| devananda | hm? was on a call | 00:03 |
|---|---|---|
| jroll | don't look up | 00:03 |
| devananda | NobodyCam: so a scheduler hint to Nova based on chassis_id -- that's fine. | 00:04 |
| devananda | heh, too late | 00:04 |
| * devananda starts sharpening a knife | 00:04 | |
| NobodyCam | jroll: Shrews: easy +a on for ironicclient.. https://review.openstack.org/#/c/91585 <- only 600 lines... | 00:04 |
| jroll | lol | 00:04 |
| NobodyCam | oh | 00:04 |
| NobodyCam | lol | 00:04 |
| NobodyCam | jroll: it has two +2 now | 00:05 |
| NobodyCam | (thats the easy part) lol | 00:05 |
| jroll | definitely not in a good place to read 600 loc right now | 00:05 |
| NobodyCam | :) | 00:06 |
| NobodyCam | it is o-beer-o-clock here in seattle | 00:06 |
| jroll | indeed | 00:07 |
| lifeless | is the cache layer in ironic on by default? | 00:07 |
| lifeless | I'm seeing qemu-img converts when deploying the same image a few minutes apart | 00:08 |
| lifeless | making me wonder if there is a bug somewhere | 00:08 |
| *** Penick has quit IRC | 00:08 | |
| jroll | lifeless: might be some bad defaults for cache size/ttl | 00:08 |
| lifeless | hahahaha | 00:10 |
| lifeless | # Maximum size (in MiB) of cache for master images, including | 00:10 |
| lifeless | # those in use (integer value) | 00:10 |
| lifeless | #image_cache_size=1024 | 00:10 |
| lifeless | let me just fix that | 00:10 |
| jroll | :) | 00:10 |
| lifeless | really we should make it like 30% of the partition its on or something | 00:11 |
| lifeless | but for now, uh, 20GB sound ok? | 00:11 |
| jroll | I don't care, I don't use that cache :) | 00:11 |
| jroll | sounds reasonable | 00:12 |
| devananda | sure | 00:13 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Set a more generous default image cache size https://review.openstack.org/106905 | 00:14 |
| NobodyCam | lifeless: thank you for the . too in that ^^^ :) | 00:16 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Push the image cache ttl way up https://review.openstack.org/106906 | 00:21 |
| lifeless | and the 106905 one is kinda urgent - its actually default to smaller than a single tripleo image | 00:23 |
| *** hemna is now known as hemna__ | 00:23 | |
| lifeless | so we were rebuilding images multiple times in a single cluster deploy | 00:23 |
| lifeless | which is a pretty bad performance impact | 00:23 |
| jroll | you can set the option in the meantime, no? | 00:24 |
| JayF | lifeless: why 10080 minutes as the TTL? | 00:24 |
| JayF | lifeless: just seems arbitrary? Not really a magic number I'd be used to seeing somewhere like there :) | 00:25 |
| lifeless | jroll: yes but we try to run defaults :) | 00:25 |
| JayF | 168 hours :) | 00:25 |
| lifeless | JayF: 60*24*7 | 00:25 |
| lifeless | 10080 | 00:25 |
| JayF | oh. 7 weeks. | 00:25 |
| JayF | er | 00:25 |
| JayF | yeah | 00:25 |
| JayF | OK, I agree | 00:25 |
| jroll | lifeless: sure thing | 00:25 |
| jroll | :) | 00:25 |
| lifeless | 60 minutes -> 1 hr; 24 hrs -> 1 day; 7 days -> 1 week | 00:25 |
| mrda | Sounds like a comment is needed :) | 00:25 |
| JayF | mrda: or I need to math harder :) | 00:26 |
| JayF | lifeless: you have +1 from me on both of those, seem like reasonable default changes :) | 00:26 |
| mrda | if it's not immediately obvious... | 00:26 |
| * JayF <3 good defaults | 00:26 | |
| lifeless | if I had more time I'd doa patch to make 0 as ttl mean 'disabled' | 00:26 |
| jroll | I'm +1 on adding a comment | 00:26 |
| lifeless | let me do that | 00:26 |
| mrda | \o/ | 00:26 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Push the image cache ttl way up https://review.openstack.org/106906 | 00:27 |
| NobodyCam | lifeless: 106905 has pep8 failure :( | 00:29 |
| rloo | NobodyCam: wrt 91585 & sort order. up to user to do the wrong, err right thing. if they're using marker, presumably they'd use the same sort order throughout :-) | 00:29 |
| jroll | lol | 00:29 |
| jroll | so much for quick patches | 00:29 |
| *** rloo has quit IRC | 00:35 | |
| devananda | right - i've gotta finish a few things... see ya'll tmw! | 00:46 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Set a more generous default image cache size https://review.openstack.org/106905 | 00:54 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Push the image cache ttl way up https://review.openstack.org/106906 | 00:55 |
| *** nosnos has quit IRC | 01:13 | |
| *** eguz has joined #openstack-ironic | 01:38 | |
| *** eghobo has quit IRC | 01:42 | |
| *** eguz has quit IRC | 02:15 | |
| *** pcrews has quit IRC | 02:16 | |
| *** dwalleck has joined #openstack-ironic | 02:17 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add versioning to Agent decommission https://review.openstack.org/106859 | 02:22 |
| *** dwalleck has quit IRC | 02:26 | |
| *** openstackgerrit has quit IRC | 02:42 | |
| *** mdorman has quit IRC | 02:42 | |
| *** morgabra has quit IRC | 02:42 | |
| *** morgabra has joined #openstack-ironic | 02:43 | |
| *** openstackgerrit has joined #openstack-ironic | 02:44 | |
| *** mdorman has joined #openstack-ironic | 02:46 | |
| *** bvivek has joined #openstack-ironic | 02:46 | |
| *** ramineni has joined #openstack-ironic | 02:51 | |
| *** Haomeng has joined #openstack-ironic | 02:53 | |
| *** harlowja is now known as harlowja_away | 02:54 | |
| *** harlowja_away is now known as harlowja | 02:58 | |
| *** vinbs has joined #openstack-ironic | 02:59 | |
| *** bvivek has quit IRC | 03:00 | |
| *** eghobo has joined #openstack-ironic | 03:11 | |
| *** mdorman has quit IRC | 03:12 | |
| *** Shrews has quit IRC | 03:12 | |
| *** anteaya has quit IRC | 03:12 | |
| *** zul has quit IRC | 03:12 | |
| *** Haomeng has quit IRC | 03:12 | |
| *** openstackgerrit has quit IRC | 03:12 | |
| *** yuriyz has quit IRC | 03:12 | |
| *** kevinbenton has quit IRC | 03:12 | |
| *** gilliard has quit IRC | 03:12 | |
| *** coolsvap has quit IRC | 03:12 | |
| *** davidlenwell has quit IRC | 03:12 | |
| *** radsy has quit IRC | 03:12 | |
| *** krtaylor has quit IRC | 03:12 | |
| *** chuckC has quit IRC | 03:12 | |
| *** mikal has quit IRC | 03:12 | |
| *** ekarlso has quit IRC | 03:12 | |
| *** mkerrin has quit IRC | 03:12 | |
| *** NobodyCam has quit IRC | 03:12 | |
| *** GheRivero has quit IRC | 03:12 | |
| *** zer0c00l has quit IRC | 03:12 | |
| *** boris-42 has quit IRC | 03:12 | |
| *** dtantsur|afk has quit IRC | 03:12 | |
| *** romcheg has quit IRC | 03:12 | |
| *** hemna__ has quit IRC | 03:12 | |
| *** zigo has quit IRC | 03:12 | |
| *** adam_g has quit IRC | 03:12 | |
| *** aignatov has quit IRC | 03:12 | |
| *** Isotopp has quit IRC | 03:12 | |
| *** aweeks has quit IRC | 03:12 | |
| *** mgagne has quit IRC | 03:12 | |
| *** BadCub_ has quit IRC | 03:12 | |
| *** enikanorov_ has quit IRC | 03:12 | |
| *** lynxman has quit IRC | 03:12 | |
| *** agordeev has quit IRC | 03:12 | |
| *** LiveOne has quit IRC | 03:12 | |
| *** comstud has quit IRC | 03:12 | |
| *** Mikhail_D_wk1 has quit IRC | 03:12 | |
| *** stevebaker has quit IRC | 03:12 | |
| *** dguerri has quit IRC | 03:12 | |
| *** soren has quit IRC | 03:12 | |
| *** morgabra has quit IRC | 03:12 | |
| *** lazy_prince has quit IRC | 03:12 | |
| *** ifarkas has quit IRC | 03:12 | |
| *** tteggel has quit IRC | 03:12 | |
| *** keekz has quit IRC | 03:12 | |
| *** Ng has quit IRC | 03:12 | |
| *** ramineni has quit IRC | 03:17 | |
| *** ramineni has joined #openstack-ironic | 03:17 | |
| *** jrist has quit IRC | 03:18 | |
| *** jrist has joined #openstack-ironic | 03:20 | |
| *** Haomeng has joined #openstack-ironic | 03:20 | |
| *** openstackgerrit has joined #openstack-ironic | 03:20 | |
| *** morgabra has joined #openstack-ironic | 03:20 | |
| *** radsy has joined #openstack-ironic | 03:20 | |
| *** lazy_prince has joined #openstack-ironic | 03:20 | |
| *** Shrews has joined #openstack-ironic | 03:20 | |
| *** krtaylor has joined #openstack-ironic | 03:20 | |
| *** yuriyz has joined #openstack-ironic | 03:20 | |
| *** dtantsur|afk has joined #openstack-ironic | 03:20 | |
| *** romcheg has joined #openstack-ironic | 03:20 | |
| *** ifarkas has joined #openstack-ironic | 03:20 | |
| *** chuckC has joined #openstack-ironic | 03:20 | |
| *** boris-42 has joined #openstack-ironic | 03:20 | |
| *** hemna__ has joined #openstack-ironic | 03:20 | |
| *** kevinbenton has joined #openstack-ironic | 03:20 | |
| *** ekarlso has joined #openstack-ironic | 03:20 | |
| *** gilliard has joined #openstack-ironic | 03:20 | |
| *** zigo has joined #openstack-ironic | 03:20 | |
| *** coolsvap has joined #openstack-ironic | 03:20 | |
| *** davidlenwell has joined #openstack-ironic | 03:20 | |
| *** mikal has joined #openstack-ironic | 03:20 | |
| *** zer0c00l has joined #openstack-ironic | 03:20 | |
| *** GheRivero has joined #openstack-ironic | 03:20 | |
| *** NobodyCam has joined #openstack-ironic | 03:20 | |
| *** mkerrin has joined #openstack-ironic | 03:20 | |
| *** comstud has joined #openstack-ironic | 03:20 | |
| *** agordeev has joined #openstack-ironic | 03:20 | |
| *** lynxman has joined #openstack-ironic | 03:20 | |
| *** LiveOne has joined #openstack-ironic | 03:20 | |
| *** enikanorov_ has joined #openstack-ironic | 03:20 | |
| *** BadCub_ has joined #openstack-ironic | 03:20 | |
| *** aweeks has joined #openstack-ironic | 03:20 | |
| *** mgagne has joined #openstack-ironic | 03:20 | |
| *** Isotopp has joined #openstack-ironic | 03:20 | |
| *** aignatov has joined #openstack-ironic | 03:20 | |
| *** adam_g has joined #openstack-ironic | 03:20 | |
| *** soren has joined #openstack-ironic | 03:20 | |
| *** dguerri has joined #openstack-ironic | 03:20 | |
| *** stevebaker has joined #openstack-ironic | 03:20 | |
| *** Mikhail_D_wk1 has joined #openstack-ironic | 03:20 | |
| *** Ng has joined #openstack-ironic | 03:20 | |
| *** keekz has joined #openstack-ironic | 03:20 | |
| *** tteggel has joined #openstack-ironic | 03:20 | |
| *** anteaya has joined #openstack-ironic | 03:20 | |
| *** zul has joined #openstack-ironic | 03:20 | |
| *** mdorman has joined #openstack-ironic | 03:22 | |
| *** ramineni has quit IRC | 03:23 | |
| *** ramineni has joined #openstack-ironic | 03:23 | |
| *** mdorman has quit IRC | 03:24 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Ironic nova driver to cache ironic client calls https://review.openstack.org/102695 | 03:56 |
| *** bmahalakshmi has joined #openstack-ironic | 03:59 | |
| *** nosnos has joined #openstack-ironic | 04:01 | |
| *** sabah has joined #openstack-ironic | 04:17 | |
| *** subah has joined #openstack-ironic | 04:18 | |
| *** sabah has quit IRC | 04:21 | |
| *** subah has quit IRC | 04:25 | |
| *** k4n0 has joined #openstack-ironic | 04:27 | |
| *** Nisha has joined #openstack-ironic | 04:44 | |
| *** vinbs has quit IRC | 04:55 | |
| *** vinbs has joined #openstack-ironic | 04:55 | |
| *** eghobo has quit IRC | 05:02 | |
| *** bvivek has joined #openstack-ironic | 05:06 | |
| *** eghobo has joined #openstack-ironic | 05:12 | |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Ironic nova driver to cache ironic client calls https://review.openstack.org/102695 | 05:12 |
| *** radsy has quit IRC | 05:28 | |
| *** shausy has joined #openstack-ironic | 05:30 | |
| *** Haomeng has quit IRC | 05:35 | |
| *** eghobo has quit IRC | 05:35 | |
| *** eghobo has joined #openstack-ironic | 05:35 | |
| *** lazy_prince is now known as killer_prince | 05:38 | |
| *** killer_prince is now known as lazy_prince | 05:39 | |
| *** rameshg87_afk is now known as rameshg87 | 05:42 | |
| *** eguz has joined #openstack-ironic | 05:43 | |
| *** eghobo has quit IRC | 05:45 | |
| openstackgerrit | Zhongyue Luo proposed a change to openstack/ironic: Moves node resource tests to a separate testcase https://review.openstack.org/106933 | 05:45 |
| *** harlowja is now known as harlowja_away | 05:49 | |
| *** Haomeng has joined #openstack-ironic | 05:55 | |
| *** chuckC has quit IRC | 05:59 | |
| *** jcoufal has joined #openstack-ironic | 06:03 | |
| *** eguz has quit IRC | 06:05 | |
| *** eghobo has joined #openstack-ironic | 06:05 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/106948 | 06:10 |
