)]}'
{"/PATCHSET_LEVEL":[{"author":{"_account_id":6618,"name":"Ruby Loo","email":"opensrloo@gmail.com","username":"rloo"},"change_message_id":"83e3b3645e62fda46bbbb6c5d692622bd1b44928","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"a7a46b7c_3293a6ce","updated":"2021-11-04 14:42:22.000000000","message":"Thanks Julia! Interesting ideas. I don\u0027t recall hearing of a desire for this at Yahoo. Which doesn\u0027t mean there isn\u0027t, I may not know of it. When the ironic nodes are active, we don\u0027t know what the tenants are using those nodes for -- i would think it would be up to them to decide to reduce power or not... (and my guess is that this might not be a high priority for them). I think there might be cases (as Arne mentioned) when ironic nodes are not in active, when we might want to crank down the power, or power off (if we don\u0027t already). eg, nodes in some error state could be powered off, then powered on when/if someone decides to look into the error. (Or maybe powered off after x amount of time if no one has looked at it?? dunno.)","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"e6bbf10d_be345ebc","updated":"2021-11-04 12:39:29.000000000","message":"Thanks, Julia! I like the overall idea to add means to Ironic to help with power savings.\n\nFrom what I understand from the spec, there would be an additional state for active nodes\nin which power saving BIOS settings are applied out-of-band, i.e. via Redfish. Is that a correct summary?\n\nWhat about nodes in fast-track which are in available, i.e. powered on but not yet in active?\nEqually, nodes in error or other state may also profit from such a mechanism: why run at full speed if they are broken at the moment?\n\nThis is why I was thinking if an additional state is not too heavy/inflexible for this and rather have\nthese nodes tagged, along with some task to adapt the power saving settings?\n\nAs an alternative/additional (maybe crazy) idea: how about a mechanism which allows Ironic to switch off active nodes and check them at regular intervals, e.g. switch them on? This could potentially save a lot more power.\n\n","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11655,"name":"Julia Kreger","email":"juliaashleykreger@gmail.com","username":"jkreger","status":"Flying to the moon with a Jetpack!"},"change_message_id":"a083a8d3b30b2840be5e1f6b6ef660a5d9046843","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"2cf596a2_4abd3441","in_reply_to":"a7a46b7c_3293a6ce","updated":"2021-11-13 00:11:55.000000000","message":"I guess, my thought on why focus on a workflow driven approach is because most of these things are *really* going to require a power cycle.\n\nI man, we *can* and *should* likely include one of the cpu power tuning daemons in IPA images so machines can idle CPU cores while not in use. I guess ultimately, no matter what, we need to kind of map out what the best operational gains would be for operators, and balance that with our sanity.\n\nPowered off after x time is likely a great addition. No sense to leave things on forever if they are broken... as long as it is *not* a cleanfail.","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11655,"name":"Julia Kreger","email":"juliaashleykreger@gmail.com","username":"jkreger","status":"Flying to the moon with a Jetpack!"},"change_message_id":"a083a8d3b30b2840be5e1f6b6ef660a5d9046843","unresolved":false,"context_lines":[],"source_content_type":"","patch_set":1,"id":"c5a3f3f7_d3d572e7","in_reply_to":"e6bbf10d_be345ebc","updated":"2021-11-13 00:11:55.000000000","message":"So in essence, that is a good summary.\n\nI hadn\u0027t thought about nodes in fast track, and truthfully there are a ton of different use cases and patterns, so I think it might be good. There is one conundrum, many of the settings which impact power utilization also change cpu/memory geometry and behavior of a system.\n\nMy primary case I kind of want to support is backup and tertiary redundant systems like those in use in telecom environments where systems need to be on, but may only take traffic if some sort of disaster occurs, in which case some time to spin up the systems seems acceptable, at least to me. One of the operational requirements that I\u0027ve heard from a few places is they need to keep the machines on, to ensure issues are identified, as opposed to off where they can\u0027t actively be aware of the state and remedy issues in advance.