{"line_number":50,"context_line":"using standard tools like perf in a virtualized enviornment to enable"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_31e3d036","line":47,"range":{"start_line":47,"start_character":7,"end_line":47,"end_character":11},"updated":"2019-06-24 21:59:28.000000000","message":"from","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":44,"context_line":""},{"line_number":45,"context_line":"As a telecoms operator building a 5G network I wish to be able to deploy"},{"line_number":46,"context_line":"a virtualias Radio access network (vRAN) applicance with minimal latency"},{"line_number":47,"context_line":"impact form my virtualiation stack."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"As a tenant I wish to be able to monitor the performace of my application"},{"line_number":50,"context_line":"using standard tools like perf in a virtualized enviornment to enable"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_51e0c426","line":47,"range":{"start_line":47,"start_character":15,"end_line":47,"end_character":28},"updated":"2019-06-24 21:59:28.000000000","message":"virtualization","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"592e66baf9db41d22a0bb7c7798b21e88e700f01","unresolved":false,"context_lines":[{"line_number":47,"context_line":"impact form my virtualiation stack."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"As a tenant I wish to be able to monitor the performace of my application"},{"line_number":50,"context_line":"using standard tools like perf in a virtualized enviornment to enable"},{"line_number":51,"context_line":"development, tuning, and profiling of my application."},{"line_number":52,"context_line":""},{"line_number":53,"context_line":"Proposed change"}],"source_content_type":"text/x-rst","patch_set":2,"id":"9fb8cfa7_f1d2185c","line":50,"range":{"start_line":50,"start_character":48,"end_line":50,"end_character":59},"updated":"2019-06-24 21:59:28.000000000","message":"environment","commit_id":"e295f32ae3e900b8e5a6b4934066cc06569d5cc9"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"042b07e4d57177ca385344dfe064cee9c4799dfe","unresolved":false,"context_lines":[{"line_number":8,"context_line":"Libvirt: tenant control of qemu performance monitoring unit (vPMU)"},{"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\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/libvirt-disable-pmu-for-realtime-guests"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"qemu/kvm supports emulation of a vPMU to enable standard performance monitoring"},{"line_number":14,"context_line":"tools such as perf to be used within a virtualisation environment. The vPMU"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9fb8cfa7_5171a443","line":11,"range":{"start_line":11,"start_character":44,"end_line":11,"end_character":83},"updated":"2019-06-24 22:09:10.000000000","message":"This doesn\u0027t match the file name. (At this point are we going to rename the blueprint?)","commit_id":"0904fd1a6ef3b9ed40de040cbc595edcfcc4b9f2"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"042b07e4d57177ca385344dfe064cee9c4799dfe","unresolved":false,"context_lines":[{"line_number":11,"context_line":"https://blueprints.launchpad.net/nova/+spec/libvirt-disable-pmu-for-realtime-guests"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"qemu/kvm supports emulation of a vPMU to enable standard performance monitoring"},{"line_number":14,"context_line":"tools such as perf to be used within a virtualisation environment. The vPMU"},{"line_number":15,"context_line":"which is available on x86 cpus emulates the hardware PMU found on Intel"},{"line_number":16,"context_line":"processors and was introduced in kvm in kernel 3.3.1."},{"line_number":17,"context_line":""}],"source_content_type":"text/x-rst","patch_set":3,"id":"9fb8cfa7_11956c33","line":14,"range":{"start_line":14,"start_character":14,"end_line":14,"end_character":18},"updated":"2019-06-24 22:09:10.000000000","message":"be nice if this was a link","commit_id":"0904fd1a6ef3b9ed40de040cbc595edcfcc4b9f2"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"042b07e4d57177ca385344dfe064cee9c4799dfe","unresolved":false,"context_lines":[{"line_number":72,"context_line":""},{"line_number":73,"context_line":"In addition to the above minimum changes the libvirt driver could be modifed"},{"line_number":74,"context_line":"to report support for a vPMU