| *** pradipta has joined #openstack-ironic | 06:20 | |
| *** lazy_prince has quit IRC | 06:31 | |
| *** nosnos has quit IRC | 06:32 | |
| *** bmahalakshmi has quit IRC | 06:32 | |
| *** bmahalakshmi has joined #openstack-ironic | 06:38 | |
| *** eguz has joined #openstack-ironic | 06:40 | |
| *** eghobo has quit IRC | 06:44 | |
| *** geekyogi has joined #openstack-ironic | 06:48 | |
| *** geekyogi has quit IRC | 06:48 | |
| *** geekyogi has joined #openstack-ironic | 06:49 | |
| *** eguz has quit IRC | 06:51 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 07:04 | |
| *** killer_prince has joined #openstack-ironic | 07:05 | |
| *** killer_prince is now known as lazy_prince | 07:05 | |
| *** rameshg87 has quit IRC | 07:12 | |
| *** rameshg87 has joined #openstack-ironic | 07:13 | |
| *** viktors has joined #openstack-ironic | 07:22 | |
| *** jistr has joined #openstack-ironic | 07:30 | |
| *** foexle has joined #openstack-ironic | 07:30 | |
| *** pradipta is now known as pradipta_away | 07:30 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: oslo.i18n migration https://review.openstack.org/105132 | 07:32 |
| *** bvivek has quit IRC | 07:47 | |
| openstackgerrit | Zhongyue Luo proposed a change to openstack/ironic: Moves node resource tests to a separate testcase https://review.openstack.org/106933 | 07:47 |
| *** derekh_ has joined #openstack-ironic | 08:01 | |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: oslo.i18n migration https://review.openstack.org/105132 | 08:13 |
| *** ndipanov_gone is now known as ndipanov | 08:18 | |
| *** athomas has joined #openstack-ironic | 08:19 | |
| *** romcheg1 has joined #openstack-ironic | 08:20 | |
| *** bvivek has joined #openstack-ironic | 08:21 | |
| *** lazy_prince has quit IRC | 08:21 | |
| dtantsur|afk | Late morning, Ironic | 08:24 |
| *** dtantsur|afk is now known as dtantsur | 08:24 | |
| *** lucasagomes has joined #openstack-ironic | 08:29 | |
| *** killer_prince has joined #openstack-ironic | 08:29 | |
| *** killer_prince is now known as lazy_prince | 08:30 | |
| yuriyz | morning Ironic dtantsur | 08:31 |
| dtantsur | yuriyz, morning :) | 08:31 |
| *** geekyogi has quit IRC | 08:32 | |
| dtantsur | Folks, do you think we require a spec for this https://review.openstack.org/#/c/106905 and https://review.openstack.org/#/c/106906/ ? | 08:32 |
| dtantsur | lifeless, ^^^ | 08:32 |
| lifeless | dtantsur: if you want one, but really? | 08:33 |
| lifeless | dtantsur: the initial numbers were AFAICT completely arbitrary, so are these, but they are based on using the code and finding it lacking | 08:34 |
| dtantsur | lifeless, I just don't know, whether these values are ok for everyone (initial are really arbitrary - I'm the author :) | 08:34 |
| lifeless | dtantsur: they probably aren't, which is why they are configurable, but do you know anyone that can buy a hard disk with < 20G of space nowadays? :) | 08:35 |
| dtantsur | heh | 08:35 |
| mrda | Night Ironic! | 08:35 |
| dtantsur | night, mrda! | 08:35 |
| *** mrda is now known as mrda-away | 08:35 | |
| lifeless | dtantsur: so, if we want more ops input, we should probably ask on the operators list about folks feelings | 08:35 |
| lifeless | dtantsur: however AFAICT there are three production deploys of Ironic, - Helion (in various places), TripleO-test-cluster, and OnMetal which doesn't use the cache at all. | 08:36 |
| dtantsur | lifeless, well, 20G and 1 week make sense to me | 08:37 |
| lifeless | dtantsur: so I'm not at all sure we'll get any useful responses, and I can speak for both Helion and TripleO to say that 20G is (probably) big enough - I'd be happy to make it bigger but worried about folk with smaller test VMs | 08:37 |
| dtantsur | lucasagomes, yuriyz are you ok with it ^^^? I am mostly convinced | 08:37 |
| yuriyz | I think 8-10GB OK | 08:38 |
| dtantsur | hmmm... lifeless, it looks like we don't have an easy agreement... What about making a spec? | 08:39 |
| yuriyz | and agree with lifeless for 1 week ttl | 08:40 |
| *** igordcard has joined #openstack-ironic | 08:40 | |
| dtantsur | brb, let me know what you agree on | 08:40 |
| lifeless | yuriyz: | 08:42 |
| lifeless | root@undercloud-undercloud-my4eb7s2dhzt:~# ls /mnt/state/var/lib/ironic/master_images/ -lh | 08:42 |
| lifeless | total 5.8G | 08:42 |
| lifeless | -rw-r--r-- 1 ironic ironic 3.4G Jul 14 23:58 2000e9f4-7df8-420e-a295-157fcbae6875 | 08:42 |
| lifeless | -rw-r--r-- 1 ironic ironic 6.2G Jul 15 00:56 31274650-f121-4ba2-9aa3-88576ad7b690 | 08:42 |
| lifeless | yuriyz: now I look, I'm thinking 20G might be too small still | 08:42 |
| lifeless | yuriyz: 8-10GB is totally useless | 08:42 |
| yuriyz | hm I see that you are right | 08:43 |
| lifeless | see the 5.8G total vs image sizes | 08:43 |
| lifeless | sparse files | 08:43 |
| lifeless | but the cache sums the stats, so it gets apparent size | 08:44 |
| lifeless | like I alluded to in the patch, if I had time I'd make it possible to disable the ttl | 08:44 |
| lifeless | image ids are write-once there's no use case for expiring content on age | 08:45 |
| yuriyz | lifeless, agree we should use production-ready defaults | 08:47 |
| * lucasagomes reads | 08:48 | |
| lucasagomes | well... yeah we have this consensus of having production-ready defaults in openstack so 20GB doesn't seems exorbitant | 08:50 |
| lucasagomes | on the tests VMs we can document/change devstack to tune down those values | 08:51 |
| lifeless | lucasagomes: I'm actually thinking 40GB might be better, and tune down the values in devtests's seed. but yeah, 1G was too small :> | 08:55 |
| lucasagomes | lifeless, I don't think I have parameter to object it... You guys are running it in production and figuring out what would be a sane default for it, thanks | 08:59 |
| lucasagomes | and yeah, thinking now 1GB may be way too small, at the begging it sounded like reasonable starting point heh | 09:00 |
| *** romcheg1 has quit IRC | 09:05 | |
| *** pelix has joined #openstack-ironic | 09:11 | |
| *** martyntaylor has joined #openstack-ironic | 09:24 | |
| *** geekyogi has joined #openstack-ironic | 09:27 | |
| *** romcheg1 has joined #openstack-ironic | 09:30 | |
| *** igordcard has quit IRC | 09:32 | |
| *** Alexei_987 has joined #openstack-ironic | 09:39 | |
| *** Nisha has quit IRC | 09:43 | |
| *** bmahalakshmi has quit IRC | 09:55 | |
| *** geekyogi has quit IRC | 09:55 | |
| *** shausy has quit IRC | 09:55 | |
| *** k4n0 has quit IRC | 09:55 | |
| *** Shrews has quit IRC | 09:55 | |
| *** anteaya has quit IRC | 09:55 | |
| *** zul has quit IRC | 09:55 | |
| *** Haomeng has quit IRC | 09:55 | |
| *** lucasagomes has quit IRC | 09:56 | |
| *** openstackgerrit has quit IRC | 09:56 | |
| *** yuriyz has quit IRC | 09:56 | |
| *** kevinbenton has quit IRC | 09:56 | |
| *** gilliard has quit IRC | 09:56 | |
| *** coolsvap has quit IRC | 09:56 | |
| *** davidlenwell has quit IRC | 09:56 | |
| *** vinbs has quit IRC | 09:56 | |
| *** jrist has quit IRC | 09:56 | |
| *** krtaylor has quit IRC | 09:56 | |
| *** viktors has quit IRC | 09:56 | |
| *** mikal has quit IRC | 09:56 | |
| *** ekarlso has quit IRC | 09:56 | |
| *** pradipta_away has quit IRC | 09:56 | |
| *** mkerrin has quit IRC | 09:56 | |
| *** NobodyCam has quit IRC | 09:56 | |
| *** GheRivero has quit IRC | 09:56 | |
| *** zer0c00l has quit IRC | 09:56 | |
| *** boris-42 has quit IRC | 09:56 | |
| *** romcheg1 has quit IRC | 09:56 | |
| *** bvivek has quit IRC | 09:56 | |
| *** athomas has quit IRC | 09:56 | |
| *** Mikhail_D_ltp has quit IRC | 09:56 | |
| *** dtantsur has quit IRC | 09:56 | |
| *** romcheg has quit IRC | 09:56 | |
| *** hemna__ has quit IRC | 09:56 | |
| *** zigo has quit IRC | 09:56 | |
| *** adam_g has quit IRC | 09:56 | |
| *** aignatov has quit IRC | 09:56 | |
| *** Isotopp has quit IRC | 09:56 | |
| *** aweeks has quit IRC | 09:56 | |
| *** mgagne has quit IRC | 09:56 | |
| *** BadCub_ has quit IRC | 09:56 | |
| *** enikanorov_ has quit IRC | 09:56 | |
| *** lynxman has quit IRC | 09:56 | |
| *** agordeev has quit IRC | 09:56 | |
| *** LiveOne has quit IRC | 09:56 | |
| *** comstud has quit IRC | 09:56 | |
| *** Mikhail_D_wk1 has quit IRC | 09:56 | |
| *** stevebaker has quit IRC | 09:56 | |
| *** dguerri has quit IRC | 09:56 | |
| *** soren has quit IRC | 09:56 | |
| *** martyntaylor has quit IRC | 09:56 | |
| *** lazy_prince has quit IRC | 09:56 | |
| *** jistr has quit IRC | 09:56 | |
| *** jcoufal has quit IRC | 09:56 | |
| *** morgabra has quit IRC | 09:56 | |
| *** ifarkas has quit IRC | 09:56 | |
| *** tteggel has quit IRC | 09:56 | |
| *** keekz has quit IRC | 09:56 | |
| *** Ng has quit IRC | 09:56 | |
| *** pelix has quit IRC | 09:56 | |
| *** pleia2 has quit IRC | 09:56 | |
| *** mmitchell_ has quit IRC | 09:56 | |
| *** Alexei_987 has quit IRC | 09:56 | |
| *** foexle has quit IRC | 09:56 | |
| *** ramineni has quit IRC | 09:56 | |
| *** rwsu has quit IRC | 09:56 | |
| *** mrda-away has quit IRC | 09:56 | |
| *** rch has quit IRC | 09:56 | |
| *** antonym has quit IRC | 09:56 | |
| *** wendar has quit IRC | 09:56 | |
| *** jroll has quit IRC | 09:56 | |
| *** toabctl has quit IRC | 09:56 | |
| *** d0ugal has quit IRC | 09:56 | |
| *** dhellmann has quit IRC | 09:56 | |
| *** JoshNang has quit IRC | 09:56 | |
| *** derekh_ has quit IRC | 09:56 | |
| *** rameshg87 has quit IRC | 09:56 | |
| *** datajerk has quit IRC | 09:56 | |
| *** annegentle has quit IRC | 09:56 | |
| *** notq has quit IRC | 09:56 | |
| *** sbadia has quit IRC | 09:56 | |
| *** sirushti has quit IRC | 09:56 | |
| *** SpamapS has quit IRC | 09:56 | |
| *** steveh has quit IRC | 09:56 | |
| *** proffalken has quit IRC | 09:56 | |
| *** devananda has quit IRC | 09:56 | |
| *** kylestev has quit IRC | 09:56 | |
| *** russell_h has quit IRC | 09:56 | |
| *** matty_dubs|gone has quit IRC | 09:56 | |
| *** pquerna has quit IRC | 09:56 | |
| *** Madasi has quit IRC | 09:56 | |
| *** harlowja_away has quit IRC | 09:56 | |
| *** JoshNang has joined #openstack-ironic | 10:03 | |
| *** dhellmann has joined #openstack-ironic | 10:03 | |
| *** toabctl has joined #openstack-ironic | 10:03 | |
| *** jroll has joined #openstack-ironic | 10:03 | |
| *** wendar has joined #openstack-ironic | 10:03 | |
| *** antonym has joined #openstack-ironic | 10:03 | |
| *** d0ugal has joined #openstack-ironic | 10:03 | |
| *** rch has joined #openstack-ironic | 10:03 | |
| *** mrda-away has joined #openstack-ironic | 10:03 | |
| *** rwsu has joined #openstack-ironic | 10:03 | |
| *** foexle has joined #openstack-ironic | 10:03 | |
| *** mmitchell_ has joined #openstack-ironic | 10:03 | |
| *** pleia2 has joined #openstack-ironic | 10:03 | |
| *** pelix has joined #openstack-ironic | 10:03 | |
| *** athomas has joined #openstack-ironic | 10:03 | |
| *** ramineni has joined #openstack-ironic | 10:03 | |
| *** Alexei_987 has joined #openstack-ironic | 10:03 | |
| *** romcheg1 has joined #openstack-ironic | 10:03 | |
| *** geekyogi has joined #openstack-ironic | 10:03 | |
| *** lazy_prince has joined #openstack-ironic | 10:03 | |
| *** lucasagomes has joined #openstack-ironic | 10:03 | |
| *** bvivek has joined #openstack-ironic | 10:03 | |
| *** jistr has joined #openstack-ironic | 10:03 | |
| *** viktors has joined #openstack-ironic | 10:03 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 10:03 | |
| *** bmahalakshmi has joined #openstack-ironic | 10:03 | |
| *** pradipta_away has joined #openstack-ironic | 10:03 | |
| *** jcoufal has joined #openstack-ironic | 10:03 | |
| *** Haomeng has joined #openstack-ironic | 10:03 | |
| *** shausy has joined #openstack-ironic | 10:03 | |
| *** vinbs has joined #openstack-ironic | 10:03 | |
| *** k4n0 has joined #openstack-ironic | 10:03 | |
| *** zul has joined #openstack-ironic | 10:03 | |
| *** anteaya has joined #openstack-ironic | 10:03 | |
| *** tteggel has joined #openstack-ironic | 10:03 | |
| *** keekz has joined #openstack-ironic | 10:03 | |
| *** Ng has joined #openstack-ironic | 10:03 | |
| *** Mikhail_D_wk1 has joined #openstack-ironic | 10:03 | |
| *** stevebaker has joined #openstack-ironic | 10:03 | |
| *** dguerri has joined #openstack-ironic | 10:03 | |