\n\nI guess, in other words, nobody wants to go over to the disaster systems and find out they are broken too.\n\nI do think some sort of mechanism does may sense like a \"usage poller\" or whatnot, but I also think that is going to vary. Like in Nobodycam\u0027s case, I suspect it may be semi-manual or human engaged for some of his infrastructure, where as for a infra operator, maybe good to shut them down, where as others, maybe they need to apply logic to take them to low power mode. I *suspect* some of that interaction can\u0027t really be Ironic driven, and maybe something like watcher is better.","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"}],"specs/backlog/power-management.rst":[{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":10,"context_line":""},{"line_number":11,"context_line":"https://storyboard.openstack.org/#!/story/XXXXXXX"},{"line_number":12,"context_line":""},{"line_number":13,"context_line":"The biggest operating cost to running your own infrastucture, is generally"},{"line_number":14,"context_line":"electricity itself. For every watt used, an amount of waste heat is generated,"},{"line_number":15,"context_line":"and for every BTU of thermal energy put into the space, that thermal energy"},{"line_number":16,"context_line":"then needs to be removed from the chassis, and ultimately the data center"}],"source_content_type":"text/x-rst","patch_set":1,"id":"c5c6f5c6_6a0d3d67","line":13,"range":{"start_line":13,"start_character":47,"end_line":13,"end_character":60},"updated":"2021-11-04 12:39:29.000000000","message":"infrastructure","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":14,"context_line":"electricity itself. For every watt used, an amount of waste heat is generated,"},{"line_number":15,"context_line":"and for every BTU of thermal energy put into the space, that thermal energy"},{"line_number":16,"context_line":"then needs to be removed from the chassis, and ultimately the data center"},{"line_number":17,"context_line":"as a whole. Even with as efficent air conditions can be, if active cooling"},{"line_number":18,"context_line":"is required, generally that power conumption is twice the power consumption"},{"line_number":19,"context_line":"by the underlying compute workload and support systems."},{"line_number":20,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"c747e9ec_6108e848","line":17,"range":{"start_line":17,"start_character":25,"end_line":17,"end_character":33},"updated":"2021-11-04 12:39:29.000000000","message":"efficient","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":15,"context_line":"and for every BTU of thermal energy put into the space, that thermal energy"},{"line_number":16,"context_line":"then needs to be removed from the chassis, and ultimately the data center"},{"line_number":17,"context_line":"as a whole. Even with as efficent air conditions can be, if active cooling"},{"line_number":18,"context_line":"is required, generally that power conumption is twice the power consumption"},{"line_number":19,"context_line":"by the underlying compute workload and support systems."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Ironic, being so close to the physical metal, is one of the best places to"}],"source_content_type":"text/x-rst","patch_set":1,"id":"f3c80442_a42ef1e1","line":18,"range":{"start_line":18,"start_character":34,"end_line":18,"end_character":44},"updated":"2021-11-04 12:39:29.000000000","message":"consumption","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":19,"context_line":"by the underlying compute workload and support systems."},{"line_number":20,"context_line":""},{"line_number":21,"context_line":"Ironic, being so close to the physical metal, is one of the best places to"},{"line_number":22,"context_line":"drive the discussion of how to improve power efficency."},{"line_number":23,"context_line":""},{"line_number":24,"context_line":"Problem description"},{"line_number":25,"context_line":"\u003d\u003d\u003d\u003d\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":"31d58c83_fb716c2c","line":22,"range":{"start_line":22,"start_character":45,"end_line":22,"end_character":54},"updated":"2021-11-04 12:39:29.000000000","message":"efficiency","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":24,"context_line":"Problem description"},{"line_number":25,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":26,"context_line":""},{"line_number":27,"context_line":"* The world is shifting from speed as the primary focus to efficency."