| *** eghobo has joined #openstack-ironic | 00:04 | |
| *** eghobo has quit IRC | 00:16 | |
| *** naohirot has joined #openstack-ironic | 00:25 | |
| *** igordcard has quit IRC | 00:36 | |
| *** mitz has joined #openstack-ironic | 00:47 | |
| *** jamielennox|away is now known as jamielennox | 01:07 | |
| *** kan_ has joined #openstack-ironic | 01:28 | |
| *** heyongli has joined #openstack-ironic | 01:40 | |
| *** jnielsen55 has quit IRC | 02:29 | |
| *** davidlenwell has quit IRC | 02:33 | |
| *** ramineni has joined #openstack-ironic | 02:48 | |
| *** eghobo has joined #openstack-ironic | 02:54 | |
| *** zhenguo has joined #openstack-ironic | 03:03 | |
| *** eghobo has quit IRC | 03:07 | |
| *** heyongli has quit IRC | 03:13 | |
| *** heyongli has joined #openstack-ironic | 03:14 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Validate capability in properties and instance_info https://review.openstack.org/174785 | 03:17 |
|---|---|---|
| openstackgerrit | Zhenguo Niu proposed openstack/ironic-specs: Add nodes tagging support https://review.openstack.org/183971 | 03:18 |
| *** mdbooth has quit IRC | 03:27 | |
| *** UForgotten has quit IRC | 03:27 | |
| *** errr has quit IRC | 03:27 | |
| *** errr has joined #openstack-ironic | 03:27 | |
| *** heyongli has quit IRC | 03:27 | |
| *** bradjones|away has quit IRC | 03:27 | |
| *** valkyrja has quit IRC | 03:27 | |
| *** mikal has quit IRC | 03:27 | |
| *** UForgotten has joined #openstack-ironic | 03:27 | |
| *** mdbooth has joined #openstack-ironic | 03:27 | |
| *** mikal has joined #openstack-ironic | 03:28 | |
| *** valkyrja has joined #openstack-ironic | 03:28 | |
| *** harlowja_at_home has joined #openstack-ironic | 03:28 | |
| *** bradjones has joined #openstack-ironic | 03:28 | |
| *** bradjones has quit IRC | 03:28 | |
| *** bradjones has joined #openstack-ironic | 03:28 | |
| *** coolsvap|afk is now known as coolsvap | 03:44 | |
| *** eghobo has joined #openstack-ironic | 03:53 | |
| *** davidlenwell has joined #openstack-ironic | 03:55 | |
| *** eghobo has quit IRC | 03:56 | |
| *** eghobo has joined #openstack-ironic | 04:03 | |
| *** naohirot has quit IRC | 04:06 | |
| *** naohirot has joined #openstack-ironic | 04:14 | |
| *** kalpase has joined #openstack-ironic | 04:19 | |
| *** harlowja_at_home has quit IRC | 04:20 | |
| *** achanda has joined #openstack-ironic | 04:33 | |
| *** saripurigopi has joined #openstack-ironic | 04:33 | |
| *** eghobo_ has joined #openstack-ironic | 04:33 | |
| *** eghobo has quit IRC | 04:37 | |
| *** eghobo has joined #openstack-ironic | 04:40 | |
| *** eghobo_ has quit IRC | 04:41 | |
| *** rameshg87 has joined #openstack-ironic | 04:42 | |
| rameshg87 | good morning (ugt) ironic | 04:43 |
| Haomeng | rameshg87: morning ramesh:) | 04:43 |
| rameshg87 | Haomeng: o/ | 04:43 |
| *** eghobo has left #openstack-ironic | 04:45 | |
| *** achanda has quit IRC | 04:55 | |
| *** achanda has joined #openstack-ironic | 05:13 | |
| saripurigopi | gm rameshg87 | 05:20 |
| rameshg87 | gm saripurigopi | 05:20 |
| *** achanda_ has joined #openstack-ironic | 05:22 | |
| *** achanda has quit IRC | 05:23 | |
| saripurigopi | rameshg87: i'm looking for the ilo secure boot demo video. | 05:24 |
| rameshg87 | saripurigopi: it is available in youtube now | 05:25 |
| * rameshg87 fetches the link | 05:25 | |
| saripurigopi | rameshg87: got it | 05:25 |
| rameshg87 | https://www.openstack.org/summit/vancouver-2015/summit-videos/presentation/how-to-make-ironic-bare-metal-provisioning-more-secure-and-reliable | 05:27 |
| rameshg87 | :) | 05:27 |
| *** achanda_ has quit IRC | 05:28 | |
| saripurigopi | rameshg87 :-) thanks | 05:28 |
| *** oomichi has joined #openstack-ironic | 05:28 | |
| *** Haomeng|2 has joined #openstack-ironic | 05:36 | |
| *** lsmola has joined #openstack-ironic | 05:36 | |
| *** Haomeng has quit IRC | 05:38 | |
| *** chenglch has joined #openstack-ironic | 05:41 | |
| *** dlpartain has joined #openstack-ironic | 05:43 | |
| *** dlpartain has left #openstack-ironic | 05:44 | |
| *** jcoufal has joined #openstack-ironic | 05:57 | |
| *** pradipta has joined #openstack-ironic | 05:58 | |
| *** tteggel has quit IRC | 06:04 | |
| *** ukalifon has joined #openstack-ironic | 06:15 | |
| *** achanda has joined #openstack-ironic | 06:19 | |
| *** saripurigopi has quit IRC | 06:38 | |
| *** dtantsur|afk is now known as dtantsur | 06:44 | |
| dtantsur | Morning Ironic | 06:44 |
| *** mordred has quit IRC | 06:46 | |
| *** heyongli has joined #openstack-ironic | 06:50 | |
| *** edcragg has joined #openstack-ironic | 07:00 | |
| *** achanda has quit IRC | 07:02 | |
| *** achanda has joined #openstack-ironic | 07:04 | |
| *** edcragg has quit IRC | 07:05 | |
| rameshg87 | dtantsur: o/ | 07:14 |
| dtantsur | o/ | 07:14 |
| *** achanda has quit IRC | 07:17 | |
| *** athomas has joined #openstack-ironic | 07:18 | |
| *** yog__ has joined #openstack-ironic | 07:24 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic: Add vendor-passthru to attach and boot an ISO https://review.openstack.org/188317 | 07:28 |
| *** lazy_prince has joined #openstack-ironic | 07:29 | |
| *** viktors|afk is now known as viktors | 07:45 | |
| *** romainh has joined #openstack-ironic | 07:52 | |
| *** jistr has joined #openstack-ironic | 07:52 | |
| openstackgerrit | Merged openstack/python-ironicclient: Updated from global requirements https://review.openstack.org/188495 | 07:57 |
| *** lucasagomes has joined #openstack-ironic | 08:05 | |
| *** davidlenwell has quit IRC | 08:06 | |
| *** davidlenwell has joined #openstack-ironic | 08:09 | |
| *** dguerri` is now known as dguerri | 08:11 | |
| *** yuanying has quit IRC | 08:13 | |
| *** derekh has joined #openstack-ironic | 08:14 | |
| *** achanda has joined #openstack-ironic | 08:14 | |
| *** ndipanov has joined #openstack-ironic | 08:15 | |
| *** bethelwell has joined #openstack-ironic | 08:17 | |
| *** davidlenwell has quit IRC | 08:23 | |
| *** achanda has quit IRC | 08:29 | |
| *** davidlenwell has joined #openstack-ironic | 08:30 | |
| *** achanda has joined #openstack-ironic | 08:30 | |
| *** yog__ has quit IRC | 08:33 | |
| *** yog__ has joined #openstack-ironic | 08:35 | |
| *** achanda has quit IRC | 08:35 | |
| *** bethelwe_ has joined #openstack-ironic | 08:38 | |
| *** bethelwell has quit IRC | 08:42 | |
| *** achanda has joined #openstack-ironic | 08:49 | |
| *** lazy_prince has quit IRC | 08:51 | |
| *** leopoldj has joined #openstack-ironic | 08:54 | |
| *** MattMan has quit IRC | 08:56 | |
| *** e0ne has joined #openstack-ironic | 08:56 | |
| *** achanda has quit IRC | 08:56 | |
| *** MattMan has joined #openstack-ironic | 08:56 | |
| *** mgoddard has joined #openstack-ironic | 08:57 | |
| *** achanda has joined #openstack-ironic | 08:57 | |
| *** yog__ has quit IRC | 08:57 | |
| *** bethelwe_ has quit IRC | 08:59 | |
| *** bethelwell has joined #openstack-ironic | 08:59 | |
| openstackgerrit | Victor Sergeyev proposed openstack/ironic: Use oslo.messeging serializers https://review.openstack.org/188701 | 09:00 |
| *** dtantsur is now known as dtantsur|bbl | 09:03 | |
| openstackgerrit | Victor Sergeyev proposed openstack/ironic: Use oslo.messeging serializers https://review.openstack.org/188701 | 09:03 |
| *** achanda has quit IRC | 09:03 | |
| *** yog__ has joined #openstack-ironic | 09:09 | |
| *** e0ne is now known as e0ne_ | 09:10 | |
| openstackgerrit | Merged openstack/ironic: Updated from global requirements https://review.openstack.org/188475 | 09:12 |
| *** dlpartain has joined #openstack-ironic | 09:19 | |
| *** dlpartain has left #openstack-ironic | 09:19 | |
| *** e0ne_ has quit IRC | 09:20 | |
| *** Nisha has joined #openstack-ironic | 09:22 | |
| *** bethelwe_ has joined #openstack-ironic | 09:23 | |
| *** bethelwe_ has quit IRC | 09:23 | |
| *** bethelwell has quit IRC | 09:23 | |
| *** bethelwell has joined #openstack-ironic | 09:23 | |
| *** e0ne has joined #openstack-ironic | 09:29 | |
| *** stendulker has joined #openstack-ironic | 09:30 | |
| *** achanda has joined #openstack-ironic | 09:31 | |
| *** alexpilotti has joined #openstack-ironic | 09:33 | |
| *** achanda has quit IRC | 09:33 | |
| *** achanda has joined #openstack-ironic | 09:34 | |
| openstackgerrit | SHIGEMATSU Mitsuhiro proposed openstack/bifrost: Correct mistakes in license notices https://review.openstack.org/189079 | 09:40 |
| *** kbyrne has quit IRC | 09:45 | |
| rameshg87 | lucasagomes: hi | 09:48 |
| *** kbyrne has joined #openstack-ironic | 09:48 | |
| lucasagomes | rameshg87, hi there! | 09:49 |
| rameshg87 | lucasagomes: one quick question | 09:49 |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic-python-agent: Add power_off command in standby extension https://review.openstack.org/189241 | 09:49 |
| lucasagomes | shoot | 09:49 |
| rameshg87 | lucasagomes: any idea why we use sysrq-trigger in https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/shell/reboot.sh | 09:49 |
| rameshg87 | instead of just invoking reboot command | 09:49 |
| rameshg87 | lucasagomes: is it because it cannot be done in a coreos container or something ? | 09:50 |
| lucasagomes | rameshg87, no idea... It was there already when I started to use it | 09:50 |
| rameshg87 | lucasagomes: I just raised a review ^^^ which just uses shutdown -h now for powering off the system | 09:50 |
| lucasagomes | the commit message doesn't seem to say much either | 09:50 |
| rameshg87 | shutdown -h works for me | 09:51 |
| lucasagomes | jroll, ^ (since he wrote the code) | 09:51 |
| *** achanda_ has joined #openstack-ironic | 09:51 | |
| rameshg87 | yeah better check with him when he comes online | 09:52 |
| lucasagomes | yeah, cause I don't know. This command was there before | 09:52 |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic-python-agent: Add power_off command in standby extension https://review.openstack.org/189241 | 09:53 |
| *** achanda has quit IRC | 09:53 | |
| *** pelix has joined #openstack-ironic | 09:55 | |
| *** kan_ has quit IRC | 09:55 | |
| *** degorenko has quit IRC | 09:55 | |
| lucasagomes | rameshg87, if you have some time https://review.openstack.org/#/c/187678/ | 09:56 |
| lucasagomes | have you tested it already? | 09:56 |
| rameshg87 | lucasagomes: still no :( | 09:56 |
| lucasagomes | :-( | 09:56 |
| rameshg87 | I still don't have my old cpu | 09:56 |
| lucasagomes | right | 09:56 |
| lucasagomes | well if you have some time take a look at the spec | 09:56 |
| lucasagomes | it's pretty short | 09:56 |
| rameshg87 | sure .. | 09:57 |
| *** naohirot has quit IRC | 09:57 | |
| *** tteggel has joined #openstack-ironic | 10:08 | |
| *** openstackgerrit has quit IRC | 10:09 | |