to improve scheduling. As the vPMU feature"},{"line_number":75,"context_line":"is supported by Nova\u0027s minimium required qemu/libvirt this would only be"},{"line_number":76,"context_line":"useful in a heterogeneous cloud. As such the desicion to expose this"},{"line_number":77,"context_line":"feature as a trait is left to the implementation and will be enabled as part"},{"line_number":78,"context_line":"of https://blueprints.launchpad.net/nova/+spec/image-metadata-prefiltering"}],"source_content_type":"text/x-rst","patch_set":3,"id":"9fb8cfa7_31f0b091","line":75,"range":{"start_line":75,"start_character":23,"end_line":75,"end_character":31},"updated":"2019-06-24 22:09:10.000000000","message":"minimum","commit_id":"0904fd1a6ef3b9ed40de040cbc595edcfcc4b9f2"},{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"ccc531125fd3203da3fe55be376097c9f0adbe4d","unresolved":false,"context_lines":[{"line_number":8,"context_line":"Libvirt: tenant control of qemu performance monitoring unit (vPMU)"},{"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\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/libvirt-pmu-configuration"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"qemu/kvm supports emulation of a vPMU to enable standard performance"},{"line_number":14,"context_line":"monitoring tools such as `Perf \u003chttps://perf.wiki.kernel.org/index.php/Main_Page\u003e`_"}],"source_content_type":"text/x-rst","patch_set":4,"id":"9fb8cfa7_eb7ae421","line":11,"range":{"start_line":11,"start_character":0,"end_line":11,"end_character":69},"updated":"2019-06-27 22:19:33.000000000","message":"cool, now make the file name, commit message tag, and topic match :)","commit_id":"bfd648dd0bde5c397e9dbb8d81c49fb9f7744fe2"},{"author":{"_account_id":6873,"name":"Matt Riedemann","email":"mriedem.os@gmail.com","username":"mriedem"},"change_message_id":"dd63f042cabdf5c3b43dea92623060387d66a441","unresolved":false,"context_lines":[{"line_number":31,"context_line":"latency of ~10us which is about 1% of the total budget for 5G end to end"},{"line_number":32,"context_line":"traffic processing latency. While this might seem small it is an appreciable"},{"line_number":33,"context_line":"portion of the total latency introduced by virtulisation and is therefore"},{"line_number":34,"context_line":"an important factor in achieving the end to end system latency target."},{"line_number":35,"context_line":""},{"line_number":36,"context_line":"As the provision of a vPMU is not currently controllable by an operator"},{"line_number":37,"context_line":"or tenant directly this creates a problem for those that want to enable"}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_fb659ce1","line":34,"updated":"2019-07-02 13:43:11.000000000","message":"tl;dr some VMs want to explicitly disable the pmu to avoid the overhead it introduces.","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"e0ebd426f0c5dde02fef57e0c5e15270cf4f967b","unresolved":false,"context_lines":[{"line_number":31,"context_line":"latency of ~10us which is about 1% of the total budget for 5G end to end"},{"line_number":32,"context_line":"traffic processing latency. While this might seem small it is an appreciable"},{"line_number":33,"context_line":"portion of the total latency introduced by virtulisation and is therefore"},{"line_number":34,"context_line":"an important factor in achieving the end to end system latency target."},{"line_number":35,"context_line":""},{"line_number":36,"context_line":"As the provision of a vPMU is not currently controllable by an operator"},{"line_number":37,"context_line":"or tenant directly this creates a problem for those that want to enable"}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_c79f9e4f","line":34,"in_reply_to":"9fb8cfa7_fb659ce1","updated":"2019-07-02 14:31:36.000000000","message":"yep","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"},{"author":{"_account_id":6873,"name":"Matt Riedemann","email":"mriedem.os@gmail.com","username":"mriedem"},"change_message_id":"dd63f042cabdf5c3b43dea92623060387d66a441","unresolved":false,"context_lines":[{"line_number":74,"context_line":"In addition to the above minimum changes the libvirt driver could be modifed"},{"line_number":75,"context_line":"to report support for a vPMU to improve scheduling. As the vPMU feature"},{"line_number":76,"context_line":"is supported by Nova\u0027s minimum required qemu/libvirt this would only be"},{"line_number":77,"context_line":"useful in a heterogeneous cloud. As such the desicion to expose this"},{"line_number":78,"context_line":"feature as a trait is left to the implementation and will be enabled as part"},{"line_number":79,"context_line":"of https://blueprints.launchpad.net/nova/+spec/image-metadata-prefiltering"},{"line_number":80,"context_line":"if desired."