},{"line_number":28,"context_line":"  * Aside from architecture chnages and shifts in industries based upon"},{"line_number":29,"context_line":"    improved power efficency, many firmware settings have been"},{"line_number":30,"context_line":"    traditionally encoded to the physical machines to ensure maximum power"}],"source_content_type":"text/x-rst","patch_set":1,"id":"a028ffc4_809faf6f","line":27,"range":{"start_line":27,"start_character":59,"end_line":27,"end_character":68},"updated":"2021-11-04 12:39:29.000000000","message":"efficiency","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":25,"context_line":"\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d\u003d"},{"line_number":26,"context_line":""},{"line_number":27,"context_line":"* The world is shifting from speed as the primary focus to efficency."},{"line_number":28,"context_line":"  * Aside from architecture chnages and shifts in industries based upon"},{"line_number":29,"context_line":"    improved power efficency, many firmware settings have been"},{"line_number":30,"context_line":"    traditionally encoded to the physical machines to ensure maximum power"},{"line_number":31,"context_line":"    and performance."}],"source_content_type":"text/x-rst","patch_set":1,"id":"9229c574_b3cde288","line":28,"range":{"start_line":28,"start_character":28,"end_line":28,"end_character":35},"updated":"2021-11-03 21:12:57.000000000","message":"changes","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":26,"context_line":""},{"line_number":27,"context_line":"* The world is shifting from speed as the primary focus to efficency."},{"line_number":28,"context_line":"  * Aside from architecture chnages and shifts in industries based upon"},{"line_number":29,"context_line":"    improved power efficency, many firmware settings have been"},{"line_number":30,"context_line":"    traditionally encoded to the physical machines to ensure maximum power"},{"line_number":31,"context_line":"    and performance."},{"line_number":32,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"936e7d3d_dc7f166c","line":29,"range":{"start_line":29,"start_character":19,"end_line":29,"end_character":28},"updated":"2021-11-04 12:39:29.000000000","message":"efficiency","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":33,"context_line":"    We need to be able to undo this from a deployed system with minimal"},{"line_number":34,"context_line":"    downtime."},{"line_number":35,"context_line":""},{"line_number":36,"context_line":"  * A running system may also need to be able to recieve operaitonal tuning,"},{"line_number":37,"context_line":"    such as changes to utilize a lower power profile based on an operational"},{"line_number":38,"context_line":"    state or task. Conceptually it is almost as if operators need a \"turbo\""},{"line_number":39,"context_line":"    button to apply new bios configuration on the dynamically."}],"source_content_type":"text/x-rst","patch_set":1,"id":"9286443b_dfedf485","line":36,"range":{"start_line":36,"start_character":57,"end_line":36,"end_character":68},"updated":"2021-11-03 21:12:57.000000000","message":"operational","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":33,"context_line":"    We need to be able to undo this from a deployed system with minimal"},{"line_number":34,"context_line":"    downtime."},{"line_number":35,"context_line":""},{"line_number":36,"context_line":"  * A running system may also need to be able to recieve operaitonal tuning,"},{"line_number":37,"context_line":"    such as changes to utilize a lower power profile based on an operational"},{"line_number":38,"context_line":"    state or task. Conceptually it is almost as if operators need a \"turbo\""},{"line_number":39,"context_line":"    button to apply new bios configuration on the dynamically."