| *** openstackgerrit has joined #openstack-ironic | 10:09 | |
| *** saripurigopi has joined #openstack-ironic | 10:11 | |
| *** tteggel has quit IRC | 10:12 | |
| *** tteggel has joined #openstack-ironic | 10:13 | |
| *** dtantsur|bbl is now known as dtantsur | 10:14 | |
| stendulker | lucasgomes: Hi | 10:15 |
| *** Nisha has quit IRC | 10:15 | |
| stendulker | lucasagomes: Hi | 10:16 |
| lucasagomes | stendulker, hi there | 10:16 |
| stendulker | lucasagomes: This is regarding the secure boot spec for the pxe driver, that we disussed during summit.. | 10:17 |
| stendulker | lucasagomes: Please review it when you get time. https://review.openstack.org/#/c/174295/ | 10:17 |
| lucasagomes | oh will do | 10:18 |
| lucasagomes | thanks | 10:18 |
| openstackgerrit | Merged stackforge/ironic-discoverd: Do not require scheduling attributes in ramdisk https://review.openstack.org/186033 | 10:19 |
| stendulker | lucasagomes: Thank you :) | 10:20 |
| *** oomichi has quit IRC | 10:25 | |
| *** achanda_ has quit IRC | 10:28 | |
| dtantsur | lucasagomes, could you also have a look at https://review.openstack.org/#/c/188370 ? It's a short spec but should be interesting to discuss | 10:30 |
| *** rameshg87 is now known as rameshg87-away | 10:31 | |
| *** pradipta has quit IRC | 10:34 | |
| lucasagomes | dtantsur, ack | 10:34 |
| *** achanda has joined #openstack-ironic | 10:35 | |
| *** achanda has quit IRC | 10:42 | |
| *** achanda has joined #openstack-ironic | 10:43 | |
| *** bethelwell has quit IRC | 10:43 | |
| *** Mark_Melb has joined #openstack-ironic | 10:44 | |
| *** achanda has quit IRC | 10:45 | |
| *** achanda has joined #openstack-ironic | 10:46 | |
| *** coolsvap is now known as coolsvap|afk | 10:47 | |
| sambetts | Good UGT Morning ironic :) | 10:51 |
| *** achanda has quit IRC | 10:52 | |
| dtantsur | sambetts, morning! | 10:52 |
| *** e0ne is now known as e0ne_ | 10:56 | |
| *** e0ne_ is now known as e0ne | 10:57 | |
| *** edcragg has joined #openstack-ironic | 10:58 | |
| *** ramineni has quit IRC | 11:01 | |
| lucasagomes | sambetts, good ugt morning | 11:01 |
| sambetts | o/ dtantsur, lucasagomes | 11:01 |
| dtantsur | sambetts, do you wanna work on https://blueprints.launchpad.net/ironic-inspector/+spec/node-not-found-hook any time soon? Otherwise I'm bumping it to the next release, as I won't have time for it. | 11:05 |
| sambetts | dtantsur: Hopefully by tomorrow or so I should have a better idea of my overall work load, I would love to work on that this cycle though if I have the time | 11:06 |
| dtantsur | sambetts, ok fine, I will keep it targeted to this version for now | 11:07 |
| sambetts | dtantsur: :D awesome, If I don't have the time for it we can bump it but I would be happy to get that in | 11:07 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic-specs: UEFI secure boot support for pxe_ilo driver https://review.openstack.org/174295 | 11:08 |
| *** wuhg has joined #openstack-ironic | 11:08 | |
| dtantsur | ok, we'll see :) I still do not have any release schedule for 2.0.0, I just wanna release it asap (i.e. when we're done renaming and doing breaking changes) | 11:08 |
| *** thrash|wknd is now known as thrash | 11:10 | |
| sambetts | dtantsur: that makes sense :) the release time frame is set by you anyway right, so 2.1.0 could be like a month after 2.0.0 if we wanted it to be ? | 11:11 |
| dtantsur | yeah, though for now I expect 2.0.0 very soon, and 2.1.0 with Liberty. but we can change it at any time :) | 11:12 |
| sambetts | :) awesome | 11:13 |
| *** dguerri is now known as dguerri` | 11:19 | |
| Mark_Melb | Morning/afternoon all: Does anyone know the best way to deal with this error from IPA writing out a config drive: http://pastebin.com/YKM9y7QY Pretty sure I am running the latest of everything and I have tried using a Live CD to remove all existing partitions with fdisk and parted. Have also tried zero’ing out the drive with DD. Use IPA with agent_ipmitool. | 11:22 |
| dtantsur | Mark_Melb, hi, you may want to wait a couple of hours until jroll or JoshNang arrive, they must know IPA much better | 11:28 |
| Mark_Melb | Will do. Cheers :) | 11:29 |
| *** yuikotakada has joined #openstack-ironic | 11:35 | |
| yuikotakada | Hi, Ironic | 11:37 |
| dtantsur | yuikotakada, hi, g'evening! | 11:37 |
| *** e0ne is now known as e0ne_ | 11:38 | |
| yuikotakada | dtantsur: Hi, g'morning(utc) | 11:38 |
| *** e0ne_ is now known as e0ne | 11:39 | |
| *** stendulker has quit IRC | 11:40 | |
| dtantsur | yuikotakada, sambetts, trown|outttypeww, I've sent an email to the ML about plans for inspector-2.0, please read/comment/etc :) | 11:43 |
| yuikotakada | dtantsur, thanks, I'll check it. | 11:43 |
| *** lucasagomes is now known as lucas-hungry | 11:44 | |
| * rameshg87-away goes home | 11:50 | |
| *** rameshg87-away has quit IRC | 11:50 | |
| *** yuikotakada has quit IRC | 11:56 | |
| *** yuikotakada has joined #openstack-ironic | 12:06 | |
| *** yuikotakada is now known as yuikotak_dinner | 12:07 | |
| *** zhenguo has quit IRC | 12:07 | |
| *** yuikotak_dinner has quit IRC | 12:11 | |
| *** yuikota__ has joined #openstack-ironic | 12:11 | |
| TheJulia | Good morning folks | 12:14 |
| *** zz_jgrimm is now known as jgrimm | 12:17 | |
| *** yuikota__ has quit IRC | 12:19 | |
| *** yuikotak_dinner has joined #openstack-ironic | 12:19 | |
| TheJulia | ThomasPB: The use context is utilizaiton of the tools to provision machines that ultimately would need to be standalone tenants on physical bare metal for performance, architecture, or scalability reasons. Granted, one can and some orgs do use ironic to alow user to provision bare metal machines to meet their needs, but your capabilities are all going to depend on the intermediate infrastucture | 12:21 |
| TheJulia | and configuration since Ironic doesn't really provide oeprational features for interction where as it helps provision the reqeusted image on available hardware. | 12:21 |
| TheJulia | yay for autowrap | 12:21 |
| dtantsur | TheJulia, morning | 12:22 |
| *** saripurigopi has quit IRC | 12:31 | |
| *** trown|outttypeww is now known as trown | 12:34 | |
| *** lazy_prince has joined #openstack-ironic | 12:34 | |
| TheJulia | cinerama: if your feeling up to it today, lets sync up on 186875 | 12:35 |
| TheJulia | Any ironic cores who have had their first cup of coffee, please have a quick glance at 186581 :) | 12:37 |
| *** lucas-hungry is now known as lucasagomes | 12:38 | |
| bcornec | Hello Julia | 12:40 |
| bcornec | Sorry TheJulia ! | 12:40 |
| bcornec | I'm trying to invoke bifrost behind the HP FW in a docker container (I like to play ;-) | 12:40 |
| *** kalpase has quit IRC | 12:41 | |
| TheJulia | Hi bcornec! Hope your doing well! | 12:41 |
| bcornec | Do you think it could be possible to split your script env-setup in 2 parts: one to install requiremetns, the other to finalize the setup ? That way the first could be repaced by installs done in the container dierctly (with proxy management) and the rest which doesn't rely on proxies would just work as documented. | 12:42 |
| bcornec | I'm doing fine yes, thanks ! Hope the same for you :-) | 12:42 |
| TheJulia | bcornec: I'm all for this, as long as we end up with something that says either "Here is how you fire this up in a docker container...." or "xyz doesn't work in a docker container, bugs are filed" :) | 12:42 |
| TheJulia | bcornec: I think that is totally do-able | 12:43 |
| TheJulia | proxy utiliztion may require us to pass http_proxy env vars down to modules though | 12:43 |
| bcornec | Ok, so let me propose a patch to you then... | 12:43 |
| openstackgerrit | Pratyusha proposed stackforge/proliantutils: Add iSCSI boot support in RIS https://review.openstack.org/187162 | 12:44 |
| bcornec | Yes, proxy is a nightmare :-(. To start wuth the pip command required it as a param on the CLI instead of understanding $http_proxy. Come on, it's XXIst century now !! | 12:44 |
| bcornec | I thought Python was great for developers ;-) | 12:44 |
| TheJulia | bcornec: if you want to just propose part of the idea, I'll be glad to take a look and discuss :) | 12:45 |
| TheJulia | Given I have a local proxy on my laptop, I can also test on it. | 12:45 |
| * TheJulia thinks there is still a caching proxy in the house network closet | 12:45 | |
| * TheJulia has the sudden ralization that she should be using this caching proxy | 12:46 | |
| bcornec | Well I'm behind a real one, the Big HP Proxy so I have that issue to solve anyway in order to go further for now. | 12:46 |
| *** dguerri` is now known as dguerri | 12:46 | |
| bcornec | I'm just invoking ansible right now, and it seems blocked at the second call, so I need to dig more in order to propose something to you. | 12:47 |
| TheJulia | bcornec: by chance did you checkout 1459084 ? | 12:47 |
| TheJulia | bug 1459084 | 12:48 |
| openstack | bug 1459084 in bifrost "biforst does not support http proxy configurations" [Medium,Triaged] https://launchpad.net/bugs/1459084 | 12:48 |
| dtantsur | lucasagomes, jroll, do you know if anyone works on promoting PXE/agent vendor passthrough to proper API? | 12:49 |
| lucasagomes | dtantsur, I'm not sure | 12:49 |
| *** thiagop has quit IRC | 12:49 | |
| lucasagomes | I also seem to don't recall what was decided in the summit properly | 12:50 |
| dtantsur | lucasagomes, I think we decided to promote them to decouple passthrough and deploy | 12:50 |
| lucasagomes | IIRC, we talked about having the vendor passthru methods as part of the other interfaces as well right? | 12:50 |
| *** gabriel-bezerra has quit IRC | 12:50 | |
| dtantsur | lucasagomes, we didn't agree on this one | 12:50 |
| *** lazy_prince has quit IRC | 12:50 | |
| lucasagomes | dtantsur, right, but here's the thing... things like "pass_deploy_info" is only used by the DIB ramdisk | 12:50 |
| lucasagomes | and specific to the pxe_* drivers | 12:51 |
| lucasagomes | not sure if we want to promote it | 12:51 |
| dtantsur | lucasagomes, in agent they have heartbeat | 12:51 |
| bcornec | TheJulia thanks for the pointer ! | 12:51 |
| dtantsur | so it may be just some "callback" method for deploy interface | 12:51 |
| *** abrito has quit IRC | 12:51 | |
| lucasagomes | dtantsur, yeah, but first I think we need to have IPA as the default ramdisk | 12:51 |
| lucasagomes | and drop the DIB | 12:52 |
| TheJulia | bcornec: enjoy! :) | 12:52 |
| lucasagomes | so it will be a common thing for all drivers | 12:52 |
| dtantsur | lucasagomes, that's good (and we should do it asap), but we still need to generalize it somehow in case we'll get more deployment methods | 12:52 |
| lucasagomes | otherwise we will endup with driver specifics in the core API | 12:52 |