}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_9b3260f1","line":77,"range":{"start_line":77,"start_character":45,"end_line":77,"end_character":53},"updated":"2019-07-02 13:43:11.000000000","message":"decision","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"},{"author":{"_account_id":6873,"name":"Matt Riedemann","email":"mriedem.os@gmail.com","username":"mriedem"},"change_message_id":"dd63f042cabdf5c3b43dea92623060387d66a441","unresolved":false,"context_lines":[{"line_number":75,"context_line":"to report support for a vPMU to improve scheduling. As the vPMU feature"},{"line_number":76,"context_line":"is supported by Nova\u0027s minimum required qemu/libvirt this would only be"},{"line_number":77,"context_line":"useful in a heterogeneous cloud. As such the desicion to expose this"},{"line_number":78,"context_line":"feature as a trait is left to the implementation and will be enabled as part"},{"line_number":79,"context_line":"of https://blueprints.launchpad.net/nova/+spec/image-metadata-prefiltering"},{"line_number":80,"context_line":"if desired."},{"line_number":81,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_3b57b403","line":78,"range":{"start_line":78,"start_character":22,"end_line":78,"end_character":48},"updated":"2019-07-02 13:43:11.000000000","message":"I\u0027d vote no on exposing a trait for this until we actually have a need for that, i.e. someone asking for it. Otherwise we could make the argument that we could expose any number of things from every driver as traits saying they can be scheduled to like this, e.g. hw watchdog etc.","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"e861e1ea934802c853570b05b35662f5b38e32a1","unresolved":false,"context_lines":[{"line_number":75,"context_line":"to report support for a vPMU to improve scheduling. As the vPMU feature"},{"line_number":76,"context_line":"is supported by Nova\u0027s minimum required qemu/libvirt this would only be"},{"line_number":77,"context_line":"useful in a heterogeneous cloud. As such the desicion to expose this"},{"line_number":78,"context_line":"feature as a trait is left to the implementation and will be enabled as part"},{"line_number":79,"context_line":"of https://blueprints.launchpad.net/nova/+spec/image-metadata-prefiltering"},{"line_number":80,"context_line":"if desired."},{"line_number":81,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_5687da13","line":78,"range":{"start_line":78,"start_character":22,"end_line":78,"end_character":48},"in_reply_to":"9fb8cfa7_3b57b403","updated":"2019-07-02 18:13:07.000000000","message":"+1","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"},{"author":{"_account_id":6873,"name":"Matt Riedemann","email":"mriedem.os@gmail.com","username":"mriedem"},"change_message_id":"dd63f042cabdf5c3b43dea92623060387d66a441","unresolved":false,"context_lines":[{"line_number":133,"context_line":"None"},{"line_number":134,"context_line":""},{"line_number":135,"context_line":"The default behavior when the flavor extra_spec and image metadata"},{"line_number":136,"context_line":"value is unset was chosen to keep backwards compatiblity on upgrade."},{"line_number":137,"context_line":""},{"line_number":138,"context_line":""},{"line_number":139,"context_line":"Implementation"}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_9bb58034","line":136,"range":{"start_line":136,"start_character":44,"end_line":136,"end_character":56},"updated":"2019-07-02 13:43:11.000000000","message":"compatibility","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"},{"author":{"_account_id":6873,"name":"Matt Riedemann","email":"mriedem.os@gmail.com","username":"mriedem"},"change_message_id":"dd63f042cabdf5c3b43dea92623060387d66a441","unresolved":false,"context_lines":[{"line_number":157,"context_line":"Dependencies"},{"line_number":158,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"If we chose to enable the reporting of vPMU emulation as a trait then"},{"line_number":161,"context_line":"the consumption of that trait would depend on the completion of"},{"line_number":162,"context_line":"https://blueprints.launchpad.net/nova/+spec/image-metadata-prefiltering"},{"line_number":163,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_1bc990aa","line":160,"updated":"2019-07-02 13:43:11.000000000","message":"I could really only see this being necessary in the case that you\u0027ve got mixed hypervisors (as noted above), want to make sure that vms created with the pmu enabled are satisfied, and don\u0027t want to use host aggregates to tie the flavors / images that enable pmu to the libvirt hosts that provide the functionality.