}],"source_content_type":"text/x-rst","patch_set":1,"id":"f12e6bc9_b0a3704e","line":36,"range":{"start_line":36,"start_character":49,"end_line":36,"end_character":56},"updated":"2021-11-03 21:12:57.000000000","message":"receive","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":36,"context_line":"  * A running system may also need to be able to recieve operaitonal tuning,"},{"line_number":37,"context_line":"    such as changes to utilize a lower power profile based on an operational"},{"line_number":38,"context_line":"    state or task. Conceptually it is almost as if operators need a \"turbo\""},{"line_number":39,"context_line":"    button to apply new bios configuration on the dynamically."},{"line_number":40,"context_line":""},{"line_number":41,"context_line":"  * An operator may need to put these changes in on a schedule, or be un-done"},{"line_number":42,"context_line":"    on a schedule, as power billing often follows a \"Time Of Use\" or \"TOU\""}],"source_content_type":"text/x-rst","patch_set":1,"id":"9846c7ae_f2bf9e8b","line":39,"range":{"start_line":39,"start_character":43,"end_line":39,"end_character":49},"updated":"2021-11-03 21:12:57.000000000","message":"remove \"on the\"","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":49,"context_line":"    spawned the creation of the MOC (https://massopen.cloud) efforts and"},{"line_number":50,"context_line":"    the MGHPCC (https://www.mghpcc.org/)."},{"line_number":51,"context_line":""},{"line_number":52,"context_line":"    Often these machines are for distater recovery or failure response,"},{"line_number":53,"context_line":"    and don\u0027t need to be in active use, but on, monitored, and \"able\""},{"line_number":54,"context_line":"    to somehow take some workloads. These machines may be best viewed as"},{"line_number":55,"context_line":"    \"warm standby\" machines."}],"source_content_type":"text/x-rst","patch_set":1,"id":"55b96c96_9a7d782c","line":52,"range":{"start_line":52,"start_character":33,"end_line":52,"end_character":41},"updated":"2021-11-03 21:12:57.000000000","message":"disaster","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":65,"context_line":"    The likely best path for these sorts of things is workload or"},{"line_number":66,"context_line":"    hypervisor \"self-management\"."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"* Operators wish make scheduling decisions based on power utilization and"},{"line_number":69,"context_line":"  power efficency of underlying hardware. For example: "},{"line_number":70,"context_line":""},{"line_number":71,"context_line":"  * Peak Power Load"}],"source_content_type":"text/x-rst","patch_set":1,"id":"3133265f_3de831c7","line":68,"range":{"start_line":68,"start_character":12,"end_line":68,"end_character":21},"updated":"2021-11-04 12:39:29.000000000","message":"wish to make","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":66,"context_line":"    hypervisor \"self-management\"."},{"line_number":67,"context_line":""},{"line_number":68,"context_line":"* Operators wish make scheduling decisions based on power utilization and"},{"line_number":69,"context_line":"  power efficency of underlying hardware. For example: "},{"line_number":70,"context_line":""},{"line_number":71,"context_line":"  * Peak Power Load"},{"line_number":72,"context_line":"  * Idle Power Load"}],"source_content_type":"text/x-rst","patch_set":1,"id":"a01d83b2_77383c60","line":69,"range":{"start_line":69,"start_character":8,"end_line":69,"end_character":17},"updated":"2021-11-04 12:39:29.000000000","message":"efficiency","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":88,"context_line":"  of glass in services such as Ironic which can realistically be utilized"},{"line_number":89,"context_line":"  to support any number of clusters or consumers."},{"line_number":90,"context_line":"  Higher level pan-\"data center\" views, as opposed to numerous isolated"},{"line_number":91,"context_line":"  ironic deployments. Higher level views allow for simpler workflows when"},{"line_number":92,"context_line":"  having to orchustrate actions such as an A/B cluster failover, or numerous"},{"line_number":93,"context_line":"  of such operations in the same data."},{"line_number":94,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"70e14a7d_aa4bc79b","line":91,"range":{"start_line":91,"start_character":2,"end_line":91,"end_character":8},"updated":"2021-11-04 12:39:29.000000000","message":"Ironic","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":89,"context_line":"  to support any number of clusters or consumers."