| lucasagomes | (which is not ideal) | 12:52 |
| *** achanda has joined #openstack-ironic | 12:52 | |
| dtantsur | lucasagomes, or do you think we should just make "agent" the only possible deployment method | 12:52 |
| lucasagomes | dtantsur, I think we should make the IPA the default ramdisk | 12:53 |
| lucasagomes | agent and iscsi methodologies should stay | 12:53 |
| lucasagomes | cause they are different | 12:53 |
| lucasagomes | and both are supported by IPA, and both uses lookup and heartbeat | 12:53 |
| dtantsur | hmm, so we can just promote them? | 12:54 |
| lucasagomes | +1 | 12:54 |
| dtantsur | aha | 12:54 |
| lucasagomes | but I think we also should drop DIB | 12:54 |
| lucasagomes | the bash ramdisk and remove those pass_deploy_info, pass_bootloader_info methods | 12:54 |
| lucasagomes | we need to deprecate it for 1 cycle (this one) | 12:54 |
| *** ukalifon2 has joined #openstack-ironic | 12:55 | |
| dtantsur | lucasagomes, I think about writing (2 more) specs: on deprecating DIB ramdisk and on promoting these methods. wdyt? | 12:55 |
| *** ukalifon has quit IRC | 12:55 | |
| lucasagomes | dtantsur, right. It works. Not sure if we need a spec to deprecate DIB tho. Perhaps it's good just to collect comments/concerns | 12:56 |
| dtantsur | yeah, to agree on some plan | 12:57 |
| lucasagomes | ML might be an option too | 12:58 |
| *** [1]cdearborn has joined #openstack-ironic | 12:58 | |
| dtantsur | lucasagomes, does agent driver even support local boot? | 13:01 |
| dtantsur | I can't find any support code in Ironic... | 13:01 |
| lucasagomes | dtantsur, it does yes | 13:02 |
| lucasagomes | it does support everything DIB ramdisk supports (when I was porting stuff) | 13:02 |
| lucasagomes | local boot, root device hints | 13:02 |
| lucasagomes | dtantsur, https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/extensions/image.py#L176 | 13:03 |
| dtantsur | lucasagomes, agent.py does not seem to call configure_local_boot: https://github.com/openstack/ironic/search?utf8=%E2%9C%93&q=configure_local_boot | 13:03 |
| * lucasagomes looks | 13:05 | |
| lucasagomes | dtantsur, oh agent.py is only full disk images | 13:05 |
| *** abrito has joined #openstack-ironic | 13:06 | |
| lucasagomes | for the iscsi methodology (pxe.py) it does support local boot just like DIB | 13:06 |
| dtantsur | ok thanks | 13:06 |
| lucasagomes | https://github.com/openstack/ironic/blob/309707bab00a8a30d473815687515ac906ffbeb2/ironic/drivers/modules/pxe.py#L625 | 13:06 |
| lucasagomes | jroll, does agent already supports partition images? | 13:06 |
| *** heyongli has quit IRC | 13:07 | |
| *** heyongli has joined #openstack-ironic | 13:08 | |
| *** gabriel-bezerra has joined #openstack-ironic | 13:08 | |
| *** dprince has joined #openstack-ironic | 13:09 | |
| trown | good ugt morning ironicistas | 13:09 |
| dtantsur | trown, morning! | 13:09 |
| trown | dtantsur: I started hacking on the edeploy plugin bug friday evening | 13:10 |
| dtantsur | great! I'd like it to be fixed with all other breaking changes | 13:10 |
| dtantsur | trown, did you see my comment about renaming? | 13:11 |
| trown | dtantsur: wdyt about swiftclient dependency? Should we use conditional import only in "extra_hardware" plugin? | 13:11 |
| trown | dtantsur: ya +1 to that | 13:11 |
| *** rloo has joined #openstack-ironic | 13:12 | |
| dtantsur | trown, I think we should put it in test-requirements, and have a usual import in the plugin | 13:12 |
| *** dguerri is now known as dguerri` | 13:12 | |
| dtantsur | and of course document the dependency in the README (where we document plugins) | 13:12 |
| dtantsur | morning, rloo! | 13:13 |
| trown | dtantsur: ah, ok that makes sense | 13:13 |
| rloo | Happy Monday morning ironic'ers! howdy dtantsur! | 13:14 |
| dtantsur | :) | 13:14 |
| trown | good morning rloo | 13:14 |
| sambetts | morning rloo | 13:17 |
| TheJulia | good morning | 13:18 |
| TheJulia | so, does being an ironicista mean I get to wears some sort of cute dress and a berret? | 13:19 |
| *** achanda has quit IRC | 13:20 | |
| trown | TheJulia: I do not see why not | 13:20 |
| *** achanda has joined #openstack-ironic | 13:21 | |
| TheJulia | I like this idea. We need pixie boots on berrets! | 13:22 |
| *** dguerri` is now known as dguerri | 13:24 | |
| jroll | lucasagomes: we use sysrq because reboot (at the time? still?) didn't work correctly from the container | 13:25 |
| jroll | Mark_Melb: are you on virt or real hardware? coreos image or DIB image? | 13:26 |
| lucasagomes | jroll, gotcha... yeah that was the assumption by ramesh | 13:26 |
| lucasagomes | jroll, trown TheJulia good ugt morning | 13:26 |
| *** bradbeam has joined #openstack-ironic | 13:26 | |
| openstackgerrit | Yuriy Zveryanskyy proposed openstack/ironic: Remove auth token saving from iLO driver https://review.openstack.org/187924 | 13:26 |
| *** achanda has quit IRC | 13:26 | |
| jroll | lucasagomes: yeah, indeed | 13:26 |
| *** yuikotak_dinner has quit IRC | 13:27 | |
| jroll | lucasagomes: and no, agent doesn't support partition images yet, someone was working on splitting partitioning code out to ironic-lib and then using that from the agent as well | 13:27 |
| jroll | and good morning to all :) | 13:27 |
| *** bradbeam has quit IRC | 13:27 | |
| lucasagomes | jroll, yup | 13:27 |
| lucasagomes | gotcha thanks | 13:27 |
| jroll | dunno the status | 13:28 |
| jroll | vdrok: were you the one working on that? | 13:28 |
| vdrok | jroll, no, these were just talks :) | 13:28 |
| vdrok | morning | 13:29 |
| jroll | hmm | 13:29 |
| * jroll checks ironic-lib commits | 13:29 | |
| jroll | faizan maybe | 13:29 |
| dtantsur | jroll, morning | 13:30 |
| dtantsur | ironic-lib commits require review :) | 13:31 |
| jroll | morning :) | 13:31 |
| jroll | yeah | 13:31 |
| jroll | it would be nice to get a chain of patches that brings ironic-lib up to date and also switches ironic to use it | 13:32 |
| jroll | and land it all at once | 13:32 |
| jroll | otherwise I fear we won't keep up | 13:32 |
| rloo | dtantsur: would it make sense to update your dashboard (http://perm.ly/ironic-review-dashboard) to include ironic-lib | 13:34 |
| *** saripurigopi has joined #openstack-ironic | 13:34 | |
| dtantsur | rloo, yeah, right. I've switched to gertty since them, I know jlvillal was patching our dashboard recently | 13:34 |
| *** krotscheck has joined #openstack-ironic | 13:34 | |
| rloo | dtantsur: ok. cuz i totally forgot about ironic-lib. | 13:35 |
| openstackgerrit | Ruby Loo proposed openstack/ironic: Don't support deprecated drivers' vendor_passthru https://review.openstack.org/188538 | 13:37 |
| *** igordcard_ has joined #openstack-ironic | 13:38 | |
| *** yuikotak_dinner has joined #openstack-ironic | 13:38 | |
| *** krotscheck has quit IRC | 13:38 | |
| *** krotscheck has joined #openstack-ironic | 13:39 | |
| openstackgerrit | Jarrod Johnson proposed stackforge/pyghmi: Provide interface for decoding PET alerts https://review.openstack.org/188920 | 13:40 |
| *** krotscheck has quit IRC | 13:40 | |
| *** mtanino has quit IRC | 13:40 | |
| *** krotscheck has joined #openstack-ironic | 13:41 | |
| *** chenglch has quit IRC | 13:42 | |
| openstackgerrit | Jarrod Johnson proposed stackforge/pyghmi: Provide interface for decoding PET alerts https://review.openstack.org/188920 | 13:42 |
| rloo | hey dtantsur, before I forget. wrt discoverd -> inspector. this should be changed: http://docs.openstack.org/kilo/config-reference/content/ch_configuring-openstack-bare-metal.html | 13:48 |
| rloo | dtantsur: see table 1.14 | 13:48 |
| rloo | dtantsur: i was going to say, remind me when it changes to inspector, but it already changed in the code, right? | 13:49 |
| dtantsur | rloo, that's for kilo, we have old configuration options there | 13:53 |
| rloo | dtantsur: yeah, I meant for the next (liberty) version. | 13:53 |
| dtantsur | rloo, I think we did change these options... | 13:53 |
| rloo | dtantsur: that process isn't fully automated. I had to change something to put the 'discoverd' table there. | 13:53 |
| *** yuikotak_dinner is now known as yuikotakada | 13:53 | |
| rloo | dtantsur: is it 'inspector' or 'ironic-inspector'? | 13:53 |
| dtantsur | rloo, oh... it's just inspector | 13:54 |
| rloo | dtantsur: i'll make the change this week or today if i get around to it ;) | 13:54 |
| dtantsur | https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/inspector.py#L36-L52 | 13:54 |
| dtantsur | rloo, thanks! | 13:54 |
| openstackgerrit | Jarrod Johnson proposed stackforge/pyghmi: Provide interface for decoding PET alerts https://review.openstack.org/188920 | 13:55 |
| *** e0ne is now known as e0ne_ | 13:56 | |
| *** bethelwell has joined #openstack-ironic | 14:01 | |
| BadCub-Hiding | morning folks | 14:01 |
| openstackgerrit | Yuiko Takada proposed stackforge/ironic-discoverd: Migrate to oslo_db https://review.openstack.org/181905 | 14:05 |
| *** BadCub-Hiding is now known as BadCub | 14:05 | |
| dtantsur | BadCub-Hiding, morning! | 14:05 |
| BadCub | morning dtantsur | 14:07 |
| *** e0ne_ has quit IRC | 14:07 | |
| *** e0ne has joined #openstack-ironic | 14:08 | |
| openstackgerrit | Merged stackforge/pyghmi: Provide interface for decoding PET alerts https://review.openstack.org/188920 | 14:10 |
| *** logan2 has quit IRC | 14:12 | |
| *** achanda has joined #openstack-ironic | 14:12 | |
| *** logan2 has joined #openstack-ironic | 14:13 | |
| Mark_Melb | jroll: Using Devstack with real hardware (Dell m6X blades). Using the latest CoreOS image. | 14:14 |
| jroll | Mark_Melb: odd, I wonder if it's something to do with the blades | 14:15 |
| openstackgerrit | Vladyslav Drok proposed openstack/ironic: Allow referring to Glance images by name https://review.openstack.org/188429 | 14:15 |
| Mark_Melb | jroll: Could be. I have been able to load up the cirros full disk before and just a centos6 image but Im stuck now. Im wondering moving away from IPA and using pxe_drac is a good idea. | 14:17 |
| jroll | Mark_Melb: IPA is the future, it's essentially our default at this point | 14:18 |
| jroll | this is a bug | 14:18 |
| *** achanda has quit IRC | 14:18 | |
| jroll | and we've seen it before and (thought we) fixed it | 14:18 |
| Mark_Melb | Maybe there is something I have missed that is out of date. I got the CoreOS image from the Openstack tarball link and it said the files were from June 4. | 14:19 |
| Mark_Melb | Devstack i deployed maybe 2-3 weeks ago. Maybe I need to re-do and get all the latest code for everything? | 14:19 |
| jroll | should be recent enough -- looking for the bug report from before | 14:20 |
| Mark_Melb | The one from the end of March? | 14:20 |