\n\nSince we have host aggregates though there is an alternative to exposing a trait so like I said above, I think we should defer exposing a trait for this until someone really asks for it.","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"e0ebd426f0c5dde02fef57e0c5e15270cf4f967b","unresolved":false,"context_lines":[{"line_number":157,"context_line":"Dependencies"},{"line_number":158,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":159,"context_line":""},{"line_number":160,"context_line":"If we chose to enable the reporting of vPMU emulation as a trait then"},{"line_number":161,"context_line":"the consumption of that trait would depend on the completion of"},{"line_number":162,"context_line":"https://blueprints.launchpad.net/nova/+spec/image-metadata-prefiltering"},{"line_number":163,"context_line":""}],"source_content_type":"text/x-rst","patch_set":5,"id":"9fb8cfa7_27e9dab6","line":160,"in_reply_to":"9fb8cfa7_1bc990aa","updated":"2019-07-02 14:31:36.000000000","message":"yep im fine with that.\nthat was teh only reason i was suggesting adding it.\nas you said we have host aggregates so its not a hard requriement and few clouds are actully multi hypervior\nignoring the ironic + vm virt driver case which will already be using flavors/aggrages to select if its a vm or baremetal instance.\n\nthe other semi common case for mixing it kvm + hyperv for windows guest but again that setup will already be using host aggregates to for hyperv for licensing reasons so this really  is jsut a nice to have that can be added later.","commit_id":"91c922a0f37482accc8591e102539a9a5f6dc6d9"}],"specs/train/libvirt-pmu-configuration.rst":[{"author":{"_account_id":14070,"name":"Eric Fried","email":"openstack@fried.cc","username":"efried"},"change_message_id":"29a8ece9da024a42e51b45cd8decec23051bc24a","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"bfb3d3c7_1e9be017","updated":"2019-05-30 16:15:26.000000000","message":"I\u0027m not reviewing this at the moment, just happened to notice that the path is wrong (needs to be specs/train/approved/...) so it won\u0027t get attention next Tuesday from people using the link I sent in http://lists.openstack.org/pipermail/openstack-discuss/2019-May/006735.html","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":30,"context_line":"While kvm/qemu support for a vPMU is generally a useful feature the requirement"},{"line_number":31,"context_line":"to collect and maintain virtual performance counter introduces additional"},{"line_number":32,"context_line":"latency of ~10us which is about 1% of the total budget for 5G  end to end"},{"line_number":33,"context_line":"traffic processing latency. While this might seam small its an appriable"},{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_f583b001","line":33,"range":{"start_line":33,"start_character":63,"end_line":33,"end_character":72},"updated":"2019-04-09 16:00:22.000000000","message":"I have no idea :D","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":30,"context_line":"While kvm/qemu support for a vPMU is generally a useful feature the requirement"},{"line_number":31,"context_line":"to collect and maintain virtual performance counter introduces additional"},{"line_number":32,"context_line":"latency of ~10us which is about 1% of the total budget for 5G  end to end"},{"line_number":33,"context_line":"traffic processing latency. While this might seam small its an appriable"},{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_7596e04b","line":33,"range":{"start_line":33,"start_character":45,"end_line":33,"end_character":49},"updated":"2019-04-09 16:00:22.000000000","message":"seem","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"89d2f9cda9dda55e470e05659dc9a398a25add9d","unresolved":false,"context_lines":[{"line_number":30,"context_line":"While