| *** soren has joined #openstack-ironic | 10:03 | |
| *** adam_g has joined #openstack-ironic | 10:03 | |
| *** aignatov has joined #openstack-ironic | 10:03 | |
| *** Isotopp has joined #openstack-ironic | 10:03 | |
| *** mgagne has joined #openstack-ironic | 10:03 | |
| *** aweeks has joined #openstack-ironic | 10:03 | |
| *** BadCub_ has joined #openstack-ironic | 10:03 | |
| *** enikanorov_ has joined #openstack-ironic | 10:03 | |
| *** LiveOne has joined #openstack-ironic | 10:03 | |
| *** lynxman has joined #openstack-ironic | 10:03 | |
| *** agordeev has joined #openstack-ironic | 10:03 | |
| *** comstud has joined #openstack-ironic | 10:03 | |
| *** mkerrin has joined #openstack-ironic | 10:03 | |
| *** NobodyCam has joined #openstack-ironic | 10:03 | |
| *** GheRivero has joined #openstack-ironic | 10:03 | |
| *** zer0c00l has joined #openstack-ironic | 10:03 | |
| *** mikal has joined #openstack-ironic | 10:03 | |
| *** davidlenwell has joined #openstack-ironic | 10:03 | |
| *** coolsvap has joined #openstack-ironic | 10:03 | |
| *** zigo has joined #openstack-ironic | 10:03 | |
| *** gilliard has joined #openstack-ironic | 10:03 | |
| *** ekarlso has joined #openstack-ironic | 10:03 | |
| *** kevinbenton has joined #openstack-ironic | 10:03 | |
| *** hemna__ has joined #openstack-ironic | 10:03 | |
| *** boris-42 has joined #openstack-ironic | 10:03 | |
| *** ifarkas has joined #openstack-ironic | 10:03 | |
| *** romcheg has joined #openstack-ironic | 10:03 | |
| *** dtantsur has joined #openstack-ironic | 10:03 | |
| *** yuriyz has joined #openstack-ironic | 10:03 | |
| *** krtaylor has joined #openstack-ironic | 10:03 | |
| *** Shrews has joined #openstack-ironic | 10:03 | |
| *** morgabra has joined #openstack-ironic | 10:03 | |
| *** openstackgerrit has joined #openstack-ironic | 10:03 | |
| *** jrist has joined #openstack-ironic | 10:03 | |
| *** sbadia has joined #openstack-ironic | 10:03 | |
| *** sirushti has joined #openstack-ironic | 10:03 | |
| *** SpamapS has joined #openstack-ironic | 10:03 | |
| *** steveh has joined #openstack-ironic | 10:03 | |
| *** proffalken has joined #openstack-ironic | 10:03 | |
| *** devananda has joined #openstack-ironic | 10:03 | |
| *** kylestev has joined #openstack-ironic | 10:03 | |
| *** russell_h has joined #openstack-ironic | 10:03 | |
| *** matty_dubs|gone has joined #openstack-ironic | 10:03 | |
| *** pquerna has joined #openstack-ironic | 10:03 | |
| *** Madasi has joined #openstack-ironic | 10:03 | |
| *** harlowja_away has joined #openstack-ironic | 10:04 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/python-ironicclient: Add pagination support to {node, port, chassis}-list https://review.openstack.org/91585 | 10:04 |
| dtantsur | lifeless, so we agreed on 20G and 1 week, right? Please update your commit messages (or even their summaries) to mention the new value | 10:07 |
| *** derekh_ has joined #openstack-ironic | 10:08 | |
| *** rameshg87 has joined #openstack-ironic | 10:08 | |
| *** datajerk has joined #openstack-ironic | 10:08 | |
| *** annegentle has joined #openstack-ironic | 10:08 | |
| *** notq has joined #openstack-ironic | 10:08 | |
| lifeless | dtantsur: I'll see where its at tomorrow, need to sleep on it | 10:09 |
| dtantsur | lifeless, g'night! | 10:09 |
| *** lazy_prince has quit IRC | 10:20 | |
| *** lazy_prince has joined #openstack-ironic | 10:21 | |
| rameshg87 | dtantsur: request you to take a look at ilo power code which you had a look earlier: https://review.openstack.org/#/c/89500/ | 10:27 |
| dtantsur | rameshg87, what's the state of the spec? | 10:28 |
| rameshg87 | dtantsur: spec has been submitted https://review.openstack.org/#/c/97455/ | 10:29 |
| dtantsur | rameshg87, ok, I'll try to find some time this evening (sorry, too much things to do) | 10:29 |
| rameshg87 | dtantsur: sure, thanks :-) | 10:30 |
| *** romcheg1 has quit IRC | 10:35 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Raise appropriate errors on duplicate Node, Port and Chassis creation https://review.openstack.org/102506 | 10:36 |
| *** lazy_prince has quit IRC | 10:36 | |
| *** Haomeng has quit IRC | 10:38 | |
| *** killer_prince has joined #openstack-ironic | 10:41 | |
| *** killer_prince is now known as lazy_prince | 10:41 | |
| *** Haomeng has joined #openstack-ironic | 10:46 | |
| openstackgerrit | A change was merged to openstack/ironic: Prevent updating UUID of Node, Port and Chassis on DB API level https://review.openstack.org/102247 | 10:56 |
| *** chuckC has joined #openstack-ironic | 10:56 | |
| *** zul has quit IRC | 11:00 | |
| *** martyntaylor has joined #openstack-ironic | 11:02 | |
| *** zul has joined #openstack-ironic | 11:05 | |
| *** ramineni has quit IRC | 11:05 | |
| *** bvivek has quit IRC | 11:11 | |
| *** bmahalakshmi has quit IRC | 11:11 | |
| *** k4n0 has quit IRC | 11:15 | |
| *** openstackgerrit has quit IRC | 11:21 | |
| *** martyntaylor has left #openstack-ironic | 11:28 | |
| *** lucasagomes is now known as lucas-hungry | 11:31 | |
| *** bvivek has joined #openstack-ironic | 11:32 | |
| *** Haomeng has quit IRC | 11:44 | |
| *** Haomeng has joined #openstack-ironic | 11:49 | |
| *** martyntaylor1 has joined #openstack-ironic | 11:53 | |
| *** jdob has joined #openstack-ironic | 12:02 | |
| *** romcheg1 has joined #openstack-ironic | 12:03 | |
| *** openstackgerrit has joined #openstack-ironic | 12:17 | |
| *** vinbs_ has joined #openstack-ironic | 12:17 | |
| *** vinbs__ has joined #openstack-ironic | 12:20 | |
| *** vinbs has quit IRC | 12:21 | |
| *** vinbs__ is now known as vinbs | 12:21 | |
| dtantsur | someone remembers, in Node.properties is `storage` in GiB or MiB? | 12:22 |
| romcheg1 | Morning Ironic! | 12:22 |
| dtantsur | romcheg1, morning! | 12:23 |
| *** romcheg1 is now known as romcheg_ltp | 12:23 | |
| romcheg_ltp | Morning dtantsur! | 12:23 |
| *** vinbs_ has quit IRC | 12:23 | |
| *** vinbs has quit IRC | 12:26 | |
| rameshg87 | dtantsur: did you mean local_gb in properties of node ? | 12:35 |
| dtantsur | rameshg87, I was confused by wrong example in db.utils. disregard my question, please :) | 12:36 |
| rameshg87 | dtantsur: okay :-) | 12:36 |
| *** ryanpetrello has quit IRC | 12:36 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Add shared functions for hardware discovery https://review.openstack.org/107010 | 12:36 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Fix wrong test fixture for Node.properties https://review.openstack.org/107031 | 12:36 |
| dtantsur | rameshg87, that's what I mean: https://review.openstack.org/#/c/107031/ | 12:37 |
| rameshg87 | dtantsur: okay, got it :-) | 12:37 |
| *** rloo has joined #openstack-ironic | 12:38 | |
| openstackgerrit | Imre Farkas proposed a change to openstack/ironic-specs: Enable PXE deploy driver for DRAC https://review.openstack.org/107033 | 12:39 |
| *** ryanpetrello has joined #openstack-ironic | 12:41 | |
| *** martyntaylor1 has left #openstack-ironic | 12:42 | |
| *** lucas-hungry is now known as lucasagomes | 12:44 | |
| *** matty_dubs|gone is now known as matty_dubs | 12:52 | |
| rloo | hello ironickers ;) | 12:53 |
| rloo | Hi ifarkas. | 12:53 |
| rloo | ifarkas -- wrt 107033. that's a new spec? | 12:53 |
| rloo | ifarkas -- fyi, in yesterday's meeting there was a discussion about not accepting new specs for juno anymore. | 12:54 |
| *** rakesh_hs has joined #openstack-ironic | 12:56 | |
| lucasagomes | rloo, morning | 12:57 |
| ifarkas | rloo, hi, yep, that's a new spec | 12:57 |
| rloo | hi lucasagomes | 12:57 |
| romcheg_ltp | Morning romcheg | 12:57 |
| romcheg_ltp | hmm | 12:57 |
| rloo | romcheg: did you just say morning to yourself? | 12:58 |
| romcheg_ltp | Tab-completion sucked :) | 12:58 |
| rloo | ha ha | 12:58 |
| dtantsur | morning, rloo! | 12:58 |
| ifarkas | rloo, I thought the new spec proposal freeze is Jul 24 as devananda wrote in the email with subject "Juno priorities and spec review timeline" | 12:58 |
| romcheg_ltp | Morning rloo! | 12:58 |
| *** jcoufal has quit IRC | 12:58 | |
| rloo | ifarkas: will bring up with devananda then, if he put it in an email. | 12:59 |
| *** jcoufal has joined #openstack-ironic | 12:59 | |
| ifarkas | rloo, ok. fyi: http://lists.openstack.org/pipermail/openstack-dev/2014-June/039031.html | 13:00 |
| rloo | ifarkas: thx! devananda ^^^ your email says new spec proposal freeze Jul 24 | 13:01 |
| *** lynxman has quit IRC | 13:02 | |
| *** jistr has quit IRC | 13:03 | |
| *** jistr has joined #openstack-ironic | 13:05 | |
| *** jbjohnso has joined #openstack-ironic | 13:08 | |
| *** ramineni has joined #openstack-ironic | 13:11 | |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Add methods to ipmitool driver https://review.openstack.org/100364 | 13:14 |
| *** stendulker has joined #openstack-ironic | 13:15 | |
| * dtantsur is going to steal some code from jroll's agent patch :) | 13:16 | |
| stendulker | dtantsur : Hi | 13:17 |
| dtantsur | stendulker, hi! | 13:17 |
| stendulker | dtantsur: I have addressed your comments related to firmware settings spec https://review.openstack.org/#/c/101122/ Can you please validate the same | 13:18 |
| jroll | morning y'all | 13:18 |
| jroll | dtantsur: what code? :) | 13:18 |
| jroll | and which patch? :P | 13:19 |
| dtantsur | jroll, morning! finding node by list of interfaces | 13:19 |
| jroll | lucasagomes: around? | 13:19 |
| lucasagomes | jroll, yo yes | 13:19 |
| jroll | dtantsur: feel free to factor that out :P | 13:19 |
| jroll | lucasagomes: hey, so I moved the config option back... and looking at this: https://review.openstack.org/#/c/100734/16/ironic/drivers/modules/pxe.py | 13:19 |
| stendulker | lucasgomes: Hi | 13:20 |
| dtantsur | jroll, thanks! while spec is not approved yet, I started on separating required common functions: https://review.openstack.org/#/c/107010/ | 13:20 |
| jroll | lucasagomes: if we decide to move that back to a class, then that patch just basically does "mv ironic/drivers/modules/image_cache.py ironic/common/image_cache.py" | 13:20 |
| jroll | lucasagomes: which I don't have a problem with... just thinking of abandoning at this point though :) | 13:20 |
| stendulker | lucasgomes: Can you please review the firmware settings spec https://review.openstack.org/#/c/101122/ | 13:20 |
| lucasagomes | jroll, yeah that's a true point... I thought about the class because it looks more consistent with what we already have | 13:21 |
| jroll | dtantsur: cool | 13:21 |
| lucasagomes | jroll, if we have a function that only creates and return an instance of a class, it looks better to have the class directly | 13:21 |
| lucasagomes | IMHO | 13:22 |
| jroll | lucasagomes: right... I agree | 13:22 |
| jroll | lucasagomes: I think I was refactoring a bit too hard here :P | 13:22 |
| lucasagomes | heh yeah, but that file def needs some refactors | 13:22 |
| lucasagomes | the names are super confusing | 13:23 |
| jroll | yeah :/ | 13:23 |
| jroll | I think you're more confused about them because you're in ipxe world | 13:23 |
| jroll | everybody else thinks pxe == tftp | 13:23 |
| lucasagomes | yeah exactly | 13:24 |
| jroll | and doesn't realize the transport layer is different than the boot mechanism :) | 13:24 |
| lucasagomes | +1 | 13:24 |
| jroll | ok, so, I think I'm going to just ditch this patch for now | 13:24 |
| lucasagomes | ack, I will rebase the iPXE work on top of ur other patch | 13:25 |
| jroll | ok, sorry about that :/ | 13:25 |
| jroll | should be pretty close at this point though | 13:25 |
| jroll | I'm still going to keep 100735 | 13:25 |
| lucasagomes | jroll, nah nothing to be sorry about it's all good :D | 13:26 |
| jroll | yeah, just I know rebasing can really suck :P | 13:26 |
| lucasagomes | hah yeah | 13:26 |
| *** jgrimm has joined #openstack-ironic | 13:27 | |
| lucasagomes | jroll, btw, you don't want to be core of the ironic-specs? | 13:28 |
| jroll | lol | 13:29 |
| lucasagomes | :P | 13:29 |
| lucasagomes | say yes say yes | 13:29 |
| lucasagomes | hah | 13:29 |
| jroll | I'll just wait for russell_h to get back and proxy my +2's through him :P | 13:29 |