| jroll | here was the fix | 14:20 |
| jroll | https://github.com/openstack/ironic-python-agent/commit/877f66826cd1b50163c67e73c8ebb4590c0f7ec8 | 14:20 |
| jroll | bug 1418833 | 14:20 |
| openstack | bug 1418833 in Ironic "check-tempest-dsvm-ironic-agent_ssh fails due to ConfigDriveWriteError" [Critical,Fix released] https://launchpad.net/bugs/1418833 - Assigned to Jay Faulkner (jason-oldos) | 14:20 |
| lucasagomes | Mark_Melb, jroll I might be out of context, but you can use pxe_drac with IPA if you want | 14:22 |
| lucasagomes | (IPA the ramdisk) | 14:22 |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: Add node() and ports() to NodeInfo https://review.openstack.org/189310 | 14:23 |
| Mark_Melb | So, it would end be agent_drac? Or, when pxe_drac boots it boots into the IPA ramdisk? | 14:30 |
| *** rameshg87 has joined #openstack-ironic | 14:31 | |
| Mark_Melb | jroll, lucasagomes: The problem I am seeing that may be related to copy_configdrive_to_disk.sh does that script run from the CoreOS ramdisk for example this: coreos_production_pxe_image-oem.cpio.gz ? | 14:32 |
| lucasagomes | Mark_Melb, so all pxe_ drivers support using the IPA ramdisk instead of the DIB one | 14:34 |
| Mark_Melb | Im just wondering because I found a copy of the config drive script on my Devstack instance that does not match what is current in github now. | 14:34 |
| lucasagomes | Mark_Melb, so the deployment will have some differences, like the image will be copied over iSCSI, the partitioning (if it's a partition image) will happen in the ironic conductor itself etc | 14:34 |
| Mark_Melb | Ok, so ironic conductor process will mount it via iSCSI and sort it out from there where as IPA will image locally via its REST calls. | 14:35 |
| rameshg87 | jroll: so do you still suggest to use sysrq to trigger power off in https://review.openstack.org/#/c/189241/ ? | 14:36 |
| * rameshg87 hasn't tested with coreos ramdisk yet | 14:37 | |
| jroll | rameshg87: yes, we do that because containers | 14:38 |
| *** jamielennox is now known as jamielennox|away | 14:38 | |
| *** jgrimm is now known as zz_jgrimm | 14:38 | |
| lucasagomes | Mark_Melb, yeah... IPA (the ramdisk) supports both deploy methodologies | 14:38 |
| jroll | Mark_Melb: yeah, that's exactly where the problem is... I'm curious where you doesnloaded the ramdisk from if it's using an old version of that | 14:39 |
| lucasagomes | Mark_Melb, I usually call them iscsi vs agent methodology | 14:39 |
| lucasagomes | where iscsi is: IPA export the local disk as iscsi and the ironic-conductor does the work of imaging it (partitioning, copying the image, copying the config drive etc...) | 14:39 |
| Mark_Melb | jroll, here is where I got them http://tarballs.openstack.org/ironic-python-agent/coreos/files/ | 14:40 |
| lucasagomes | and agent it's done inband by the IPA ramdisk, fetching the image from HTTP (be it a temp url or just an external http server), copying this image to the disk (only full disk supported) and creating an config drive partition | 14:40 |
| jroll | Mark_Melb: ok, great, and where did you find the older version? | 14:41 |
| rameshg87 | jroll: JayF: JoshNang: can you guys review https://review.openstack.org/#/c/187111/ | 14:41 |
| Mark_Melb | jroll, ‘opt/stack/ironic-python-agent/ironic_python_agent/shell/copy_configdrive_to_disk.sh’ | 14:41 |
| Mark_Melb | I may have cloned that to look at. I’m not sure. | 14:41 |
| rameshg87 | eventhough we might not use reboot command, we need to enable sysrq function control for operations because it's not enabled by default on some systems | 14:41 |
| jroll | Mark_Melb: and in local.conf or whatever, what is IRONIC_BUILD_DEPLOY_RAMDISK set to? | 14:42 |
| jroll | rameshg87: yeah, I'm happy to review | 14:42 |
| rameshg87 | jroll: and please confirm if my reply comment is correct | 14:42 |
| jroll | rameshg87: though I want to get away from the culture of "only rackspace folks know IPA well enough to review", I think other people know it well enough too, whether they believe it or not :) | 14:42 |
| rameshg87 | yeah I agree :) | 14:43 |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: API: Get a subset of fields from Ports and Chassis https://review.openstack.org/183549 | 14:43 |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: API: Get a subset of fields from Nodes https://review.openstack.org/183491 | 14:43 |
| Mark_Melb | jroll that is not currently set in l | 14:43 |
| * rameshg87 should ping other people too :) | 14:43 | |
| Mark_Melb | jroll: That is not currently set in my localrc. | 14:44 |
| rameshg87 | dtantsur: can you plesae restore your +2 if it looks fine on this patch - https://review.openstack.org/#/c/187111/ . only commit message was updated | 14:44 |
| * dtantsur is looking | 14:44 | |
| jroll | rameshg87: +2 with a link to docs :) | 14:44 |
| rameshg87 | jroll: thanks :) | 14:44 |
| *** boris-42 has joined #openstack-ironic | 14:45 | |
| boris-42 | dtantsur: ping | 14:45 |
| jroll | Mark_Melb: ok, default is true on that, which means it builds from source. which means it should be grabbing from your local copy -- could be the problem | 14:45 |
| dtantsur | boris-42, hi! | 14:45 |
| jroll | Mark_Melb: I'd set that to False, but also blow away devstack to be sure there's no caching going on etc | 14:45 |
| Mark_Melb | jroll: Excellent, thank you will try that. | 14:46 |
| jroll | Mark_Melb: np :) | 14:46 |
| * dtantsur brb | 14:46 | |
| boris-42 | dtantsur: I am back LOL | 14:46 |
| *** NobodyCam has joined #openstack-ironic | 14:46 | |
| boris-42 | looking for Ironic expert=) | 14:46 |
| boris-42 | I hope right chat=) | 14:47 |
| *** dlpartain has joined #openstack-ironic | 14:47 | |
| jroll | boris-42: no ironic experts here, sorry :) | 14:47 |
| jroll | what's up? | 14:47 |
| boris-42 | haha=) | 14:47 |
| TheJulia | jroll: I think we left them in that OTHER chatroom ;) | 14:48 |
| NobodyCam | good morning Ironic | 14:48 |
| TheJulia | good morning NobodyCam | 14:48 |
| BadCub | lol | 14:48 |
| jroll | morning NobodyCam BadCub TheJulia etc etc :) | 14:49 |
| NobodyCam | morning TheJulia jroll BadCub and boris-42 | 14:49 |
| boris-42 | NobodyCam: hi there | 14:49 |
| *** saripurigopi_ has joined #openstack-ironic | 14:49 | |
| BadCub | mornign jroll TheJulia NobodyCam | 14:49 |
| boris-42 | so actually we find time to work on Ironic benchmarks | 14:49 |
| boris-42 | in Rally | 14:49 |
| boris-42 | so I need some kind of audit from Ironic team | 14:49 |
| boris-42 | do we do the right things | 14:49 |
| *** saripurigopi has quit IRC | 14:50 | |
| TheJulia | to help us understand context, what are you intending to benchmark, and on what substate will you be benchmarking? | 14:50 |
| *** saripurigopi_ is now known as saripurigopi | 14:50 | |
| openstackgerrit | Yuriy Zveryanskyy proposed openstack/ironic: Remove auth token saving from iLO driver https://review.openstack.org/187924 | 14:50 |
| lucasagomes | boris-42, also, https://review.openstack.org/#/c/137171/ | 14:51 |
| NobodyCam | morning lucasagomes | 14:51 |
| lucasagomes | NobodyCam, good ugt morning | 14:51 |
| NobodyCam | :) | 14:51 |
| boris-42 | TheJulia: the smallest set of what I am going to benchmark is *everything* | 14:51 |
| *** dlpartain has left #openstack-ironic | 14:52 | |
| boris-42 | TheJulia: actually rally can be used for functional tesing as well (it's like common tool for all kinds of testing) | 14:52 |
| boris-42 | here is the start of https://review.openstack.org/#/c/186064/ | 14:52 |
| TheJulia | so you have hardware on hand to perform deployment benchmarks across all driver/configuration configurations? | 14:52 |
| boris-42 | TheJulia: Cray has | 14:53 |
| boris-42 | TheJulia: I don't care about end results | 14:53 |
| boris-42 | I am making the process simple so others will do this | 14:53 |
| lucasagomes | boris-42, that spec is about adding metrics in various parts of the ironic code, api, db, rpc etc... Can you check if the work you want to do will somehow conflict with that | 14:54 |
| boris-42 | lucasagomes: Rally just calls Ironic as a Serivce under load of N users | 14:54 |
| boris-42 | lucasagomes: btw why not integrating osprofiler? | 14:54 |
| boris-42 | lucasagomes: seems like you are reinventing wheel... | 14:55 |
| lucasagomes | boris-42, not my spec, it's under review | 14:55 |
| lucasagomes | that's why I want to find out what are the other works going on in this area | 14:55 |
| boris-42 | lucasagomes: so there is osprofiler cross project / service distrubted tracer | 14:56 |
| boris-42 | I believe it's better choice=) | 14:56 |
| *** achanda has joined #openstack-ironic | 14:56 | |
| boris-42 | but in any case it's not ralted to benchmarking stuff | 14:56 |
| boris-42 | I will left some comments in that spec* | 14:56 |
| lucasagomes | boris-42, ack. thanks | 14:56 |
| boris-42 | so about benchmarking | 14:56 |
| *** lazy_prince has joined #openstack-ironic | 14:57 | |
| boris-42 | could somebody help us to write proper benchmarks? | 14:57 |
| boris-42 | from Ironic team? | 14:57 |
| lucasagomes | boris-42, sure, I believe people will help but we need something more substantial about what is being done (reads 186064) | 15:00 |
| * lucasagomes reads 186064 (I mean) | 15:00 | |
| *** Mark_Melb has quit IRC | 15:01 | |
| openstackgerrit | Ruby Loo proposed openstack/ironic: Install guide reflects changes on master branch https://review.openstack.org/189327 | 15:01 |
| *** achanda has quit IRC | 15:02 | |
| NobodyCam | oh nice catch rloo :) | 15:03 |
| dtantsur | NobodyCam, morning | 15:03 |
| NobodyCam | and good (UGT) morning :) | 15:03 |
| NobodyCam | morning dtantsur | 15:03 |
| rloo | NobodyCam: :) | 15:03 |
| NobodyCam | :) | 15:03 |
| jroll | boris-42: with the statsd metrics stuff, we want to track every request, not a subset, in production. this allows us to do so. last I read, osprofiler was not meant for this case | 15:04 |
| rloo | NobodyCam: that was cuz I was looking at the agenda for the meeting this week and docs is there. | 15:04 |
| NobodyCam | :) just one question: should the link to the conf options point to : http://docs.openstack.org/kilo/config-reference/content/ch_configuring-openstack-bare-metal.html | 15:05 |
| NobodyCam | or is that too kilo only | 15:05 |
| rloo | NobodyCam: no, cuz we most assuredly will be updating the configs on master, and that is kilo only | 15:05 |
| NobodyCam | ++ | 15:05 |
| NobodyCam | was just a thought | 15:05 |
| NobodyCam | :-p | 15:06 |
| rloo | NobodyCam: I did wonder about putting a link to kilo too but i think maybe we should leave that for when/if we have a kilo-specific version. | 15:06 |