kvm/qemu support for a vPMU is generally a useful feature the requirement"},{"line_number":31,"context_line":"to collect and maintain virtual performance counter introduces additional"},{"line_number":32,"context_line":"latency of ~10us which is about 1% of the total budget for 5G  end to end"},{"line_number":33,"context_line":"traffic processing latency. While this might seam small its an appriable"},{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_785b3521","line":33,"range":{"start_line":33,"start_character":63,"end_line":33,"end_character":72},"in_reply_to":"5fc1f717_f583b001","updated":"2019-04-09 16:13:14.000000000","message":"appriable-\u003eappreciable :)","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"as the provision of a vPMU is not currently contolable by an operator"},{"line_number":38,"context_line":"or tenant directly this creates a problem for those why want to enable"},{"line_number":39,"context_line":"or disable the vPMU to either avoid the latency overhaed or rely on it"},{"line_number":40,"context_line":"to monitor the performace of there workload."}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_1590bc42","line":37,"range":{"start_line":37,"start_character":0,"end_line":37,"end_character":2},"updated":"2019-04-09 16:00:22.000000000","message":"As","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":34,"context_line":"portion of the total latency intoduced by virtulisation an is therefore"},{"line_number":35,"context_line":"an important factor in aciving the end to end system latancy target."},{"line_number":36,"context_line":""},{"line_number":37,"context_line":"as the provision of a vPMU is not currently contolable by an operator"},{"line_number":38,"context_line":"or tenant directly this creates a problem for those why want to enable"},{"line_number":39,"context_line":"or disable the vPMU to either avoid the latency overhaed or rely on it"},{"line_number":40,"context_line":"to monitor the performace of there workload."}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_b579a82f","line":37,"range":{"start_line":37,"start_character":44,"end_line":37,"end_character":54},"updated":"2019-04-09 16:00:22.000000000","message":"controllable","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":37,"context_line":"as the provision of a vPMU is not currently contolable by an operator"},{"line_number":38,"context_line":"or tenant directly this creates a problem for those why want to enable"},{"line_number":39,"context_line":"or disable the vPMU to either avoid the latency overhaed or rely on it"},{"line_number":40,"context_line":"to monitor the performace of there workload."},{"line_number":41,"context_line":""},{"line_number":42,"context_line":"Use Cases"},{"line_number":43,"context_line":"---------"}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_756fc05f","line":40,"range":{"start_line":40,"start_character":29,"end_line":40,"end_character":34},"updated":"2019-04-09 16:00:22.000000000","message":"their","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":42,"context_line":"Use Cases"},{"line_number":43,"context_line":"---------"},{"line_number":44,"context_line":""},{"line_number":45,"context_line":"As a telecoms operator building a 5G network i wish to be able to deploy"},{"line_number":46,"context_line":"a virtualias Radio access network (vRAN) applicance with minimal latency"},{"line_number":47,"context_line":"impact form my virtualiation stack."},{"line_number":48,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_d5955433","line":45,"range":{"start_line":45,"start_character":45,"end_line":45,"end_character":46},"updated":"2019-04-09 16:00:22.000000000","message":"I (here and elsewhere)","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":47,"context_line":"impact form my virtualiation stack."},{"line_number":48,"context_line":""},{"line_number":49,"context_line":"As a tenant i wish to be able to monitor the performace of my application"},{"line_number":50,"context_line":"using standard tools like perf in a virutalised enviornment to enable"},{"line_number":51,"context_line":"developement, tuneing and profileing of my application."