| jroll | I'm not opposed to being core there, but I prefer code over specs | 13:30 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Factor out deploy info from PXE driver https://review.openstack.org/100735 | 13:31 |
| *** openstackgerrit has quit IRC | 13:31 | |
| lucasagomes | jroll, heh, fair :) | 13:33 |
| *** 17SAAF480 has joined #openstack-ironic | 13:33 | |
| lucasagomes | maybe Shrews will want :P | 13:33 |
| jroll | heh | 13:33 |
| jroll | I'm happy to +2 some specs if someone gives me that power | 13:33 |
| jroll | I just don't want to read them | 13:34 |
| jroll | not sure how desirable that is | 13:34 |
| lucasagomes | lol not very | 13:34 |
| jroll | :P | 13:34 |
| jroll | so... agent driver is going to be blocked on this: https://review.openstack.org/#/c/102632/ | 13:36 |
| jroll | fun. | 13:36 |
| 17SAAF480 | Victor Sergeyev proposed a change to openstack/ironic: Use oslo.db library https://review.openstack.org/42159 | 13:36 |
| 17SAAF480 | Victor Sergeyev proposed a change to openstack/ironic: Use opportunistic approach from migration testing https://review.openstack.org/107053 | 13:36 |
| *** ramineni1 has joined #openstack-ironic | 13:37 | |
| lucasagomes | jroll, will review it in a bit | 13:38 |
| 17SAAF480 | Jim Rollenhagen proposed a change to openstack/ironic: Add missing docstrings https://review.openstack.org/106202 | 13:38 |
| stendulker | dtantsur: Hi | 13:38 |
| *** 17SAAF480 has quit IRC | 13:38 | |
| jroll | lucasagomes: no worries... going to wait on swift people anyway | 13:38 |
| viktors | devananda: hi! Could you please, when you'll have a time, take a look at patch https://review.openstack.org/107053 (Use opportunistic approach from migration testing) - this is attempt to fix bugs 1327397 and 1328997. Thanks! | 13:39 |
| *** openstackgerrit has joined #openstack-ironic | 13:40 | |
| *** rameshg87 has left #openstack-ironic | 13:40 | |
| *** ramineni has quit IRC | 13:41 | |
| *** pcrews has joined #openstack-ironic | 13:42 | |
| lucasagomes | If ur willing to review something, here's something we need and is blocking us from releasing a new version of the python-ironicclient: https://review.openstack.org/#/c/91585/ | 13:42 |
| openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: oslo.i18n migration https://review.openstack.org/105132 | 13:42 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Add missing docstrings https://review.openstack.org/106202 | 13:43 |
| jroll | lucasagomes: sure... let me finish up iterating on some patches | 13:44 |
| lucasagomes | np | 13:44 |
| dtantsur | stendulker, hi, I saw your request, but we have already a huge prioritized queue of reviews, and I don't know when I'll have time to get back to it, sorry. | 13:45 |
| *** ramineni1 has quit IRC | 13:46 | |
| stendulker | dtantsur: I understand. Just wanted to inform you status from my side. Do review it, when you get time. | 13:46 |
| stendulker | dtantsur: Thank you. | 13:47 |
| dtantsur | np | 13:47 |
| dtantsur | Folks, mind reviewing 3 lines patch for tests? https://review.openstack.org/#/c/107031/ | 13:48 |
| *** kevinbenton has quit IRC | 13:52 | |
| *** kevinbenton has joined #openstack-ironic | 13:53 | |
| *** ramineni has joined #openstack-ironic | 13:53 | |
| jroll | dtantsur: +2'd... curious why that doesn't break tests. :) | 13:55 |
| *** ramineni1 has joined #openstack-ironic | 13:55 | |
| jroll | dtantsur: as in, do we even need a properties dict in that fixture? | 13:55 |
| NobodyCam | good morning Ironic | 13:56 |
| *** lynxman has joined #openstack-ironic | 13:56 | |
| rloo | dtantsur: approved. | 13:56 |
| NobodyCam | I will be starting a little late this mornig | 13:56 |
| *** ramineni has quit IRC | 13:58 | |
| dtantsur | jroll, I need it, that's why I encountered this problem :) | 13:59 |
| dtantsur | jroll, rloo, thanks! | 14:00 |
| *** romcheg_ltp has left #openstack-ironic | 14:00 | |
| jroll | NobodyCam: not allowed. | 14:00 |
| rloo | dtantsur: yw. (although I always feel a bit bad fast-tracking someone's patch over others.) | 14:01 |
| dtantsur | oh, I understand :) | 14:03 |
| *** bvivek has quit IRC | 14:06 | |
| jroll | lucasagomes: hey, about https://review.openstack.org/#/c/91585/7/ironicclient/v1/chassis.py | 14:07 |
| NobodyCam | jroll: :) have to take bubbie to the vets, this morning :( | 14:07 |
| jroll | lucasagomes: I don't see what you're doing with filters etc there... because _path(path) just should return '/v1/chassis'.... oh I see | 14:08 |
| jroll | lucasagomes: I don't like it :| | 14:08 |
| jroll | NobodyCam: :( | 14:08 |
| * lucasagomes is in a call 1 min will answer it | 14:09 | |
| jroll | lucasagomes: no worries, I see what it's doing, I just don't like it | 14:09 |
| lucasagomes | lol ok | 14:10 |
| jroll | why can't we just use requests? :((( | 14:13 |
| *** Mikhail_D_ltp has quit IRC | 14:16 | |
| lucasagomes | :( | 14:17 |
| lucasagomes | jroll, u mean the mechanism to actually build the URL? | 14:17 |
| jroll | lucasagomes: yeah | 14:18 |
| lucasagomes | yeah that's very simplistic right now | 14:18 |
| jroll | like, in a sane http client, you pass params as a dict | 14:18 |
| jroll | but our client doesn't have that | 14:18 |
| lucasagomes | yeah that's true | 14:18 |
| jroll | /shrug | 14:19 |
| lucasagomes | :( | 14:19 |
| * jroll rewrites everything | 14:19 | |
| lucasagomes | lol | 14:21 |
| lucasagomes | many parts of our driver, this base class for e.g | 14:21 |
| lucasagomes | is a copy&paste from other openstack clients | 14:21 |
| lucasagomes | :( | 14:21 |
| jroll | yeah... | 14:22 |
| *** lynxman has quit IRC | 14:23 | |
| Shrews | NobodyCam: any reason why you didn't approve 101765? | 14:24 |
| *** lynxman has joined #openstack-ironic | 14:25 | |
| *** annegentle is now known as anneonthebus | 14:26 | |
| *** geekyogi has quit IRC | 14:31 | |
| GheRivero | lucasagomes: you have time to talk about https://review.openstack.org/#/c/103685 | 14:34 |
| lucasagomes | GheRivero, hey yea | 14:35 |
| GheRivero | do you have any other concerns besides revising all the InvalidParameterValue (and confirm they are indeed MissingParameterValue) | 14:36 |
| jroll | I have a different approach here... what if validate() also accepted the method name ('deploy', 'tear_down', etc), and could change its validation based on that? | 14:40 |
| * lucasagomes looks at the patch again | 14:40 | |
| jroll | no matter the exception raised, I don't think an error should pop up on tear_down just because some metadata is missing (that isn't used for tear_down) | 14:41 |
| lucasagomes | GheRivero, but afair that's my only concern there | 14:41 |
| lucasagomes | and also making the validate_driver_interfaces() handle that new expcetion | 14:41 |
| GheRivero | ok. Will address that today. | 14:43 |
| jroll | lucasagomes: approved https://review.openstack.org/#/c/91585/ :) | 14:43 |
| jroll | GheRivero, lucasagomes, wdyt about validate() taking the method name | 14:43 |
| jroll | I think it's silly to do the same validation for deploy and tear_down | 14:43 |
| jroll | in our case, it's turned into moving validation into those functions | 14:44 |
| GheRivero | jroll: I think it should be the parent method the one dealing with the error, but it just and opinion | 14:44 |
| lucasagomes | \o/ | 14:45 |
| jroll | GheRivero: I guess I disagree that missing e.g. image_source is an error at all, for tear_down | 14:45 |
| lucasagomes | jroll, yeah, it sounded like a good idea to have a validate method for the whole interface | 14:45 |
| GheRivero | validation should do just what it said. It shouldn't care what is the purpose os the validation. Today is easy just with a couple of methods... but if it grows, checking every option inside validate could be insane | 14:45 |
| lucasagomes | but I agree that we may need something more specific for each method instead | 14:46 |
| lucasagomes | but again, that validate is for the whole interface | 14:46 |
| jroll | GheRivero: also, imagine validate() adds some check for a parameter that tear_down needs. raises MissingParameterError if not there. now your tear_down method doesn't catch it. | 14:46 |
| NobodyCam | Shrews: Ahh Just wanted lucasagomes to have a quick look see | 14:46 |
| jroll | s/doesn't catch it/keeps going/ | 14:47 |
| lucasagomes | NobodyCam, morning | 14:47 |
| NobodyCam | morning lucasagomes | 14:47 |
| jroll | GheRivero: my bigger question is, how do you validate something, if you don't know what that something is? | 14:47 |
| Shrews | interesting that the docstring for validate() in base.py says it's for validating *deploy*. nothing about tear_down. | 14:50 |
| GheRivero | i got it now. | 14:50 |
| jroll | Shrews: dat's what I'm sayin' | 14:50 |
| Shrews | jroll: yeah. the interface predates me, but i'm just wondering if it morphed over time | 14:51 |
| GheRivero | different methods need to validate different params :/ | 14:51 |
| jroll | "oh shit... ironic needs to tear down nodes, too, huh?" -- deva, 18 months ago | 14:51 |
| jroll | :P | 14:51 |
| Shrews | lol | 14:52 |
| GheRivero | lol | 14:52 |
| lucasagomes | heh | 14:52 |
| lucasagomes | yeah i think that docstring is wrong | 14:52 |
| lucasagomes | lol | 14:52 |
| jroll | the docstring is totally right... none of that validation is needed for tear_down | 14:52 |
| jroll | bbiab | 14:53 |
| lucasagomes | right, so on the do_node_tear_down method from the manager we may not want to run validate() on the deploy interface | 14:55 |
| lucasagomes | well urgh... | 14:56 |
| dtantsur | NobodyCam, morning! | 14:59 |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Set a more generous default image cache size https://review.openstack.org/106905 | 15:00 |
| NobodyCam | morning dtantsur | 15:00 |
| *** romcheg1 has joined #openstack-ironic | 15:00 | |
| NobodyCam | dtantsur: hahaha I just fixed the pep8 issue on that.. :-p | 15:03 |
| dtantsur | :) | 15:03 |
| lucasagomes | dtantsur, what we are doing with the inventory_timestamp? | 15:03 |
| lucasagomes | brb quickly | 15:04 |
| dtantsur | lucasagomes, NobodyCam suggested it will be helpful for UI to actually know, when this HW information was last updated and even search/sort by it (e.g. to quickly find newly added nodes) | 15:04 |
| *** mdorman has joined #openstack-ironic | 15:05 | |
| *** ramineni has joined #openstack-ironic | 15:06 | |
| *** cdent has joined #openstack-ironic | 15:07 | |
| cdent | lucasagomes: ping? | 15:08 |
| *** ramineni1 has quit IRC | 15:09 | |
| lucasagomes | dtantsur, we have the updated_at field on nodes as well | 15:10 |
| lucasagomes | cdent, pong | 15:10 |
| cdent | howdy, I just have a quick question about that stub ipmitool host you told me about | 15:10 |
| dtantsur | lucasagomes, yeah, but it can be changed for a lot of reasons. Anyway, this field is more by request of reviewers, than invented by me :) | 15:10 |
| cdent | it's pretty slow, is that expected? | 15:10 |
| cdent | single digit minutes to gather sensor data | 15:11 |
| lucasagomes | cdent, ouch... hmm it shouldn't be that slow | 15:11 |
| lucasagomes | lemme check it | 15:11 |
| lucasagomes | actually matty_dubs may now better | 15:11 |
| lucasagomes | know* | 15:12 |
| lucasagomes | dtantsur, yeah, so uff... it looks that it complicates more :/ it sounds redundant as well | 15:13 |
| jroll | perhaps have the timestamp in the dict that's stored | 15:13 |
| jroll | next to the hardware | 15:14 |
| jroll | I don't see any reason to sort by it... if you want to find newly added nodes, use the flag that's been added | 15:14 |
| dtantsur | lucasagomes, jroll: here it was introduced by request of jdob: https://review.openstack.org/#/c/102565/9/specs/juno/generic-hardware-discovery.rst | 15:15 |
| dtantsur | (link leads to his comment) | 15:15 |
| dtantsur | jroll, no dict is stored :) | 15:15 |
| dtantsur | only Node.properties are updated | 15:15 |
| jroll | dtantsur: ohhh, right | 15:16 |
| jroll | blah | 15:16 |
| jdob | dtantsur: which comment, the one about the explicit parameter for newly_discovered? | 15:17 |
| lucasagomes | cdent, I'm trying another node here | 15:17 |