| jroll | boris-42: and statsd/graphite are well-tested, well-understood systems with lots of options for real-time graphing, aggregation queries, etc etc | 15:06 |
| *** killer_prince has joined #openstack-ironic | 15:06 | |
| rloo | NobodyCam: eg, we've already moved discoverd to inspector | 15:06 |
| NobodyCam | thats on the index page of the wiki so I think it okay as is | 15:06 |
| NobodyCam | and yep !!! | 15:06 |
| boris-42 | jroll: so you are not interested in tracing | 15:06 |
| boris-42 | jroll: just tracking seprated operation in production on each request? | 15:07 |
| jroll | boris-42: no, I'm interested in live performance metrics | 15:07 |
| jroll | boris-42: 99th percentile api response time etc | 15:07 |
| boris-42 | jroll: not sure that API response time should be part of project itself* | 15:07 |
| *** jcoufal has quit IRC | 15:08 | |
| jroll | boris-42: and rpc request time and so on | 15:08 |
| boris-42 | jroll: btw this won't scale =) | 15:08 |
| TheJulia | I kind of think it must be in an operational toolkit when any difference in your substrate is going to cause differences that cannot be compared to a different configuration except through quite a bit of math and brainpower | 15:08 |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: Naming clean up https://review.openstack.org/189329 | 15:09 |
| cinerama | hi folks | 15:09 |
| *** lazy_prince has quit IRC | 15:09 | |
| jroll | boris-42: what won't scale? | 15:09 |
| boris-42 | jroll: I really do not why osprofiler was rejected in case of overhead and this won't be=) | 15:09 |
| TheJulia | good morning cinerama | 15:09 |
| BadCub | Morning cinerama | 15:09 |
| jroll | boris-42: we're running this in production today | 15:09 |
| dtantsur | cinerama, morning | 15:09 |
| boris-42 | jroll: what scale? 2 nodes? 20? 200? 2k? 200k? | 15:10 |
| jroll | boris-42: thousands | 15:10 |
| boris-42 | jroll: on thousands nodes you are collecting all this data in runtime and it works well? | 15:10 |
| jroll | boris-42: it's small udp payloads, and a couple of time.time() calls, nothing really blocking | 15:10 |
| jroll | boris-42: yes | 15:11 |
| boris-42 | jroll: not bad=) | 15:11 |
| boris-42 | jroll: in any case why you are doing this only for ironic?) | 15:11 |
| boris-42 | jroll: plus agree this is another case from osprofiler | 15:11 |
| jroll | boris-42: my team works on ironic, other teams here do this for other projects | 15:12 |
| jroll | e.g. swift does relatively the same thing | 15:12 |
| jroll | or do you mean for upstream? | 15:12 |
| boris-42 | jroll: yep | 15:12 |
| boris-42 | jroll: upstream | 15:12 |
| jroll | boris-42: only so much time in the day | 15:13 |
| boris-42 | jroll: =) | 15:13 |
| NobodyCam | morning cinerama | 15:13 |
| * dtantsur also remembers sending insane amount of stats to Zabbix for hundreds of instances | 15:13 | |
| jroll | boris-42: I'm an ironic developer and operator, my primary concern is how ironic is performing | 15:13 |
| boris-42 | jroll: okay so in any case | 15:13 |
| boris-42 | jroll: osprofiler is something different from this | 15:14 |
| jroll | boris-42: right, just pointing that out :) | 15:14 |
| boris-42 | jroll: it puprose is getting cross service trace (not measuring seprated operations) | 15:14 |
| boris-42 | jroll: lemme remove -1 | 15:14 |
| jroll | boris-42: indeed, that's why I wanted to make sure that was clear :) | 15:14 |
| *** dprince has quit IRC | 15:17 | |
| jroll | boris-42: I might ask why not stacktach for collection, but I need to step away and that could be a long conversation :) | 15:17 |
| boris-42 | jroll: in case of osprofiler? | 15:17 |
| boris-42 | jroll: or this thing?) | 15:17 |
| jroll | boris-42: osprofiler | 15:17 |
| boris-42 | jroll: it's just different | 15:17 |
| boris-42 | jroll: it doesn't allow you to create traces | 15:18 |
| boris-42 | jroll: sandywalsh even agree on that | 15:18 |
| jroll | boris-42: ok :) | 15:18 |
| boris-42 | jroll: the goal of osprofiler is nested traces over all land of projects of openstack | 15:18 |
| boris-42 | jroll: e.g. this http://boris-42.github.io/ngk.html | 15:18 |
| jroll | boris-42: wow, that's more than I assumed | 15:19 |
| jroll | awesome work | 15:19 |
| *** lazy_prince has joined #openstack-ironic | 15:19 | |
| boris-42 | jroll: it's clickable | 15:19 |
| boris-42 | jroll: and the idea is that this stuff extends openstack API | 15:21 |
| boris-42 | jroll: so I can integrate it with Rally for example=) and do load with tracing=) | 15:22 |
| *** killer_prince has quit IRC | 15:22 | |
| boris-42 | lucasagomes: thank you for review | 15:22 |
| lucasagomes | boris-42, no problem | 15:22 |
| *** killer_prince has joined #openstack-ironic | 15:23 | |
| jroll | boris-42: indeed, good stuff, I'd be excited to get it running in the gate and get perf impact on patches etc | 15:23 |
| *** yuikotakada has quit IRC | 15:24 | |
| *** killer_prince has quit IRC | 15:24 | |
| *** lazy_prince has quit IRC | 15:25 | |
| *** lazy_prince has joined #openstack-ironic | 15:26 | |
| boris-42 | jroll: so first step is let's finish rally-ironic integration | 15:27 |
| boris-42 | jroll: so you will have perf jobs in ironic like for example cinder has | 15:27 |
| *** saripurigopi has quit IRC | 15:28 | |
| boris-42 | jroll: e.g. http://logs.openstack.org/35/189135/2/check/gate-rally-dsvm-cinder/546cb99/rally-plot/results.html.gz | 15:28 |
| *** absubram has joined #openstack-ironic | 15:28 | |
| boris-42 | jroll: they run this set of benchmarks against each patch (if you use fake drivers, you can test infrastracture scalability) | 15:28 |
| *** rameshg87 has left #openstack-ironic | 15:29 | |
| *** leopoldj has quit IRC | 15:31 | |
| *** killer_prince has joined #openstack-ironic | 15:31 | |
| *** lazy_prince has quit IRC | 15:35 | |
| openstackgerrit | Vladyslav Drok proposed openstack/ironic-specs: Allow to refer to Glance image using its name https://review.openstack.org/189341 | 15:35 |
| *** yog__ has quit IRC | 15:37 | |
| *** mtanino has joined #openstack-ironic | 15:38 | |
| openstackgerrit | Merged openstack/ironic-python-agent: Make sure sysrq functions are enabled in reboot.sh https://review.openstack.org/187111 | 15:38 |
| *** Sukhdev has joined #openstack-ironic | 15:39 | |
| *** naohirot has joined #openstack-ironic | 15:42 | |
| *** saripurigopi has joined #openstack-ironic | 15:43 | |
| *** Nisha has joined #openstack-ironic | 15:44 | |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: Rework processing hook interface for 2.0.0 https://review.openstack.org/189346 | 15:45 |
| openstackgerrit | Stephanie Miller proposed openstack/bifrost: Ensure dependencies are present for dib image build https://review.openstack.org/186875 | 15:47 |
| dtantsur | g'night all! | 15:54 |
| *** dtantsur is now known as dtantsur|afk | 15:54 | |
| jlvillal | Good morning Ironic | 15:54 |
| *** wuhg has quit IRC | 15:57 | |
| *** killer_prince is now known as lazy_prince | 15:58 | |
| TheJulia | goood morning | 15:58 |
| *** devananda has joined #openstack-ironic | 15:59 | |
| pshige | morning! | 16:00 |
| lucasagomes | dtantsur|afk, g'night | 16:00 |
| lucasagomes | jlvillal, good ugt morning | 16:00 |
| devananda | g'morning, all | 16:00 |
| devananda | seems my bouncer got rebooted over the weekend | 16:01 |
| jroll | morning pshige, jlvillal, devananda :) | 16:01 |
| jlvillal | devananda: If your OS came up but not the bouncer. You might check out 'monit', if you haven't already. | 16:02 |
| jlvillal | I use 'monit' to make sure that znc gets started if it is ever not running. | 16:02 |
| lucasagomes | devananda, morning | 16:02 |
| devananda | jlvillal: mostly i'm too lazy, and this bouncer had been up for a couple years | 16:02 |
| jlvillal | :) | 16:02 |
| jroll | devananda: friendly reminder for neutron/ironic meeting that I totally forgot about until now | 16:02 |
| jlvillal | Argh, I forgot too | 16:03 |
| devananda | jroll: hah. yea, well, I have a standing conflicting meeting, but I will lurk | 16:03 |
| jroll | #openstack-meeting-4 | 16:03 |
| jroll | ha | 16:03 |
| jroll | ok | 16:03 |
| devananda | jroll: channel is m... thanks | 16:03 |
| *** romainh has left #openstack-ironic | 16:04 | |
| *** degorenko has joined #openstack-ironic | 16:07 | |
| *** jistr has quit IRC | 16:10 | |
| *** achanda has joined #openstack-ironic | 16:16 | |
| *** hemna_ is now known as hemna | 16:17 | |
| *** ukalifon2 has quit IRC | 16:17 | |
| *** yuriyz has quit IRC | 16:21 | |
| *** yuriyz has joined #openstack-ironic | 16:29 | |
| *** dprince has joined #openstack-ironic | 16:31 | |
| *** davideagnello has joined #openstack-ironic | 16:37 | |
| *** ijw has joined #openstack-ironic | 16:39 | |
| *** ijw has quit IRC | 16:40 | |
| *** ijw has joined #openstack-ironic | 16:40 | |
| *** achanda has quit IRC | 16:41 | |
| openstackgerrit | Vladyslav Drok proposed openstack/ironic-specs: Add pluggable credentials storage https://review.openstack.org/186056 | 16:42 |
| *** achanda has joined #openstack-ironic | 16:44 | |
| *** valkyrja has quit IRC | 16:44 | |
| JayF | mariojv: ^ | 16:44 |
| *** bnemec has quit IRC | 16:45 | |
| JayF | vdrok: mariojv is working on that similar thing (pluggable credentials storage) right now as well | 16:45 |
| *** bnemec has joined #openstack-ironic | 16:46 | |
| *** bethelwell has quit IRC | 16:50 | |
| Madasi | yep, was going to point that out | 16:50 |
| Madasi | both specs are pluggable solutions to storing creds, Vladyslav's focuses on credentials currently in the DB and looks at keystone as the first implemented storage option, Mario's focuses on config file credentials and Barbican as the first implemented option | 16:52 |
| Madasi | son't see any reason they couldn't work together | 16:52 |
| Madasi | https://review.openstack.org/#/c/188609 and https://review.openstack.org/#/c/186056 | 16:53 |
| vdrok | JayF, Madasi, yup, saw that, and then when mariojv could also add barbican as credentials storage option | 16:54 |
| vdrok | for in-DB credentials | 16:54 |
| Madasi | I'm all about pluggable credential storage in general | 16:55 |
| *** killer_prince has joined #openstack-ironic | 16:56 | |
| *** saripurigopi has quit IRC | 16:58 | |
| *** lazy_prince has quit IRC | 17:00 | |
| lucasagomes | folks calling it a day, have a good night everyone | 17:00 |
| TheJulia | goodnight lucasagomes | 17:00 |
| lucasagomes | night :-) | 17:00 |
| *** lucasagomes is now known as lucas-dinner | 17:01 | |
| *** derekh has quit IRC | 17:01 | |
| NobodyCam | night lucas-dinner | 17:01 |
| Madasi | vdrok: question aobut your spec, what would you think about storing something like the driver name in place of the credential in the db, instead of just a generic value like /*** ? | 17:03 |