},{"line_number":90,"context_line":"  Higher level pan-\"data center\" views, as opposed to numerous isolated"},{"line_number":91,"context_line":"  ironic deployments. Higher level views allow for simpler workflows when"},{"line_number":92,"context_line":"  having to orchustrate actions such as an A/B cluster failover, or numerous"},{"line_number":93,"context_line":"  of such operations in the same data."},{"line_number":94,"context_line":""},{"line_number":95,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"5f73f64e_2512b837","line":92,"range":{"start_line":92,"start_character":12,"end_line":92,"end_character":23},"updated":"2021-11-03 21:12:57.000000000","message":"orchestrate","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":93,"context_line":"  of such operations in the same data."},{"line_number":94,"context_line":""},{"line_number":95,"context_line":""},{"line_number":96,"context_line":"* Operators wish to consolidate workloads dynamically for efficency, and"},{"line_number":97,"context_line":"  then on demand spread them out, adjust settings, and workloads."},{"line_number":98,"context_line":""},{"line_number":99,"context_line":"  In other words, if the compute node doesn\u0027t need to be on, and we have"}],"source_content_type":"text/x-rst","patch_set":1,"id":"51544d14_6e830a07","line":96,"range":{"start_line":96,"start_character":58,"end_line":96,"end_character":67},"updated":"2021-11-04 12:39:29.000000000","message":"efficiency","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":100,"context_line":"  reserve, turn it off. If it needs to be in \"warm standby\", then put it"},{"line_number":101,"context_line":"  in such state until it needs to be brought out."},{"line_number":102,"context_line":""},{"line_number":103,"context_line":"  In all liklihood, functionality, at least at thie highest levels of"},{"line_number":104,"context_line":"  orchustration, are largely best solved by projects such as"},{"line_number":105,"context_line":"  OpenStack Watcher, yet at the lower levels, the machines still need power"},{"line_number":106,"context_line":"  management, and/or settings management workflows and changes to enable the"}],"source_content_type":"text/x-rst","patch_set":1,"id":"de286a4b_d1c0532a","line":103,"range":{"start_line":103,"start_character":9,"end_line":103,"end_character":18},"updated":"2021-11-04 12:39:29.000000000","message":"likelyhood","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":100,"context_line":"  reserve, turn it off. If it needs to be in \"warm standby\", then put it"},{"line_number":101,"context_line":"  in such state until it needs to be brought out."},{"line_number":102,"context_line":""},{"line_number":103,"context_line":"  In all liklihood, functionality, at least at thie highest levels of"},{"line_number":104,"context_line":"  orchustration, are largely best solved by projects such as"},{"line_number":105,"context_line":"  OpenStack Watcher, yet at the lower levels, the machines still need power"},{"line_number":106,"context_line":"  management, and/or settings management workflows and changes to enable the"}],"source_content_type":"text/x-rst","patch_set":1,"id":"10cd9322_29b184dc","line":103,"range":{"start_line":103,"start_character":47,"end_line":103,"end_character":51},"updated":"2021-11-04 12:39:29.000000000","message":"the","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":101,"context_line":"  in such state until it needs to be brought out."},{"line_number":102,"context_line":""},{"line_number":103,"context_line":"  In all liklihood, functionality, at least at thie highest levels of"},{"line_number":104,"context_line":"  orchustration, are largely best solved by projects such as"},{"line_number":105,"context_line":"  OpenStack Watcher, yet at the lower levels, the machines still need power"},{"line_number":106,"context_line":"  management, and/or settings management workflows and changes to enable the"},{"line_number":107,"context_line":"  ability to support a union of operational requirements and energy savings."}],"source_content_type":"text/x-rst","patch_set":1,"id":"da5e3024_8bf96f7b","line":104,"range":{"start_line":104,"start_character":2,"end_line":104,"end_character":15},"updated":"2021-11-03 21:12:57.000000000","message":"orchestration","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":133,"context_line":"  an expansive task/capability, it could ultimately be useful,"},{"line_number":134,"context_line":"  and underlying logic must always supply aggregate numbers when"},{"line_number":135,"context_line":"  possible if power supplies are visible in a machine *as* there are"},{"line_number":136,"context_line":"  always multiple devices."