| openstackgerrit | Chris Krelle proposed a change to openstack/ironic: Set a more generous default image cache size https://review.openstack.org/106905 | 15:17 |
| dtantsur | jdob, no, the 4th, about timestamp | 15:17 |
| jdob | ah, gotcha | 15:17 |
| lucasagomes | cdent, it seems very slow as well :( | 15:17 |
| cdent | foo | 15:18 |
| cdent | It's not the end of the world, it's just making my manual test cycle rather painful :) | 15:18 |
| lucasagomes | yeah | 15:18 |
| lucasagomes | cdent, real1m58.163s | 15:19 |
| lucasagomes | :/ | 15:19 |
| NobodyCam | dtantsur: added to the commit message for that patch ^^ | 15:20 |
| *** jistr has quit IRC | 15:20 | |
| dtantsur | NobodyCam, thanks! | 15:20 |
| *** jistr has joined #openstack-ironic | 15:21 | |
| *** openstackgerrit has quit IRC | 15:21 | |
| NobodyCam | brb | 15:22 |
| dtantsur | jdob, do you really need this timestamp for UI? | 15:23 |
| *** ramineni has quit IRC | 15:23 | |
| * cdent is still waiting from time results | 15:24 | |
| cdent | s/from/for | 15:24 |
| lucasagomes | cdent, seems it's the location that is affecting us... matty_dubs just tried it from wexford and it took 16 seconds to run | 15:24 |
| jdob | dtantsur: that's a better question for jcoufal and tzumainn; from what I remember, that suggestion was largely based on what I saw elsewhere in the spec. I think they do want to show the hardware config has changed, but the specifics on how to do that are open to debate | 15:24 |
| lucasagomes | westford* | 15:24 |
| cdent | feh | 15:24 |
| cdent | real 5m6.172s | 15:25 |
| cdent | (that's for sdr -v, which is what the code is calling) | 15:25 |
| lucasagomes | holy cow! | 15:25 |
| lucasagomes | yeah | 15:25 |
| lucasagomes | here it took 2 min | 15:25 |
| jdob | my comment was more about the internal consistency of the spec. the problem description says it can detect changes to the hardware, but I didn't see how it would actually track that | 15:25 |
| lucasagomes | cdent, maybe if we get a machine in Brno it will be quicker | 15:25 |
| * cdent nods | 15:25 | |
| cdent | I thought perhaps I could fake something with some hardware I have around the house, but nothing has the right knobs | 15:26 |
| *** rameshg87 has joined #openstack-ironic | 15:28 | |
| jroll | jdob: I think at that point we're going back into CMDB territory again :/ | 15:28 |
| dtantsur | jroll, got two shared functions for discovery, you may be interested: https://review.openstack.org/#/c/107010/ | 15:29 |
| matty_dubs | So I was thinking about the CMDB stuff last night. I wonder if it makes sense to start to envision, even if only theoretical, a separate CMDB project and see if that solves needs | 15:29 |
| NobodyCam | :( its such a fine line between what we need and becomming a cmdb | 15:30 |
| jdob | to be fair, i'm not overly familiar at all with ironic, so if things I'm suggesting aren't in line with its vision, I'm not going to argue strongly for them | 15:30 |
| matty_dubs | I know with the NetApp stuff, I kept coming back to "Well, Ironic is clearly the best fit of the existing projects" | 15:30 |
| dtantsur | lucasagomes, NobodyCam, jroll: finally, should I delete this timestamp field? | 15:30 |
| jroll | matty_dubs: this has some cmdb related things https://wiki.openstack.org/wiki/Fuel | 15:31 |
| * lucasagomes thinking... | 15:31 | |
| jroll | dtantsur: probably | 15:31 |
| jroll | dtantsur: I don't see a valid use case, others may | 15:31 |
| lucasagomes | dtantsur, if you do delete it... why we have a new method for updating the inventory? | 15:31 |
| jdob | my only request then is to also delete the line from the spec, or to clarify how "Responding to basic hardware changes, e.g. new hard drive, added RAM etc" is met | 15:31 |
| lucasagomes | why not just PATCH [{'path': '/properties'...}, {}] /nodes/<uuid> | 15:32 |
| jdob | either how it's met, or what it entails, because I read that as saying ironic will report or otherwise make available knowledge that changes have been made | 15:32 |
| dtantsur | lucasagomes, to share the code (see https://review.openstack.org/#/c/107010/ ). Maybe it was a bot overkill to put it to the spec, as it's more of implementation detail, but I wanted to show it to some people doing the same | 15:32 |
| dtantsur | lucasagomes, ehhhhhr... provided you know UUID for just-booted node - yes | 15:33 |
| jdob | the timestamp was just a suggestion as a (presumably, which doesn't appear to be the case) simple way of saying "Something last changed on X" | 15:33 |
| lucasagomes | dtantsur, you're going to find the UUID by looking the NIC addr no? | 15:33 |
| * lucasagomes still reading | 15:33 | |
| * gilliard out - see you in the morning | 15:34 | |
| jroll | lucasagomes: the discovery ramdisk would need to know it to send a patch request | 15:34 |
| jroll | jdob: it "responds" to hardware changes by storing that data :P | 15:34 |
| dtantsur | lucasagomes, yeah, but it will be inside conductor, too late to do PATCH :) | 15:34 |
| NobodyCam | night gilliard | 15:34 |
| dtantsur | gilliard, g'night | 15:34 |
| matty_dubs | cdent: I could also just set you up with shell access on a box in Westford, local to that IPMI instance | 15:34 |
| jdob | jroll: fair enough :) | 15:35 |
| matty_dubs | Since transatlantic IPMI is apparently unbearable ;) | 15:35 |
| jdob | maybe i was just reading into it | 15:35 |
| jroll | you guys aren't exposing IPMI to the internet... are you? | 15:35 |
| * jroll hacks into red hat | 15:36 | |
| matty_dubs | jroll: Haha, no, over corporate WAN | 15:36 |
| dtantsur | ok, I'm deleting the timestamp. I suggest reintroducing it later on demand. | 15:36 |
| jroll | good, only RH employees can pwn it :) | 15:36 |
| jdob | so dtantsur, check with jcoufal to see what he was thinking in terms of being able to report changes to the user, otherwise I don't have a strong argument for any sort of reporting | 15:36 |
| jdob | and am cool with jroll's approach of interpretting that as that ironic will keep its inventory updated | 15:36 |
| matty_dubs | jroll: Heh, I have (at home, not the office) some Dell boxes with a vulnerable BMC, where you can get remote console without having to log in | 15:36 |
| matty_dubs | And Dell has never patched it, since it was an old/oddball setup | 15:37 |
| jroll | ha | 15:37 |
| cdent | matty_dubs: Thanks, but I think I'll cope with what I've got: environment tweaked and all that to get everything talking to one another in a devstack | 15:37 |
| jcoufal | dtantsur: can we catch up tomorrow? | 15:37 |
| jroll | I just assume all BMCs are vulnerable :) | 15:37 |
| NobodyCam | jroll: they are :) | 15:37 |
| dtantsur | jcoufal, whenever. I suggest making it as an addition to the spec, because we can't discuss the details forever :) | 15:37 |
| lucasagomes | jroll, dtantsur... idk I may be thinking differently... how a machine already registered in Ironic will endup booting the discovery ramdisk? | 15:37 |
| lucasagomes | the idea is not a pxe fallback config file that machines _not registered_ will fallback to and boot the deploy ramdisk? | 15:38 |
| *** rameshg87 has quit IRC | 15:38 | |
| dtantsur | lucasagomes, Nisha is prototyping doing it on request of a user, for example | 15:38 |
| lucasagomes | if the machine is already registered, the request will come from a known mac addr | 15:38 |
| jroll | lucasagomes: I think that's unclear, but we're leaving it open-ended | 15:38 |
| *** jcoufal has quit IRC | 15:38 | |
| lucasagomes | and this address will ahve a specific pxe config for it, so no autodiscovery ramdisk | 15:38 |
| lucasagomes | right | 15:39 |
| lucasagomes | idk sounds like over engineering | 15:39 |
| jroll | lol | 15:39 |
| jroll | dtantsur's specs just talks about how the discovery ramdisk will talk to ironic, and how ironic will handle it | 15:40 |
| lucasagomes | right, but my point is ... we already have those methods in the api | 15:40 |
| dtantsur | right, I have exactly nothing about how ramdisk will be there. but I'm consolidating several specs that are floating in the air and how they are going to work with Ironic | 15:41 |
| dtantsur | lucasagomes, where? | 15:41 |
| lucasagomes | properties can be updated by using PATCH, and creation of the node using POST | 15:41 |
| lucasagomes | you can GET the mac to see the node uuid | 15:41 |
| jroll | mmm, I see | 15:41 |
| lucasagomes | GET <macs> ... not registerd... register node POST, update the prorperties PUT | 15:41 |
| lucasagomes | boom | 15:41 |
| lucasagomes | no changes to the api, no redundant methods to create or update nodes | 15:42 |
| jroll | hmm, that's a good point | 15:42 |
| dtantsur | lucasagomes, 1. too many logic to put into ramdisk; 2. what if we change the way we discover node UUID; 3. how the ramdisk will now e.g. default driver? | 15:42 |
| lucasagomes | or I may be missing something here | 15:42 |
| dtantsur | now = know | 15:42 |
| dtantsur | lucasagomes, also you suggest at least twice more calls to the network (actually 1 + number of macs) | 15:42 |
| lucasagomes | dtantsur, right default driver could be something passed via the kernel cmd line to the ramdisk? | 15:43 |
| lucasagomes | and a config option in Ironic? | 15:43 |
| dtantsur | lucasagomes, and who is overengineering? ;) | 15:43 |
| lucasagomes | heh | 15:43 |
| *** rameshg87 has joined #openstack-ironic | 15:43 | |
| lucasagomes | I'm tlaking about 1 config option being added | 15:43 |
| lucasagomes | which is also present on that spec | 15:43 |
| lucasagomes | 2) I agree we may need something better later | 15:44 |
| lucasagomes | and for 2) we have something | 15:44 |
| *** shausy has quit IRC | 15:44 | |
| dtantsur | lucasagomes, how are you going to pass options to ramdisk, provided that DHCP configurations is static? | 15:44 |
| lucasagomes | once we have the arbitrary key and value pairs indexable by ironic | 15:44 |
| lucasagomes | we could save the idk some ID of the node | 15:44 |
| lucasagomes | so instead of GET macs you would GET that ID see if there's a node which contains it | 15:44 |
| lucasagomes | or multiple IDs | 15:44 |
| lucasagomes | "arbitrary tag support" | 15:45 |
| dtantsur | you're talking about some future feature that does not even have it's spec :-/ | 15:45 |
| lucasagomes | yeah | 15:45 |
| lucasagomes | but for now we can use MAC addrs | 15:45 |
| dtantsur | yes. and I don't see any reason your suggestion is better | 15:45 |
| lucasagomes | dtantsur, options are passed using the kernel cmd line | 15:45 |
| dtantsur | lucasagomes, how? should user update two places instead of one? | 15:46 |
| lucasagomes | dtantsur, it will be in the pxe configuration file | 15:46 |
| lucasagomes | dtantsur, the kernel options are there | 15:46 |
| *** viktors is now known as viktors|afk | 15:46 | |
| dtantsur | ok, that may be possible. I still see no reason why | 15:46 |
| lucasagomes | dtantsur, yeah u right about the network calls | 15:46 |
| lucasagomes | it's definately more calls | 15:46 |
| jroll | discovery should be a rare case, I wouldn't worry about api calls | 15:47 |
| dtantsur | moving logic from Ironic to the ramdisk is not making anything simple | 15:47 |
| lucasagomes | but almost no changes to ironic, less code to maintain, less redundancy | 15:47 |
| dtantsur | exactly the same amount of code | 15:47 |
| dtantsur | but moved to ramdisk | 15:47 |
| lucasagomes | no man | 15:47 |
| lucasagomes | ramdisk will do like at least 3 api calls | 15:47 |
| lucasagomes | get 1 mac, post and a put | 15:47 |
| lucasagomes | could be more depending on the number of macs | 15:47 |
| dtantsur | i.e. the same that I'm doing right now. Where is simplification? | 15:48 |
| lucasagomes | but definetely not the same ammount of code | 15:48 |
| jroll | + POST /ports for each mac | 15:48 |
| lucasagomes | jroll, yeah | 15:48 |
| jroll | if the node is new | 15:48 |
| dtantsur | + some folks will want to update ports, yeah | 15:48 |
| jroll | we maintain official ramdisks, yes? | 15:48 |
| lucasagomes | well the PUT is also not necessary you can create the nodes with the properties on it already | 15:48 |
| jroll | right | 15:48 |
| lucasagomes | u right about the POST to /ports | 15:49 |