| *** naohirot has quit IRC | 17:03 | |
| Madasi | it's just a thought, but might make it a little easier on an operator who wasn't aware of how the credential storage was setup | 17:04 |
| *** e0ne has quit IRC | 17:06 | |
| *** saripurigopi has joined #openstack-ironic | 17:07 | |
| *** bradjones has quit IRC | 17:10 | |
| *** bradjones has joined #openstack-ironic | 17:11 | |
| *** bradjones has quit IRC | 17:11 | |
| *** bradjones has joined #openstack-ironic | 17:11 | |
| JayF | interesting, because I think for mariojv's thing, we wanted creds for BMCs stored there too | 17:15 |
| JayF | maybe a misunderstanding? | 17:15 |
| mariojv | looking at the other spec | 17:15 |
| *** lauramoore has joined #openstack-ironic | 17:15 | |
| killer_prince | good night all.. | 17:16 |
| *** zz_natorious is now known as natorious | 17:16 | |
| *** killer_prince has quit IRC | 17:17 | |
| mariojv | vdrok: looks like you already have an implementation in progress, that's great | 17:18 |
| Madasi | It (mariojv's) did both config file and db creds, it just spent the most time on how it would work for config file creds, so I summarized as focusing there :) | 17:18 |
| mariojv | vdrok: i can modify my spec to just implement barbican off of your spec if that sounds good | 17:18 |
| Madasi | I think, as writtern, both will handle both types of creds | 17:19 |
| mariojv | Madasi, JayF: yep, that's what i was thinking, you can put creds in either | 17:19 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add command metrics to IPA API https://review.openstack.org/119981 | 17:19 |
| JayF | mariojv: that sounds like a good idea | 17:22 |
| mariojv | k will update later today | 17:22 |
| *** Marga_ has joined #openstack-ironic | 17:23 | |
| *** Marga_ has quit IRC | 17:27 | |
| *** Marga_ has joined #openstack-ironic | 17:27 | |
| *** trown is now known as trown|lunch | 17:29 | |
| *** dguerri is now known as dguerri` | 17:31 | |
| NobodyCam | JayF: see you had a +1 on https://review.openstack.org/#/c/187415 (BiFrost doc patch) happen to have a free minute to look at the latest rev? | 17:31 |
| openstackgerrit | Merged openstack/bifrost: Preserve logs in logs/ folder https://review.openstack.org/186581 | 17:31 |
| NobodyCam | and good morning :) | 17:31 |
| *** lsmola has quit IRC | 17:32 | |
| JayF | +1d | 17:32 |
| NobodyCam | JayF: :) Ty :) | 17:32 |
| *** vdrok_ has joined #openstack-ironic | 17:33 | |
| *** edcragg has quit IRC | 17:37 | |
| jlvillal | jroll: I see you got a +2 on: https://review.openstack.org/#/c/188652/ :) | 17:37 |
| jlvillal | jroll: Do you think a recheck is all that is needed? Or could there be a valid Jenkins failure? | 17:37 |
| jroll | jlvillal: hooray, validation from my peers | 17:38 |
| jlvillal | :) | 17:38 |
| *** lauramoore has quit IRC | 17:38 | |
| jroll | jlvillal: that job doesn't use the ironic driver, so probably a recheck | 17:38 |
| * jroll does it | 17:38 | |
| jlvillal | Thanks | 17:38 |
| vdrok_ | Madasi: mariojv JayF yes, there is an implementation here - https://review.openstack.org/185074 , mariojv if you don't have an your implementation please have a look at this, I'll try to make it work with configs | 17:38 |
| mariojv | sounds good vdrok_ | 17:39 |
| *** Sukhdev has quit IRC | 17:39 | |
| vdrok_ | mariojv: ok, thanks | 17:39 |
| *** Sukhdev has joined #openstack-ironic | 17:40 | |
| rloo | hey JoshNang, what's the relationship between https://review.openstack.org/#/c/188873/ and https://review.openstack.org/#/c/159322/ | 17:41 |
| *** spandhe has joined #openstack-ironic | 17:43 | |
| *** pelix has quit IRC | 17:51 | |
| *** zz_jgrimm is now known as jgrimm | 17:53 | |
| *** ijw_ has joined #openstack-ironic | 17:58 | |
| *** achanda_ has joined #openstack-ironic | 17:59 | |
| *** dontalton has joined #openstack-ironic | 18:00 | |
| *** ijw has quit IRC | 18:01 | |
| *** achanda has quit IRC | 18:02 | |
| *** e0ne has joined #openstack-ironic | 18:08 | |
| *** openstackgerrit has quit IRC | 18:09 | |
| *** openstackgerrit has joined #openstack-ironic | 18:09 | |
| *** vdrok_ has quit IRC | 18:13 | |
| *** krotscheck is now known as krotscheck_confe | 18:21 | |
| *** krotscheck_confe is now known as krotsch_at_con | 18:21 | |
| openstackgerrit | Jarrod Johnson proposed stackforge/pyghmi: Acknowledge PET alert on decode https://review.openstack.org/189392 | 18:23 |
| *** e0ne is now known as e0ne_ | 18:25 | |
| *** trown|lunch is now known as trown | 18:29 | |
| JoshNang | rloo: good question. i think the Add /nodes/<uuid>/cleaning/steps should be based on the exposing clean_step. clean_step is the current step, cleaning step is all steps it will take (which is dynamically generated and/or cached) | 18:29 |
| *** e0ne_ has quit IRC | 18:30 | |
| JayF | Current clean step not existing is a way larger operational pain than not being able to fetch steps | 18:31 |
| JayF | because nodes that *are* cleaning list all their steps on the node | 18:31 |
| JayF | but nothing api exposed to show which one its on | 18:31 |
| *** achanda_ has quit IRC | 18:32 | |
| *** erelwell has joined #openstack-ironic | 18:34 | |
| *** erelwell is now known as bethelwell | 18:34 | |
| *** Marga_ has quit IRC | 18:40 | |
| *** Sukhdev has quit IRC | 18:41 | |
| *** achanda has joined #openstack-ironic | 18:41 | |
| rloo | JoshNang: I don't know when to bump the microversion. Do we bump it now to expose node.clean_step, and bump it again when ../cleaning/steps API is available? | 18:43 |
| rloo | JoshNang: seems like if it was one feature, we'd bump the microversion when 'cleaning' feature was complete. | 18:44 |
| JoshNang | rloo: no idea. it depends on if we want to bump for purely additive changes or not | 18:44 |
| rloo | JoshNang: how hard is it to add in one patch (which is what I thought the original one did) exposing node.clean_step AND the new API endpoint? | 18:45 |
| *** e0ne has joined #openstack-ironic | 18:46 | |
| *** krotsch_at_con is now known as krotsck_at_con | 18:46 | |
| JoshNang | rloo: not hard. i just split it because they're related but independent features and smaller patches general > larger patches | 18:46 |
| JoshNang | *generally | 18:46 |
| *** achanda has quit IRC | 18:46 | |
| rloo | JoshNang: ok. in the old world, we'd allow two patches for them, so let's see what happens... | 18:47 |
| NobodyCam | brb | 18:47 |
| *** Marga_ has joined #openstack-ironic | 18:47 | |
| jroll | rloo: I think of it more like... people are already debating whether the new api endpoint should wait for zapping stuff etc, so let's split out the not-so-contentious bits in the meantime | 18:48 |
| JoshNang | rloo: so i think if we're going to do one bump for both, it makes the most sense to do it on the first patch. though i don't see a problem with bumping for both. | 18:48 |
| rloo | jroll, JoshNang: so it isn't clear how to deal with microversions and we shouldn't delay development cuz of it so it is ok, we can always bump the microversion :-) | 18:49 |
| jroll | rloo: yeah, I'm still on the side of "bump it always" especially after sdague's message on the list | 18:49 |
| JoshNang | ++ | 18:49 |
| JoshNang | people are either going to pin to one version or use latest. more granularity doesn't hurt | 18:50 |
| *** edcragg has joined #openstack-ironic | 18:51 | |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add command metrics to IPA API https://review.openstack.org/119981 | 18:53 |
| *** saripurigopi has quit IRC | 18:56 | |
| rloo | JoshNang: ok, i'm good with that. just a couple minor things in that patch. | 18:57 |
| *** romcheg has joined #openstack-ironic | 19:01 | |
| *** ijw_ has quit IRC | 19:01 | |
| *** ijw has joined #openstack-ironic | 19:02 | |
| *** saripurigopi has joined #openstack-ironic | 19:06 | |
| devananda | spreading a feature across multiple patches -> good. exposing only part of the API of an incomplete feature -> not so good. | 19:13 |
| devananda | JoshNang: imagine what happens if someone downstream were to deploy at a random point in the tree between the patch adding node.clean_step and the patch which adds /cleaning/steps -- seems like a rather strange API, maybe one we don't want to support, right? | 19:14 |
| devananda | JoshNang: so perhaps we should hide node.clean_step until the feature is complete (enough) to expose all the API | 19:14 |
| devananda | *changes | 19:14 |
| *** igordcard has joined #openstack-ironic | 19:15 | |
| jroll | ehhhhhhh | 19:17 |
| jroll | devananda: the api already returns the clean steps assigned to the node in driver_internal_info | 19:17 |
| jroll | there just isn't a "what clean steps can I expect for this node when not in cleaning" api | 19:17 |
| jroll | all this does is expose the clean_step field already on the node | 19:17 |
| devananda | jroll: I still occasionally question why we exposed that field in the API | 19:17 |
| jroll | devananda: as an operator, I care a whole lot about what clean step is currently executing on a node | 19:18 |
| devananda | jroll: totally agree with ^ | 19:18 |
| jroll | devananda: as an operator, I don't want to wait for zapping features to be finished to know that | 19:19 |
| devananda | jroll: I clearly missed something. why would /clean/steps wait on zapping? | 19:19 |
| jroll | idk, that's what I see happening. I think maybe that's the api to queue up zap steps or something too? | 19:19 |
| jroll | I need to look at the zapping stuff more | 19:20 |
| *** BadCub is now known as BadCub-Hiding | 19:20 | |
| devananda | ditto | 19:20 |
| jroll | but I don't see why we'd wait for anything to expose a simple db field | 19:20 |
| jroll | idk what this /clean/steps thing is meant to do, other than maybe control zapping | 19:21 |
| devananda | I would find it very strange if /cleaning/steps affected zapping -- shouldn't that be /zapping/foo ? | 19:21 |
| *** dguerri` is now known as dguerri | 19:21 | |
| *** edcragg has quit IRC | 19:22 | |
| jroll | I mean, underneath they are clean steps with a certain priority | 19:22 |
| *** [1]cdearborn has quit IRC | 19:22 | |
| devananda | ? | 19:22 |
| jroll | priority:0 == executed at operator's will == zap step | 19:23 |
| jroll | AIUI | 19:23 |
| jroll | https://review.openstack.org/#/c/185122/5/specs/liberty/implement-zapping-states.rst | 19:24 |
| devananda | my initial response: that is a confusing API | 19:24 |
| jroll | that's what I said when people first proposed a zap/clean split :) | 19:25 |
| *** mgoddard has quit IRC | 19:28 | |
| *** mgoddard has joined #openstack-ironic | 19:29 | |
| rloo | jroll, devananda: node.clean_step shows the current clean step being performed (if any) | 19:30 |