},{"line_number":137,"context_line":""},{"line_number":138,"context_line":".. TODO:: does our snmp power interface have a defficency in it\u0027s model"},{"line_number":139,"context_line":"   in that it likely only supports one plug per node?"}],"source_content_type":"text/x-rst","patch_set":1,"id":"c632c2b2_4d475244","line":136,"updated":"2021-11-03 21:12:57.000000000","message":"Does this imply we need to model PDUs and their relationship to chassis and nodes? This would feed into my suggestion below.","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11292,"name":"Arne Wiebalck","email":"Arne.Wiebalck@cern.ch","username":"wiebalck"},"change_message_id":"4c97522e4ec4e4c23837e18bd929e58bf3de8ad2","unresolved":true,"context_lines":[{"line_number":135,"context_line":"  possible if power supplies are visible in a machine *as* there are"},{"line_number":136,"context_line":"  always multiple devices."},{"line_number":137,"context_line":""},{"line_number":138,"context_line":".. TODO:: does our snmp power interface have a defficency in it\u0027s model"},{"line_number":139,"context_line":"   in that it likely only supports one plug per node?"},{"line_number":140,"context_line":""},{"line_number":141,"context_line":".. TODO::"}],"source_content_type":"text/x-rst","patch_set":1,"id":"4aced7b2_47829e61","line":138,"range":{"start_line":138,"start_character":47,"end_line":138,"end_character":57},"updated":"2021-11-04 12:39:29.000000000","message":"deficiency","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":4571,"name":"Steve Baker","email":"sbaker@redhat.com","username":"steve-stevebaker"},"change_message_id":"7487d964b62da5c81cd9e753bd6186a8b0662a95","unresolved":true,"context_lines":[{"line_number":157,"context_line":"another project? In another language? Why aren\u0027t we using those? This doesn\u0027t"},{"line_number":158,"context_line":"have to be a full literature review, but it should demonstrate that thought has"},{"line_number":159,"context_line":"been put into why the proposed solution is an appropriate one."},{"line_number":160,"context_line":""},{"line_number":161,"context_line":"Data model impact"},{"line_number":162,"context_line":"-----------------"},{"line_number":163,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"1eeaf4ab_3c5364ad","line":160,"updated":"2021-11-03 21:12:57.000000000","message":"How about extending the chassis and allocation model so that is possible to implement these sorts of scheduling policies:\n- pack one chassis/PDU completely before activating another\n- modelling the physical relationship of chassis within the data warehouse, so that activated chassis are evenly distributed within the space to avoid overworking the cooling in one area (or packed together in one cooling domain so others can be put in standby)\n\nMaybe these policies would require the allocation API to be backed by the placement service to be feasible?","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"},{"author":{"_account_id":11655,"name":"Julia Kreger","email":"juliaashleykreger@gmail.com","username":"jkreger","status":"Flying to the moon with a Jetpack!"},"change_message_id":"a083a8d3b30b2840be5e1f6b6ef660a5d9046843","unresolved":true,"context_lines":[{"line_number":157,"context_line":"another project? In another language? Why aren\u0027t we using those? This doesn\u0027t"},{"line_number":158,"context_line":"have to be a full literature review, but it should demonstrate that thought has"},{"line_number":159,"context_line":"been put into why the proposed solution is an appropriate one."},{"line_number":160,"context_line":""},{"line_number":161,"context_line":"Data model impact"},{"line_number":162,"context_line":"-----------------"},{"line_number":163,"context_line":""}],"source_content_type":"text/x-rst","patch_set":1,"id":"45d21b78_0aa875d9","line":160,"in_reply_to":"1eeaf4ab_3c5364ad","updated":"2021-11-13 00:11:55.000000000","message":"I *think* these are all great ideas. I also suspect it would be a good idea to discuss and try to hash something like that out, even if it is just a thought experiment.","commit_id":"207188082766da3474cc197eb7fd47ba69eec0fb"}]}