| dtantsur | you're replacing code in Ironic making DB calls with code in _a lot of_ ramdisks doing the same by network call | 15:49 |
| lucasagomes | we gotta register those after the node | 15:49 |
| jroll | but you'll also want to do a GET for each mac in the new node case | 15:49 |
| *** matty_dubs is now known as matty_dubs|lunch | 15:49 | |
| lucasagomes | maybe we should do a hangout about it | 15:49 |
| jroll | dtantsur: why "a lot of ramdisks" | 15:49 |
| lucasagomes | ? | 15:49 |
| dtantsur | jroll, IPA, iLO, probably simple ramdisk - this is minimum what we have | 15:49 |
| dtantsur | yeah, iLO folks have there own ideas on OOB discovery on so on, which I was trying to embrace as well | 15:50 |
| jroll | hrm | 15:50 |
| dtantsur | ask Nisha or see his spec | 15:50 |
| jroll | right, oob would need to be different | 15:51 |
| lucasagomes | dtantsur, yeah mostly are replacing the calls in ironic for network calls from the ramdisk | 15:51 |
| dtantsur | and I didn't receive answer to question, how we have simplification by replacing Python functions call with network calls from ramdisk | 15:51 |
| jroll | but your spec doesn't help for oob discovery | 15:51 |
| dtantsur | jroll, it does provide some required bits | 15:51 |
| BadCub_ | morning all | 15:52 |
| jroll | dtantsur: kind of... but it all relies on /.../discovery being hit | 15:52 |
| dtantsur | at least it helped us agree on some common approaches, e.g. format of inventory to pass in | 15:52 |
| jroll | right | 15:52 |
| jroll | but OOB discovery won't pass anything in | 15:52 |
| jroll | ironic needs to pull the data | 15:52 |
| dtantsur | jroll, yeah, and he's introducing new mgmt function with _some_ result format. We agreed on standardized format for it | 15:53 |
| jroll | sure | 15:53 |
| jroll | I'm not totally sold on putting the logic in the ramdisk... but I also don't know why I didn't think of it :) | 15:53 |
| lucasagomes | I see the benefits on both approachs | 15:54 |
| dtantsur | jroll, unrelated. but we're starting discussion about storing inventory again in comments to IPA spec.... | 15:54 |
| jroll | dtantsur: yay. | 15:55 |
| dtantsur | well, folks, that's how I see it. You're free to -2 it, if you don't like. And I have my Czech lesson very soon, so have to go for now... | 15:57 |
| * lucasagomes feels a bit bad | 15:58 | |
| lucasagomes | I don't wanna be a blocker really | 15:58 |
| lucasagomes | I just think it could be simpler than it is being proposed | 15:59 |
| *** jcoufal has joined #openstack-ironic | 15:59 | |
| lucasagomes | simpler I mean, ironic offering basic functions: register and update {nodes, ports} | 15:59 |
| jroll | it is simpler on the ironic side | 15:59 |
| lucasagomes | ramdisk having the logic to interrogate the hardware and talk to the ironic api using that basic functions | 15:59 |
| jroll | and we could even code the ramdisk-side logic into the client | 16:00 |
| jroll | def create_or_update_node(...): | 16:00 |
| lucasagomes | and then we can start tunning the ironic api to help with it. e.g arbitrary tags to not have to GET macs, being able to register all the nodes ports within the same request to register the node itself | 16:00 |
| dtantsur | I would rather see the simplest possible ramdisk, because it's so harder to debug... | 16:00 |
| lucasagomes | but baby steps | 16:00 |
| lucasagomes | that's true | 16:01 |
| jroll | dtantsur: a ramdisk that has an init system isn't as hard to debug ;) | 16:01 |
| lucasagomes | but I don't think that X GET request and Y POST requests will make it _way_ more complicated than it will be | 16:01 |
| lucasagomes | we are talking about few conditionals | 16:02 |
| lucasagomes | and it may even be better, because if we detect that the mac is already registered | 16:02 |
| *** cdent_ has joined #openstack-ironic | 16:02 | |
| lucasagomes | we don't even need to interrogate the node | 16:02 |
| lucasagomes | nor send the inventory over the network | 16:02 |
| jroll | touché | 16:02 |
| jroll | well | 16:03 |
| jroll | we still will want to update the node | 16:03 |
| lucasagomes | yeah... and for that it's fine, cause we cna do partial updates on our api | 16:03 |
| *** amitpp has joined #openstack-ironic | 16:03 | |
| lucasagomes | so ok mac is registered u have the node id | 16:03 |
| lucasagomes | update only what needs to be updated | 16:03 |
| lucasagomes | shutdown | 16:03 |
| *** eghobo has joined #openstack-ironic | 16:04 | |
| jroll | right, partial updates, but we'll still need to interrogate the node | 16:04 |
| *** romcheg1 has quit IRC | 16:04 | |
| *** cdent has quit IRC | 16:04 | |
| *** cdent_ is now known as cdent | 16:04 | |
| lucasagomes | yeah true | 16:04 |
| lucasagomes | but there's nothing to do about it really heh | 16:04 |
| jroll | right | 16:04 |
| jroll | I'm going to shower / head to the office... I'll think about this some more | 16:05 |
| lucasagomes | yeah I will think about it as well. I may be missing something | 16:05 |
| *** hemna has joined #openstack-ironic | 16:05 | |
| dtantsur | I already imagine how to debug getting NodeLocked in the middle of discovery on 5 of 100 new machines... gonna be fun | 16:06 |
| dtantsur | with a new endpoint you can at least ensure it will fail only for serious reasons | 16:07 |
| lucasagomes | dtantsur, GET doesn't need locks... POST either | 16:07 |
| lucasagomes | neither* | 16:07 |
| dtantsur | lucasagomes, you mean patching node does not lock it? | 16:07 |
| lucasagomes | dtantsur, ah... yeah PATCH does lock the node :/ | 16:07 |
| lucasagomes | but that only will aftect updates to the node | 16:08 |
| lucasagomes | for creation you can POST with all properties | 16:08 |
| jroll | and for PATCH: | 16:08 |
| dtantsur | lucasagomes, yes, so updating existing node will need retrying | 16:08 |
| lucasagomes | dtantsur, yup, we also have a spec about having our API to retry it | 16:08 |
| jroll | if the node is locked... you're probably about to get deployed | 16:09 |
| lucasagomes | so that would alleviate that | 16:09 |
| jroll | which means your ramdisk is about to go away | 16:09 |
| jroll | or | 16:09 |
| lucasagomes | the NodeLock is a bigger problem than discovery it affects other stuff | 16:09 |
| lucasagomes | and need to be worked | 16:09 |
| jroll | better yet, put the node into maintenance when doing discovery | 16:09 |
| lucasagomes | jroll, +1 | 16:09 |
| jroll | that way it won't get locked for other things | 16:09 |
| lucasagomes | yeah maintenance will make the sync stop | 16:09 |
| jroll | won't get deployed etc | 16:09 |
| lucasagomes | on that node, so no lock | 16:09 |
| jroll | ok bbl for real | 16:10 |
| dtantsur | ok, I'm close to thinking that your approach is possible. Still don't see any advantage. | 16:10 |
| lucasagomes | jroll, ack cheers | 16:10 |
| *** lazy_prince is now known as killer_prince | 16:10 | |
| lucasagomes | mostly is keeping the ironic code base simpler | 16:11 |
| lucasagomes | I know we are shifting some to the ramdisk | 16:11 |
| dtantsur | by increasing some other code base :) | 16:11 |
| lucasagomes | few lines of code afaiui | 16:12 |
| lucasagomes | no new api endpoints | 16:12 |
| dtantsur | but much more API calls | 16:12 |
| lucasagomes | == no new tempest tests | 16:12 |
| lucasagomes | which can be worked later | 16:12 |
| lucasagomes | as I said, this is the first version | 16:12 |
| dtantsur | I would not call "no tempest tests" an advantage | 16:12 |
| lucasagomes | I would | 16:12 |
| Shrews | I WOULD! *giggle* | 16:13 |
| lucasagomes | because there's no new endpoints | 16:13 |
| lucasagomes | so no new tests needed | 16:13 |
| lucasagomes | less code to maintain | 16:13 |
| dtantsur | so you suggest to make feature non-testable, right? | 16:13 |
| lucasagomes | faster tests on gate | 16:13 |
| Shrews | but only b/c i've been in them so much lately | 16:13 |
| lucasagomes | dtantsur, no | 16:13 |
| lucasagomes | I mean, what is needed to have that feature is already being tested | 16:13 |
| lucasagomes | and already available to use | 16:13 |
| lucasagomes | which is creation of {nodes, ports} and updates of nodes | 16:13 |
| dtantsur | lucasagomes, yes, but not their integration as a discovery feature | 16:13 |
| dtantsur | I mean, we have everythings covered by unit tests, but we have also tempest for reasons | 16:14 |
| lucasagomes | that's the thing and why I think that keeping it simple is important | 16:14 |
| lucasagomes | because ironic only will have basic functions | 16:14 |
| lucasagomes | there's no difference between registering a node with the client | 16:15 |
| lucasagomes | or having the ramdisk doing it | 16:15 |
| lucasagomes | all we want is to have the node registered | 16:15 |
| dtantsur | ok, I'm also really going for the lesson. We anyway can't agree on it and have to ask devananda. | 16:15 |
| *** dtantsur is now known as dtantsur|afk | 16:16 | |
| lucasagomes | dtantsur|afk, ack, alright it's just an idea | 16:16 |
| lucasagomes | we can talk more about it | 16:16 |
| *** jistr has quit IRC | 16:16 | |
| *** athomas has quit IRC | 16:24 | |
| *** eghobo has quit IRC | 16:26 | |
| *** eghobo has joined #openstack-ironic | 16:26 | |
| *** foexle has quit IRC | 16:26 | |
| *** Alexei_987 has quit IRC | 16:30 | |
| *** ndipanov is now known as ndipanov_gone | 16:30 | |
| *** athomas has joined #openstack-ironic | 16:38 | |
| *** Nisha has joined #openstack-ironic | 16:40 | |
| *** killer_prince is now known as lazy_prince | 16:41 | |
| *** derekh_ has quit IRC | 16:42 | |
| *** stendulker has quit IRC | 16:43 | |
| cdent | lucasagomes: in case you're curious: it's "mostly working" https://tank.peermore.com/tanks/cdent-rhat/20140715 | 16:46 |
| lucasagomes | cdent, w00t! will take a look | 16:47 |
| devananda | morning, all | 16:56 |
| NobodyCam | good morning devananda | 16:56 |
| devananda | I'm attending the OSSG meetup today & tomorrow, will be semi offline | 16:56 |
| lucasagomes | morning devananda | 16:57 |
| devananda | lucasagomes: good evening! | 16:57 |
| devananda | rloo: err, if my email said Jul 24, then I goofed in the meeting yesterday | 16:58 |
| devananda | rloo: yesterday was the 14th.... heh | 16:58 |
| rloo | devananda: ok, so no -2s until the 24th then! | 16:59 |
| devananda | lucasagomes: i see you and dtantsur|afk wanted my opinion on something... reading scrollback | 16:59 |
| lucasagomes | devananda, me dtantsur|afk and jroll had some discussion about the autodiscovery, if you get a free time there take a look at it, food for though really, because we don't seem to agree much | 16:59 |
| lucasagomes | yeah | 16:59 |
| lucasagomes | it's about the autodiscovery | 17:00 |
| rloo | ifarkas: ^^^ the deadline is the 24th ;) | 17:00 |
| *** romcheg1 has joined #openstack-ironic | 17:02 | |
| *** pelix has quit IRC | 17:07 | |
| *** harlowja_away is now known as harlowja | 17:08 | |
| devananda | lucasagomes: ok, i think i've mostly digested the conversation | 17:08 |
| *** igordcard has joined #openstack-ironic | 17:09 | |
| devananda | lucasagomes: it sounds like there are a few very differnt approaches being proposed: OOB discovery (eg, iLO) where conductor interrogates the hw and adds/updates it in DB | 17:09 |
| devananda | - in-band method by a ramdisk that does not need to change the ironic API (any more than we already plan to) | 17:10 |
| devananda | - in-band method taht would make significant API changes | 17:10 |
| lucasagomes | yes | 17:10 |
| lucasagomes | the first one, the iLO thing is another spec | 17:11 |
| *** matty_dubs|lunch is now known as matty_dubs | 17:11 | |
| devananda | so (1) and (3) both propose API changes, but different ones | 17:11 |
| devananda | (2) proposes no API changes | 17:11 |
| lucasagomes | but I dtantsur|afk was working with Nisha to make sure their work won't overlap | 17:11 |
| devananda | well | 17:11 |
| devananda | that's a problem | 17:11 |
| devananda | I don't want us to end up with two different APIs for discovery | 17:12 |
| devananda | that's silly | 17:12 |
| lucasagomes | + | 17:12 |
| lucasagomes | +1 | 17:12 |