| openstackgerrit | Christopher Dearborn proposed openstack/ironic: Fix DRAC driver job completion detection https://review.openstack.org/189409 | 19:30 |
| rloo | jroll, devananda: or zap step I guess | 19:30 |
| rloo | jroll, devananda: the cleaning/steps API is to show all clean steps with their priorites or whatever. | 19:30 |
| jroll | rloo: right. I think these are two (slightly) separate things, and I'm not opposed to a separate patch for them. | 19:30 |
| jroll | as a note: the lack of node.clean_step being exposed took me by surprise when we (recently) deployed this code | 19:31 |
| rloo | jroll: right, they are separate, but they are both part of the 'cleaning' feature. I would have liked one version bump for the 'clean' feature. but I'm not fussy. | 19:31 |
| jroll | and is a huge pain point for me as an operator right now | 19:31 |
| rloo | jroll: so getting that info in driver_internal_info isn't good enough? | 19:32 |
| jroll | but like... if people don't like the separate patch, whatever, I'll hang on to it downstream even though it makes me sad | 19:32 |
| jroll | rloo: that doesn't expose the step currently runnign | 19:32 |
| rloo | jroll: I think it does. I thought it was the first clean step in there. | 19:32 |
| jroll | rloo: that's what I thought too, apparently not? | 19:32 |
| jroll | rloo: unless we did something broken in our shim that munges things from downstream to upstream... but I don't think we did | 19:33 |
| rloo | jroll: I've never run the code and looked. i thought so based on something JoshNang said and/or based on looking at the code. | 19:33 |
| jroll | rloo: me too :) | 19:33 |
| * jroll looks at code | 19:34 | |
| rloo | jroll: OH | 19:34 |
| * jroll listens | 19:35 | |
| openstackgerrit | Merged openstack/ironic: Don't support deprecated drivers' vendor_passthru https://review.openstack.org/188538 | 19:37 |
| devananda | rloo: is ^ documented and clearly messaged somewhere as a "hey, we're breaking the driver API" ? | 19:38 |
| *** dprince has quit IRC | 19:39 | |
| rloo | devananda: there was a warning in the logs, and it was documented in kilo notes somewhere. sec. | 19:39 |
| devananda | rloo: cool. still, there should be an email to the ML notifying people that it was finally removed | 19:39 |
| rloo | devananda: https://wiki.openstack.org/wiki/ReleaseNotes/Kilo#OpenStack_Bare_Metal_service_.28Ironic.29 | 19:40 |
| rloo | devananda: under 'Upgrade Notes' | 19:40 |
| rloo | devananda: although I see that it doesn't say *when* it will be removed | 19:40 |
| rloo | devananda: oh, i can do that. i don't recall doing that in the past. | 19:41 |
| rloo | devananda: that == send an email to the ML. | 19:41 |
| rloo | devananda: should we be sending email when something is deprecated too? | 19:41 |
| devananda | rloo: thanks. we're not perfect -- but yes, we should | 19:41 |
| devananda | both to note "hey, this will be removed" and then "hey, we removed this thing, like we told you we would" | 19:41 |
| rloo | devananda: ok, email when deprecated & email after removal. gotcha. | 19:42 |
| devananda | cheers | 19:42 |
| rloo | devananda: I should have said 'yeah, I was waiting for it to be merged and then was going to send out an email'. ha ha. | 19:42 |
| *** Nisha has quit IRC | 19:44 | |
| devananda | rloo: :-D | 19:47 |
| devananda | jroll: I'm digging into that spec now | 19:47 |
| devananda | jroll: also, I hate that some drivers can't return the clean/zap steps outside of when the agent is booted | 19:47 |
| devananda | jroll: can we please just cache that info? | 19:47 |
| jroll | devananda: I'm pretty far removed from that spec, don't ask me. I think that was proposed at one point but I am not surwe | 19:48 |
| JoshNang | sorry, was at lunch | 19:48 |
| JoshNang | devananda: yup, i added the caching bit to the zapping spec. its painful otherwise | 19:48 |
| devananda | JoshNang: it's unusable otherwise. we shouldn't even support not caching it | 19:52 |
| JoshNang | it does work with always running agents, but yeah, we don't actually support that yet. i'll drop the not-cached bit | 19:53 |
| openstackgerrit | Merged openstack/python-ironicclient: Install the ironicclient into the 'venv' virtualenv https://review.openstack.org/188550 | 19:56 |
| *** Marga_ has quit IRC | 19:57 | |
| devananda | JoshNang: how do you feel about the complexity of this JSON document? | 20:00 |
| *** dprince has joined #openstack-ironic | 20:01 | |
| JoshNang | devananda: not terribly happy | 20:01 |
| devananda | seems like we're asking every client to design and specify the steps for zapping every time -- rather than having the operator do it once | 20:01 |
| devananda | s/seems like// | 20:01 |
| devananda | and that seems like not the API I would want, as either an operator or a user | 20:01 |
| JoshNang | right | 20:03 |
| JoshNang | it's very flexible, but probably too flexible/not well defined | 20:03 |
| devananda | I'm going to think through this (in IRC) for a few minutes -- and I'd like to know if ya'll have a different POV | 20:04 |
| JoshNang | alright | 20:04 |
| devananda | as an operator-of-an-ironic-cloud, I want to define the process that each node goes through to make it available to Nova, and the process it goes through during provisioning/deprovisioning | 20:05 |
| devananda | and I want some guarantee that it will be consistent (at least for all nodes of the same "type") | 20:05 |
| openstackgerrit | Stephanie Miller proposed openstack/bifrost: Add cookiecutter-generated files for project https://review.openstack.org/188959 | 20:05 |
| devananda | do I want that process defined via OpenStack's REST API? or do I want that process defined via JSON files which I keep under version control, which I then pass to Ironic every time? | 20:06 |
| JoshNang | i lean towards API driven whenever possible | 20:08 |
| devananda | ^^^ Case #1 | vvv Case #2 | 20:08 |
| devananda | as an operator-using-ironic-to-manage-hardware, I want to do <things> to some subset of <machines> | 20:09 |
| devananda | and i want to automate it, so I can do it to some other subset of machines, again and again | 20:10 |
| devananda | --- end | 20:11 |
| *** pcrews has quit IRC | 20:12 | |
| devananda | JoshNang: oh, another aspect - API visible auditing. How do I, as a user of Ironic, know that the driver actually did the things I asked it to do? | 20:14 |
| openstackgerrit | Christopher Dearborn proposed openstack/ironic: Add additional required RPMs to dev instructions https://review.openstack.org/185993 | 20:14 |
| devananda | JoshNang: I'm not sure thta's important, but if I'm POSTing a JSON doc to Ironic, I'm going to want some way to tell that it actually did each of the things I asked for | 20:15 |
| JoshNang | devananda: other than logs? i'm not sure we do that anywhere other than showing state changes | 20:16 |
| devananda | JoshNang: right. so, if it failed part way through, how does the APi reflect the full list that I requested? | 20:16 |
| *** bethelwell has quit IRC | 20:16 | |
| openstackgerrit | John Trowbridge proposed stackforge/ironic-discoverd: [WIP] Fix edeploy plugin puts too much data in Ironic extra column https://review.openstack.org/189427 | 20:16 |
| devananda | assume that, since it's just a JSON document, I will forget it immediately after passing it to the API, and expect that I can later retrieve it from the API | 20:17 |
| JoshNang | devananda: they have to be stored in the node to be able to iterate them, we'll have to keep track of where in the list we are, so yeah it can be shown | 20:17 |
| devananda | cool | 20:17 |
| devananda | didn't see that in the spec | 20:17 |
| JoshNang | are you asking for the exact doc you passed in back? or how ironic translates it to a list of zap steps (which should be very similar) | 20:18 |
| JoshNang | the latter is what i was going for | 20:18 |
| devananda | the latter -- though, the difference in formatting between those is going to be a pain point for users | 20:19 |
| *** romcheg has left #openstack-ironic | 20:19 | |
| devananda | who will need to automate three things: parsing the list of available clean steps, parsing that ^ document, and preparing the requested zapping steps | 20:20 |
| devananda | it'd be, you know, nice if those were all consistent :) | 20:20 |
| JoshNang | mmm i think they could be exactly the same looking at it | 20:21 |
| devananda | woot | 20:21 |
| *** dprince has quit IRC | 20:22 | |
| devananda | JoshNang: replied to spec just now - some more comments in line | 20:24 |
| JoshNang | devananda: been back and forth on the endpoints. whether /nodes/<>/cleaning/steps and /nodes/<>/zapping/steps, use query param, or clean_steps vs all_steps | 20:25 |
| devananda | query param really sits poorly with me | 20:25 |
| devananda | otoh, this makes sense from a usability POV to me: /cleaning/steps == list of things that happen during cleaning; /zapping/steps == list of available things that could happen during zapping | 20:26 |
| devananda | again, thinking from user POV, if I see a node in cleaning and want to know what it will do next, it makes sense to query /cleaning/steps | 20:27 |
| JoshNang | agreed. and if i want to figure out zapping, might not be clear i should look at the cleaning api | 20:27 |
| devananda | right | 20:28 |
| JoshNang | i'll it switch back | 20:28 |
| devananda | certainly if I see a node in state=ZAPPING, I'm going to think to look at /nodes/zapping/steps | 20:28 |
| JoshNang | the rest of the comments are reasonable | 20:30 |
| rloo | JoshNang, does /nodes/zapping/steps show all clean steps, even if they aren't used/configured for this particular node for zapping? | 20:31 |
| JoshNang | rloo: it should only show steps available for that node | 20:31 |
| rloo | JoshNang: seems like there are 3 things. 1. all possible steps; 2. steps configured for node; 3. current (if any) step | 20:32 |
| JoshNang | is all possible steps useful? | 20:32 |
| rloo | JoshNang: for discovery, yes. For me personally, don't care. | 20:32 |
| rloo | JoshNang: how does someone know what to specify as the zap steps? | 20:33 |
| JoshNang | ohh i misread | 20:33 |
| *** chuckC_ has joined #openstack-ironic | 20:34 | |
| JoshNang | so all 3 are met here: /nodes/<>/zapping/steps gives all possible steps for that node, steps being executed/already execute will be stored on the node in driver_internal_info, and current step as node.zap_step | 20:34 |
| devananda | JoshNang: I dislike using driver_internal_info for this | 20:35 |
| devananda | it's not internal info -- it's something the user asked for | 20:35 |
| JoshNang | mmm. do you want to do something like storing it in DII but exposing it as its own field in the api? or just create a new db field? | 20:36 |