},{"line_number":52,"context_line":""},{"line_number":53,"context_line":"Proposed change"}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_95a34c9e","line":50,"range":{"start_line":50,"start_character":36,"end_line":50,"end_character":47},"updated":"2019-04-09 16:00:22.000000000","message":"virtualized","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":48,"context_line":""},{"line_number":49,"context_line":"As a tenant i wish to be able to monitor the performace of my application"},{"line_number":50,"context_line":"using standard tools like perf in a virutalised enviornment to enable"},{"line_number":51,"context_line":"developement, tuneing and profileing of my application."},{"line_number":52,"context_line":""},{"line_number":53,"context_line":"Proposed change"},{"line_number":54,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_15b99cca","line":51,"range":{"start_line":51,"start_character":0,"end_line":51,"end_character":36},"updated":"2019-04-09 16:00:22.000000000","message":"development, tuning, and profiling","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":15334,"name":"Stephen Finucane","display_name":"stephenfin","email":"stephenfin@redhat.com","username":"sfinucan"},"change_message_id":"87312a8f050d7ac3750d6e4223a5cd6b3b866f87","unresolved":false,"context_lines":[{"line_number":53,"context_line":"Proposed change"},{"line_number":54,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":55,"context_line":""},{"line_number":56,"context_line":"This spec proposes adding a tri state image metadata key"},{"line_number":57,"context_line":"hw_pmu\u003dTrue|false|\u003cunset\u003e and a corresponding flavor extra spec"},{"line_number":58,"context_line":"hw:pmu\u003dtrue|false|\u003cunset\u003e to enable/disable the pmu explcitly."},{"line_number":59,"context_line":""},{"line_number":60,"context_line":"The defualt value will be unset meaning the propry is not present in"},{"line_number":61,"context_line":"either the image or flavor."}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_b53a0831","line":58,"range":{"start_line":56,"start_character":0,"end_line":58,"end_character":62},"updated":"2019-04-09 16:00:22.000000000","message":"I\u0027m not sure why we need this. Realistically, the use cases you\u0027ve described necessitate real-time support. That being the case, why can\u0027t we just disable this by default if using real-time and leave it switched on if not? It\u0027s one less knob to configure and will achieve the same thing for the 99% of users who will be able to manage not using perf in their real-time VMs.","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"},{"author":{"_account_id":11604,"name":"sean mooney","email":"smooney@redhat.com","username":"sean-k-mooney"},"change_message_id":"89d2f9cda9dda55e470e05659dc9a398a25add9d","unresolved":false,"context_lines":[{"line_number":53,"context_line":"Proposed change"},{"line_number":54,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":55,"context_line":""},{"line_number":56,"context_line":"This spec proposes adding a tri state image metadata key"},{"line_number":57,"context_line":"hw_pmu\u003dTrue|false|\u003cunset\u003e and a corresponding flavor extra spec"},{"line_number":58,"context_line":"hw:pmu\u003dtrue|false|\u003cunset\u003e to enable/disable the pmu explcitly."},{"line_number":59,"context_line":""},{"line_number":60,"context_line":"The defualt value will be unset meaning the propry is not present in"},{"line_number":61,"context_line":"either the image or flavor."}],"source_content_type":"text/x-rst","patch_set":1,"id":"5fc1f717_d86ca9d1","line":58,"range":{"start_line":56,"start_character":0,"end_line":58,"end_character":62},"in_reply_to":"5fc1f717_b53a0831","updated":"2019-04-09 16:13:14.000000000","message":"my understanding is that if you dont have the pmu avaiable then a large amount of the performace metric that you get form perf and other tools like collectd or prometious dont provide accuate cpu info anymore.\n\nSo you would not be able to profile and tune realtiem application unless you could monitor the performacne from the host or extrnally at a end to end system level.\n\nbut even if we ignore that the main concern i woudl have with just disable the PMU when realtiem is enabled would be its a guest viable backward incompatible change. we could choose to do that and i can add it to the alternitives section but i did not want to start there.\n\nif that is the perferd solumtion i will updated teh propsoed change section and move teh current content to the alternivies section.","commit_id":"751c2c72c920ca6c1562e6fcfb8a31fe14cc355d"}]}