| *** tzumainn has joined #openstack-ironic | 17:12 | |
| devananda | I think Nisha's proposal conflates two things | 17:12 |
| devananda | - discovery | 17:12 |
| devananda | - re-inventory | 17:12 |
| lucasagomes | my arguments is mostly because we already have functions to create and update resources, I don't think we need to create other functions that will do the same thing | 17:13 |
| lucasagomes | but in a different endpoint | 17:13 |
| devananda | if we table any discussion of "how do we re-inventory known hardware" then things are much clearer | 17:13 |
| lucasagomes | URI* | 17:13 |
| devananda | and I think both IB and OOB discovery can be _initiated_ using a single API | 17:14 |
| devananda | under /v1/drivers/<driver_name> | 17:14 |
| lucasagomes | right | 17:14 |
| devananda | pass the driver enough information to tell it where to go look | 17:14 |
| lucasagomes | and by initiated you mean, tell ironic to for e.g prepare the pxe env for the autodiscovery ramdisk? | 17:14 |
| devananda | for an IB / DHCP based discovery, like IPA, that would be just a boolean | 17:14 |
| devananda | "enable discovery ramdisk for DHCP BOOT requests from unknown MAC" or "disable ..." | 17:15 |
| lucasagomes | +1 | 17:15 |
| devananda | for an OOB / IPMI based discovery, like iLO, that might be an IP range on the IPMI LAN | 17:15 |
| Nisha | devananda:yes | 17:15 |
| devananda | "go poke 10.0.0.0/24 using username=foo,password=bar and see if there are new nodes" | 17:15 |
| devananda | this is a single new API endpoint taht covers both IB and OOB discovery | 17:16 |
| devananda | updates to known hardware can then be done in two stages without any further changes | 17:16 |
| devananda | - delete node $UUID | 17:16 |
| devananda | - discover node based on <known property> | 17:16 |
| *** rameshg871 has joined #openstack-ironic | 17:17 | |
| devananda | or if the operator already knows what has changed, he can simply issue PATCH request to change it | 17:17 |
| devananda | or she | 17:17 |
| devananda | :) | 17:17 |
| lucasagomes | right this tihng about update known hardware to me, seems like a diff feature than auto discovery | 17:18 |
| devananda | yes | 17:18 |
| devananda | it clouds the discussion, has a different API, and I'm not even sure who wants it | 17:19 |
| devananda | none of the operators at the last summit brought that up | 17:19 |
| lucasagomes | yeah | 17:19 |
| devananda | in fact, some even said "delete and re-discover is fine" | 17:19 |
| devananda | ** assuming taht discovery works :) | 17:19 |
| lucasagomes | so I agree with you in adding a api point to enable/disable the discovery | 17:19 |
| *** rameshg87 has quit IRC | 17:19 | |
| lucasagomes | I first thought about a config otpion but an api to make it dynamic may be better | 17:19 |
| lucasagomes | like PUT /drivers/<,,,>/discovery to enable and DELETE /drivers/<,,,>/discovery to disable | 17:20 |
| devananda | starting/stopping discovery should be possible without restarting cluster | 17:20 |
| lucasagomes | yeah | 17:20 |
| devananda | and should be possible per-driver | 17:20 |
| devananda | like ^ | 17:20 |
| lucasagomes | sounds good | 17:20 |
| devananda | it would be great to have a standard parameter set for that endpoint | 17:21 |
| devananda | like IP range, user, pass | 17:21 |
| *** igordcard has quit IRC | 17:21 | |
| devananda | but i'm not sure we can | 17:21 |
| Nisha | devananda: This will not require new CLI? | 17:21 |
| devananda | it might need to be another driver-specific dict, like the driver_info fields | 17:21 |
| devananda | which drivers need to expose | 17:21 |
| *** Penick has joined #openstack-ironic | 17:22 | |
| *** kylestev_ has joined #openstack-ironic | 17:22 | |
| lucasagomes | devananda, right, or simply being passed as the body request | 17:22 |
| devananda | Nisha: I'm suggesting adding a new API endpoint, so yes, it will need a change to the CLI to communicate with that | 17:22 |
| lucasagomes | and each driver will know how to handle it | 17:22 |
| lucasagomes | kinda like the vendor passthru | 17:22 |
| devananda | lucasagomes: sure - but again, the driver needs to expose /what/ it expects in that body | 17:22 |
| lucasagomes | yeah | 17:22 |
| lucasagomes | right... ok this needs to be sorted | 17:23 |
| devananda | whereas vendor passthru does not expose what it expects, I think this needs to, since I'd like it to be a standard part of the API | 17:23 |
| devananda | it won't be "core" since not all drivers can support discovery, but it should be "common" | 17:23 |
| devananda | actually. maybe they can | 17:24 |
| *** Penick has quit IRC | 17:24 | |
| devananda | anyway. that's a tangent | 17:24 |
| lucasagomes | yeah, we can iron that out later | 17:24 |
| *** Penick has joined #openstack-ironic | 17:26 | |
| *** cdent has quit IRC | 17:28 | |
| lucasagomes | devananda, alright thanks for the input... I will think on it | 17:28 |
| lucasagomes | see if I can put on an etherpad or something | 17:28 |
| Nisha | devananda: lucasagomes so from CLI perspective you plan to introduce a new CLI or use existing CLI's for autodiscovery? | 17:29 |
| *** rameshg871 has quit IRC | 17:29 | |
| devananda | Nisha: I dont understand your question. why would we create a nwe command line interface? | 17:29 |
| lucasagomes | Nisha, well if we add that endpoint to enable/disbale the autodiscovery in the API we will add it to the CLI as well | 17:30 |
| devananda | Nisha: also, CLI changes are a secondary effect of API changes, which need to be very deliberate | 17:30 |
| lucasagomes | def not creating a new CLI for it | 17:30 |
| devananda | lucasagomes: thanks. It'd be great if we can sort this out by / at the midcycle, otherwise I doubt we'll be able to land it in Juno | 17:31 |
| devananda | lucasagomes: given how much disagreement there seems to be right now | 17:31 |
| Nisha | devananda: ohk...actually i meant that how OOB will invoke autodiscovery since it will require IP range etc as input.... | 17:32 |
| devananda | Nisha: right. I already discussed that | 17:32 |
| lucasagomes | devananda, yup, we really want to have this feature in J so we probably are going to work hard on it to solve the misunderstands and all | 17:32 |
| devananda | 17:14:00 < devananda> and I think both IB and OOB discovery can be _initiated_ using a single API | 17:32 |
| devananda | 17:14:07 < devananda> under /v1/drivers/<driver_name> | 17:32 |
| devananda | 17:15:30 < devananda> for an OOB / IPMI based discovery, like iLO, that might be an IP range on the IPMI LAN | 17:33 |
| Nisha | ok | 17:33 |
| devananda | lucasagomes: quick poke re: python-ironicclient reviews. | 17:34 |
| devananda | lucasagomes: anything I shouldn't approve? :) | 17:34 |
| *** igordcard has joined #openstack-ironic | 17:34 | |
| lucasagomes | devananda, oh, the pagination was approved but jenkins is failing | 17:34 |
| devananda | ugh | 17:34 |
| lucasagomes | devananda, other than that, there's also the mrda patch adding the cache | 17:34 |
| devananda | I'd love to get a new release tagged with that | 17:34 |
| devananda | and that | 17:34 |
| devananda | :) | 17:34 |
| lucasagomes | devananda, yeah it's a werid error when fetching the packages from pip | 17:34 |
| lucasagomes | I will recheck that | 17:34 |
| Nisha | devananda: Ok i will work with lucasagomes , dtantsur|afk and see how it goes | 17:35 |
| devananda | Nisha: thanks! | 17:35 |
| lucasagomes | the error: http://logs.openstack.org/85/91585/7/check/gate-python-ironicclient-python26/2939d4e/console.html | 17:35 |
| lucasagomes | kg_resources.DistributionNotFound: virtualenv>=1.9.1 | 17:35 |
| *** dtantsur|afk is now known as dtantsur | 17:35 | |
| dtantsur | devananda, lucasagomes: is there some conclusion on what we started with, i.e. autodiscovery? | 17:36 |
| *** geekyogi has joined #openstack-ironic | 17:36 | |
| devananda | dtantsur: read scrollback :) | 17:36 |
| devananda | dtantsur: not a conclusion, but I have some suggestions | 17:37 |
| devananda | dtantsur: which I think make both the discussion and the implementation simpler and achievable in Juno | 17:37 |
| devananda | if everyone agrees | 17:37 |
| lucasagomes | will approve mrda patches on the client | 17:38 |
| dtantsur | it's probably evening, but I can't find it :( it's ok, we moved OOB out of the discussion, but it's not clear for me, whether we want logic on ramdisk leevel or on ironic level | 17:38 |
| *** rloo has quit IRC | 17:38 | |
| devananda | rloo: https://blueprints.launchpad.net/ironic/+spec/get-required-driver-info approved and targeted | 17:39 |
| lucasagomes | dtantsur, afaiui, Ironic needs an endpoint to enable/disable the autodiscovery | 17:40 |
| lucasagomes | for e.g for PXE someone will do a POST /drivers/pxe/autodiscovery | 17:40 |
| lucasagomes | and enable it | 17:40 |
| dtantsur | lucasagomes, why not use a configuration option? | 17:41 |
| dtantsur | True or False :) | 17:41 |
| *** tzumainn has left #openstack-ironic | 17:41 | |
| lucasagomes | so the PXE driver will prepare the PXE enviroment and generate the default config files where requests from mac _not registered_ in ironic will boot from | 17:41 |
| lucasagomes | dtantsur, cause we want it to be dynamic and per driver | 17:41 |
| lucasagomes | dtantsur, if it's a config you will have to stop start the service | 17:41 |
| dtantsur | lucasagomes, ah, ok, I though you're discussing my spec | 17:42 |
| dtantsur | because this part is completely irrelevant to it | 17:42 |
| lucasagomes | (our services doesn't understand KILL -HUP to re-read conf unfortunately) | 17:42 |
| lucasagomes | our == openstack | 17:42 |
| lucasagomes | dtantsur, yes, that's irrelevant for that spec | 17:42 |
| lucasagomes | but that's also a common group, _all_ drivers will need a way to disable/enable dynamically the autodiscovery | 17:42 |
| lucasagomes | so that endpoint should be in the spec, not the driver implementation | 17:43 |
| lucasagomes | afaiui | 17:43 |
| *** bvivek has joined #openstack-ironic | 17:43 | |
| dtantsur | lucasagomes, my spec does not touch initiating the autodiscovery at all. It discusses part after ramdisk discovered something | 17:43 |
| dtantsur | so it's very good, but needs one more spec written by someone (maybe me) | 17:43 |
| lucasagomes | right... | 17:44 |
| lucasagomes | after the ramdisk is discovered? after it's boot? | 17:44 |
| dtantsur | lucasagomes, that's what my spec and today's arguments is about :) I though you were discussing it... | 17:45 |
| devananda | ifarkas: hi! drac power driver BP approved. do you think code will be be ready by J2 (next thursday) ? | 17:45 |
| dtantsur | I mean, for the simpliest case we don't need to have any initialization, just a DHCP/PXE server with appropriate configuration | 17:45 |
| lucasagomes | dtantsur, right... so one thing about the _after_ | 17:45 |
| lucasagomes | dtantsur, discovery vs updating known hardware are diff things | 17:46 |
| dtantsur | that's why I'm thinking you're discussing step #2 | 17:46 |
| dtantsur | while we don't have step #1 | 17:46 |
| lucasagomes | dtantsur, we need initialization | 17:46 |
| devananda | so let's set "update known hardware" completely aside. For now, it can be approximated with "delete then rediscover" | 17:46 |
| *** rameshg87 has joined #openstack-ironic | 17:47 | |
| dtantsur | lucasagomes, we will, but we can live without it at first, because static configuration is a good baby step | 17:47 |
| dtantsur | devananda, will we account for the case, when ramdisk is by mistake booted on a known node? | 17:47 |
| dtantsur | or will we create a new node inconditionally? | 17:48 |
| lucasagomes | dtantsur, static configuration is not a good idea IOM | 17:48 |
| lucasagomes | IMO* | 17:48 |
| devananda | also, static external DHCP-based discovery is a trivial oversimplification. a proposal handling /only/ that case will impact the ability of OOB systems' discovery | 17:48 |
| lucasagomes | cause there's no way to disable it | 17:48 |
| lucasagomes | also u will have to create the pxe configs by hand? | 17:48 |
| lucasagomes | but let's put it aside | 17:48 |