| devananda | JoshNang: what about: GET /nodes/NNN/zapping/steps for the full list, GET /nodes/NNN/zapping/current for the current plan (or empty set if no plan) | 20:38 |
| devananda | JoshNang: oh, and this spec does not outline the changes to the node detail view (ie, the call which returns all subresources) | 20:38 |
| JoshNang | devananda: that could work | 20:39 |
| rloo | I was thinking GET drivers/zap_steps for the full list | 20:39 |
| devananda | rloo: it may vary by node, not by driver | 20:39 |
| rloo | I thought it depended on the driver. I mean the possible steps, not the steps used by a particular driver. | 20:39 |
| rloo | argh. not the steps used by a particular node. | 20:40 |
| rloo | so eg driver A supports steps 1,2,3,4. Node N uses driver A and is configured to do steps 1,3. | 20:40 |
| openstackgerrit | Christopher Dearborn proposed openstack/ironic: Add additional required RPMs to dev instructions https://review.openstack.org/185993 | 20:41 |
| devananda | rloo: agent_ipmitool for example -- on Node X, the agent does not detect a RAID card, and on Node Y it does | 20:42 |
| devananda | same driver, different node -> different available Zapping steps | 20:42 |
| JoshNang | right | 20:43 |
| JoshNang | rloo: are you looking for the complete list of zap step that could be executed, regardless of whether an individual node supports it? | 20:44 |
| rloo | JoshNang: yes. | 20:44 |
| devananda | JoshNang: IIUC, the way ya'll are implementing the agent[-driver], the actual driver code won't know what any of the possible steps are, and relies completely on the agent. | 20:45 |
| rloo | JoshNang: I thought that's what /nodes/<id>/zap-steps did based on your spec. But I thought that you wanted to change that to only show what steps are going to be done on that particular node. | 20:45 |
| devananda | right? | 20:45 |
| JoshNang | devananda: correct | 20:45 |
| rloo | it is the same problem with cleaning. (that only the agent knows) | 20:46 |
| JoshNang | rloo: so if a node doesn't have a raid card, it seems reasonable that the driver wouldn't return 'configure_raid' in /nodes/<id>/zap-steps | 20:46 |
| rloo | JoshNang: I'm probably confused. | 20:47 |
| *** pcrews has joined #openstack-ironic | 20:47 | |
| rloo | JoshNang: so ..nodes/id/zap-steps shows all possible steps for that node | 20:47 |
| rloo | JoshNang: and PUT nodes/../provision/zap <body> is subset from nodes/id/zap_steps | 20:47 |
| JoshNang | correct | 20:47 |
| rloo | JoshNang: so after you do a PUT how do you know what you had set? | 20:48 |
| *** [1]cdearborn has joined #openstack-ironic | 20:48 | |
| rloo | JoshNang: is that the /current that devananda suggested? | 20:48 |
| JoshNang | yeah: GET /nodes/NNN/zapping/current | 20:48 |
| JoshNang | was proposed | 20:48 |
| rloo | JoshNang: ok, just the wording bothers me then. | 20:49 |
| rloo | JoshNang: 'current' could be interpreted as the 'current step being processed' | 20:49 |
| *** Sukhdev has joined #openstack-ironic | 20:50 | |
| JoshNang | i don't see a reason we can't or wouldnt return 'these are the steps you going to/have alread executed, this is the step you are currently executing' in that response | 20:50 |
| *** achanda has joined #openstack-ironic | 20:51 | |
| rloo | JoshNang: ok, that would make it clear. | 20:51 |
| JoshNang | rloo: ok! or this info should be available in one way or another in the node details api, imo | 20:52 |
| JoshNang | s/or/and | 20:52 |
| rloo | JoshNang: I hesitate to use 'state' but that's what I think of it, the current state of zapping. ha ha. | 20:52 |
| JoshNang | heh | 20:52 |
| rloo | JoshNang: I think I had this concern before. That within a state, we could have various steps and folks might want to know what those are and where things are at. | 20:53 |
| JoshNang | devananda: back to the original question, do you have a better api than passing in json documents? | 20:53 |
| rloo | JoshNang: and I was worried that we'd start adding new variables for each different state to track those things. | 20:53 |
| JoshNang | heh and we have | 20:54 |
| devananda | JoshNang: this approach requires the client to interpret the JSON document and build a new one | 20:54 |
| *** Marga_ has joined #openstack-ironic | 20:54 | |
| devananda | JoshNang: what if each available clean|zap step was a resource? | 20:54 |
| JoshNang | devananda: that seems more REST-y | 20:54 |
| devananda | JoshNang: and I put a list of those resources? | 20:54 |
| devananda | JoshNang: that's what I'm thinking :) | 20:54 |
| devananda | JoshNang: however -- that alone won't work, because those resources can change. what if I switch to a new Agent, and it doesn't find the RAID card any longer? -- what does the resource whose URL I used last time mean now? | 20:55 |
| devananda | JoshNang: so I'm not sure how to make this more RESTy, but it feels quite unRESTy at the moment | 20:55 |
| JoshNang | devananda: it certainly is | 20:56 |
| *** achanda has quit IRC | 20:56 | |
| JoshNang | there's definitely some friction between ironic driving the bus and ipa being very configurable/dynamic | 20:59 |
| *** Marga_ has quit IRC | 20:59 | |
| *** Marga_ has joined #openstack-ironic | 20:59 | |
| devananda | JoshNang: yup | 21:05 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add command metrics to IPA API https://review.openstack.org/119981 | 21:10 |
| *** pcrews has quit IRC | 21:11 | |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add docs tests to default tox envlist in IPA https://review.openstack.org/189455 | 21:13 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/187680 | 21:14 |
| *** jgrimm is now known as zz_jgrimm | 21:17 | |
| jlvillal | I was looking at: http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html#step-3-install-the-client | 21:19 |
| jlvillal | Does the: tox -evenv -- echo 'done' | 21:19 |
| jlvillal | Work for anyone? | 21:19 |
| jlvillal | I always get an error about the echo part | 21:19 |
| jlvillal | The environment gets installed fine. | 21:19 |
| jroll | jlvillal: sounds like it depends on your definition of "work" :) | 21:20 |
| jlvillal | jroll: My definition must be too strict ;) | 21:20 |
| jroll | :P | 21:20 |
| jroll | `echo $?` probably says no, "venv works" says yes | 21:20 |
| *** SpamapS has quit IRC | 21:22 | |
| openstackgerrit | John L. Villalovos proposed openstack/ironic: Update docs for usage of python-ironicclient https://review.openstack.org/189469 | 21:22 |
| jlvillal | jroll: Thanks. Glad it isn't just me. | 21:22 |
| *** trown is now known as trown|outttypeww | 21:24 | |
| *** Marga_ has quit IRC | 21:37 | |
| openstackgerrit | John L. Villalovos proposed openstack/python-ironicclient: Make 'echo' work in the 'venv' environment https://review.openstack.org/189477 | 21:37 |
| *** lucas-dinner has quit IRC | 21:37 | |
| openstackgerrit | John L. Villalovos proposed openstack/ironic: Make 'echo' work in the 'venv' environment https://review.openstack.org/189482 | 21:40 |
| jlvillal | jroll: Hack to make the documentation correct :) ^^^^^^^^ | 21:41 |
| jroll | jlvillal: lol | 21:41 |
| *** EmilienM is now known as EmilienM|afk | 21:41 | |
| jroll | jlvillal: I think another way around that is something like commands = /bin/true || {posargs} | 21:42 |
| jroll | (maybe) | 21:42 |
| jlvillal | jroll: I'm not sure... | 21:43 |
| jlvillal | jroll: Or more accurately. I have no idea. | 21:43 |
| jroll | the only reason the echo is there is to make it not bail with "no commands given" or whatever, afaik | 21:43 |
| jlvillal | jroll: Okay. With that method the 'echo' doesn't actually get run though. Which maybe isn't important. | 21:46 |
| jroll | jlvillal: right, it's hacky | 21:47 |
| *** BadCub-Hiding is now known as BadCub | 21:47 | |
| *** ijw has quit IRC | 21:47 | |
| *** ijw has joined #openstack-ironic | 21:48 | |
| *** dprince has joined #openstack-ironic | 21:49 | |
| *** lucas-dinner has joined #openstack-ironic | 21:50 | |
| *** lucas-dinner has quit IRC | 21:50 | |
| *** pcrews has joined #openstack-ironic | 21:50 | |
| *** [1]cdearborn has quit IRC | 21:50 | |
| *** achanda has joined #openstack-ironic | 21:51 | |
| mrda | Morning Ironic | 21:54 |
| *** SpamapS has joined #openstack-ironic | 21:55 | |
| * mrda would welcome help from jroll and lucasagomes on nova reviews | 21:56 | |
| jlvillal | mrda: Good morning :) | 21:56 |
| jroll | mrda: hit me! | 21:57 |
| mrda | oh hai! | 21:57 |
| *** achanda has quit IRC | 21:57 | |
| * mrda returns from a long-weekend downunder, so it's a virtual Monday today | 21:57 | |
| jlvillal | mrda: It is a Monday here also ;) | 21:58 |
| mrda | ;) | 21:58 |
| mrda | So we align on day for once :) | 21:58 |
| *** achanda has joined #openstack-ironic | 21:59 | |
| *** achanda has quit IRC | 22:08 | |
| *** krotsck_at_con is now known as krotscheck | 22:08 | |
| NobodyCam | morning mrda :) | 22:09 |
| TheJulia | good morning mrda | 22:10 |
| mrda | o/ | 22:13 |
| *** ijw has quit IRC | 22:28 | |
| *** ijw has joined #openstack-ironic | 22:29 | |
| *** ijw has quit IRC | 22:30 | |
| jroll | mrda: so uh, what reviews did you need? | 22:30 |
| jroll | damn if I'm going to look at all nova reviews :P | 22:31 |
| *** penick has joined #openstack-ironic | 22:31 | |
| NobodyCam | jroll: ++ | 22:32 |
| NobodyCam | :-p | 22:32 |
| *** absubram has quit IRC | 22:38 | |
| *** ijw has joined #openstack-ironic | 22:44 | |
| mrda | jroll: sorry, catching up with email. jlvillal and I will push some reviews your way when they come about :) | 22:48 |
| jroll | mrda: no worries, was hoping to get some in before end of today | 22:48 |
| * jroll reviews ironic-y things instead | 22:48 | |
| * jlvillal thinks mrda is reading scrollback from Friday.... | 22:48 | |
| * mrda thinks jlvillal is right (so it was my Saturday :) | 22:48 | |
| *** Sukhdev has quit IRC | 22:50 | |
| *** e0ne has quit IRC | 23:05 | |
| *** dprince has quit IRC | 23:18 | |
| *** natorious is now known as zz_natorious | 23:19 | |
| *** Sukhdev has joined #openstack-ironic | 23:19 | |
| *** EmilienM|afk is now known as EmilienM | 23:20 | |
| *** ijw has quit IRC | 23:23 | |
| *** yuanying has joined #openstack-ironic | 23:25 | |
| *** hemna is now known as hemnafk | 23:32 | |
| *** zhenguo has joined #openstack-ironic | 23:35 | |
| *** chlong has joined #openstack-ironic | 23:36 | |
| *** rloo has quit IRC | 23:39 | |
| *** dontalton has quit IRC | 23:45 | |
| openstackgerrit | Stephanie Miller proposed openstack/bifrost: Add cookiecutter-generated files for project https://review.openstack.org/188959 | 23:49 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!