| dtantsur | so, you folks don't like baby steps and what to prototype all the abilities at once? | 17:48 |
| devananda | no | 17:48 |
| dtantsur | that we have to think about authentication as well | 17:48 |
| *** kylestev_ has left #openstack-ironic | 17:48 | |
| devananda | I'm proposing an incremental approach | 17:49 |
| devananda | whcih doesn't block further work | 17:49 |
| dtantsur | me too | 17:49 |
| lucasagomes | alright... ok even if it's static | 17:49 |
| lucasagomes | the way we register the nodes after the ramdisk is being booted | 17:49 |
| rameshg87 | devananda, dtantsur, lucasagomes, request your review on ilo power code: https://review.openstack.org/#/c/89500/ , thanks ... | 17:49 |
| lucasagomes | IMO we should still use POST | 17:50 |
| lucasagomes | on the normal /nodes | 17:50 |
| lucasagomes | instead of having a new endpoint for it | 17:50 |
| devananda | dtantsur: of course we need to handle the case where a known node is re-discovered. but we already do -- MAC address is unique | 17:50 |
| *** rameshg87 has left #openstack-ironic | 17:50 | |
| devananda | dtantsur: though POST /v1/nodes/ to create a new node doesn't (and shouldn't) have any uniqueness checks | 17:50 |
| devananda | dtantsur: this simply means that IB-based (ramdisk-based) discovery needs to do some checking (like GET /v1/ports?MAC=) before creating new node | 17:51 |
| dtantsur | devananda, yes, you have to make ramdisk to do all the checks with ports before | 17:51 |
| devananda | which I think is what lucasagomes was already suggesting | 17:51 |
| Nisha | devananda: could u explain what u mean by "uniqueness checks"? ^^^^ | 17:51 |
| dtantsur | I hate moving logic to the ramdisk, but as you wish | 17:51 |
| dtantsur | I'm abandoning my spec, right? | 17:51 |
| devananda | dtantsur: so I didn't say the ramdisk itself has to do that work :) | 17:51 |
| *** rloo has joined #openstack-ironic | 17:51 | |
| devananda | dtantsur: jsut taht that check needs to be done before the creation | 17:52 |
| devananda | but... | 17:52 |
| devananda | er, so... | 17:52 |
| devananda | dtantsur: what about this | 17:52 |
| devananda | ramdisk POSTs all the data to API service | 17:52 |
| lucasagomes | devananda, yes! ramdisk does the logic to discovery if the node id based on the ports | 17:52 |
| devananda | API service contains all the logic of | 17:52 |
| devananda | - interpret and validate POSTed data (sanity checking, auth, etc) | 17:52 |
| devananda | - look up all MAC addresses contained in the body | 17:53 |
| devananda | - if no duplicates, create new node, create port(s) associated to node | 17:53 |
| dtantsur | devananda, that is _exactly_ what I am proposing, I guess. lucasagomes is against creating an API service for it. | 17:53 |
| devananda | - respond success | 17:53 |
| devananda | lucasagomes: so taht doesn't require any conductor work -- it's just API and DB interactions | 17:53 |
| lucasagomes | but using the same endpoint right? to create the node with its ports within the same request | 17:54 |
| devananda | dtantsur: i haven't read your current spec version .... sorry | 17:54 |
| lucasagomes | <lucasagomes> and then we can start tunning the ironic api to help with it. e.g arbitrary tags to not have to GET macs, being able to register all the nodes ports within the same request to register the node itself | 17:54 |
| lucasagomes | I suggested something like that as an improvement | 17:54 |
| dtantsur | lucasagomes, our current endpoint does not accept MAC's... | 17:54 |
| lucasagomes | yeah... as an improvement | 17:54 |
| lucasagomes | later | 17:54 |
| lucasagomes | for now using the MACs seems resonable | 17:54 |
| lucasagomes | I mean using GET on ports to find the node id | 17:55 |
| lucasagomes | before creation | 17:55 |
| devananda | so, right now, all this can be done using the CLI, by the ramdisk | 17:55 |
| devananda | adding a nwe API endpoint which encapsulates all the logic of "create a node with these ports, if one does not exist" | 17:55 |
| devananda | would be generally a useful improvement | 17:55 |
| devananda | and has nothing intrisically to do with autodiscovery | 17:55 |
| devananda | operators would in general benefit from that | 17:56 |
| devananda | as it makes enrollment of known inventory faster, too | 17:56 |
| devananda | eg, writing a script to import a bunch of server data in a .csv file would be easier if you could just POST once per server, rather than once per server + once for every MAC | 17:56 |
| dtantsur | devananda, maybe you'll have a quick look at my spec? I already do not know what we're actually arguing, it looks like we're proposing the same... | 17:57 |
| jroll | I haven't read scrollback yet, but is there any operator with a sizable deployment that won't be scripting their bootstrap? | 17:57 |
| devananda | I get why lucasagomes doesn't want to tie that API change to autodiscevery -- I think they're separate things | 17:57 |
| devananda | jroll: nope | 17:57 |
| jroll | and is 1+n*macs slow enough to actually matter, and need to create a new endpoint for? | 17:57 |
| jroll | (that many POST requests) | 17:57 |
| devananda | jroll: i don't think that the perf here matters, honestly, but OTOH it's a shortcut | 17:58 |
| jroll | like, if I'm enrolling 1000 nodes... I don't care if it takes 10 minutes or 30 | 17:58 |
| dtantsur | my point is not only speed, but in context of ramdisk, complexity of ramdisks costs a lot more than complexity of Ironic itself | 17:58 |
| devananda | POST (server and all relevant data) || POST (server) + N * POST (NICS in server) | 17:58 |
| dtantsur | in terms of debugging and testing | 17:58 |
| JayF | dtantsur: O | 17:58 |
| JayF | dtantsur: I'm not sure I completely agree with that | 17:58 |
| JayF | dtantsur: logic run in a ramdisk is distributed across all your nodes | 17:58 |
| JayF | dtantsur: so instead of having a small cluster of conductors doing things, I have idle hardware doing things, which scales more naturally | 17:59 |
| jroll | devananda: right... I think it's negligible if the operator does any sort of capacity planning, at all | 17:59 |
| *** tatyana has joined #openstack-ironic | 17:59 | |
| JayF | devananda: if you're going to make a shortcut, make a multiple-create endpoint | 17:59 |
| dtantsur | JayF, not it does not, because for database access you still go to the conductor | 17:59 |
| jroll | there will never be a hair on fire situation where I need to add nodes 3 times as fast | 17:59 |
| JayF | devananda: that would be one call to register multiple nodes | 17:59 |
| devananda | dtantsur: not for creation you don't | 17:59 |
| dtantsur | devananda, checking NICs? | 17:59 |
| devananda | dtantsur: only for update, which is one reason I suggested we table any discussion of using discovery for update | 17:59 |
| devananda | dtantsur: correct -- taht's just a DB lookup | 18:00 |
| devananda | dtantsur: the API can do that jsut fine | 18:00 |
| devananda | only updating of existing resoruces requires lock coordination, ergo only updates require the conductor | 18:00 |
| lucasagomes | hey folks I will have to step back a little | 18:00 |
| dtantsur | sorry, I meant API service, not conductor | 18:00 |
| devananda | (delete being a type of update, too) | 18:01 |
| lucasagomes | step out* | 18:01 |
| devananda | lucasagomes: cheers, ttyl | 18:01 |
| *** romcheg1 has quit IRC | 18:01 | |
| NobodyCam | night lucasagomes | 18:01 |
| devananda | JayF: the problem with multi-node-create is what to do if one fails | 18:01 |
| dtantsur | I mean, doing logic in ramdisk won't scale better, because the bottleneck is mostly database and database accesses will be the same (but over REST API) | 18:01 |
| lucasagomes | thanks I will read the scrollback later | 18:02 |
| *** lucasagomes is now known as lucas-afk | 18:02 | |
| devananda | so no one is going to be enrolling nodes so quickly that scalability will seriously matter | 18:02 |
| NobodyCam | anyone up for a quick review of: https://review.openstack.org/#/c/106905 | 18:02 |
| JayF | devananda: I'm with jroll in saying I don't think we need a new endpoint. But if you really wanted to make it more convienient, multi-create would be more useful imo :) | 18:02 |
| devananda | seriously - compared to the scaling issues we already know we have with deploying a fleet, | 18:02 |
| devananda | registering a fleet takes miniscule amount of time | 18:03 |
| jroll | heh | 18:03 |
| *** romcheg1 has joined #openstack-ironic | 18:03 | |
| JayF | Yes, exactly | 18:03 |
| devananda | dtantsur's point is that putting the logic into a ramdisk is harder to debug than putting it in the API service | 18:03 |
| dtantsur | this is my major point ^^^ | 18:04 |
| devananda | lucas' point is that we dont need to make any API changes to accomplish the same functionality | 18:04 |
| jroll | ok, so | 18:04 |
| jroll | if the log is in python-ironicclient... | 18:04 |
| jroll | is that harder to debug? | 18:04 |
| jroll | s/log/logic | 18:04 |
| jroll | (answer is likely yes, but less so than logic in ramdisk) | 18:04 |
| dtantsur | that does not solve problem with ramdisk exiting in the middle of the process on 5 of 20 machines... that's why I want to leave minimum logic on ramdisk side | 18:05 |
| devananda | that's python code with unit tests. compared to a bash script, I'd say it's probably easier to debug and maintain | 18:05 |
| jroll | gah, ramdisk is all bash, isn't it | 18:05 |
| * jroll sadface | 18:05 | |
| devananda | dtantsur: huh? what does "5 of 20 machiens" have to do with this? | 18:06 |
| jroll | dtantsur: but, does that happen today? why would that happen by adding this? | 18:06 |
| dtantsur | debugging PXE faults is already a not-too-pleasant experience | 18:06 |
| devananda | dtantsur: enrollment is per-machine. if the ramdisk crashes halfway through enrollment, THATs gonna be harder to debug regardless. | 18:06 |
| devananda | *regardless of what the ramdisk is doing | 18:06 |
| dtantsur | ok, 5/20 is irrelevant, debugging even one may be a problem | 18:06 |
| devananda | dtantsur: also, "leave minimum logic on ramdisk side" is counter to all the work IPA is donig | 18:07 |
| devananda | doing | 18:07 |
| NobodyCam | brb | 18:07 |
| devananda | to make the ramdisk more programatic, testable, and, well, generate logging that we can consume externally | 18:07 |
| *** overlayer has joined #openstack-ironic | 18:07 | |
| jroll | well, to be fair, we have an init system, logging, and ssh server in IPA :) | 18:07 |
| dtantsur | devananda, minimum != none. | 18:07 |
| dtantsur | my view on IPA is that it encapsulates all hard moments with dealing with remote hardware | 18:08 |
| dtantsur | (like partitioning, firmware updates etc) | 18:08 |
| devananda | dtantsur: IPA already has several call-and-response interactions with their driver | 18:08 |
| devananda | adding this logic for discovery to it is not a significant change AIUI | 18:09 |
| *** jbjohnso_ has joined #openstack-ironic | 18:09 | |
| devananda | jroll: please tell me if I"m wrong :) | 18:09 |
| jroll | right, this logic would be nothing for ipa | 18:09 |
| dtantsur | than where's the line between what we put to Ironic and what we put to IPA? | 18:10 |
| devananda | dtantsur: OTOH, adding this to the current PXE ramdisk, whcih is bash and lacks a multi-user environment, would certanly be possible, but not easy to debug, i agree | 18:10 |
| jroll | although, this logic is more than we have now | 18:10 |
| dtantsur | yeah, it ruins hopes to have a simple bash-based discovery before IPA is done... | 18:10 |
| jroll | ironic side currently does the mac -> node mapping | 18:10 |
| jroll | OTOH, that doesn't do node creation or anything | 18:11 |
| *** jbjohnso has quit IRC | 18:11 | |
| dtantsur | I actually borrowed a lot of mapping code from jroll's implementation :) so IPA already has some of this logic in Ironic | 18:11 |
| jroll | right | 18:11 |
| devananda | dtantsur: no. it accepts the limitations of the current ramdisk -- it's harder to debug failures | 18:11 |
| jroll | dtantsur: but that's only because it makes it easier to change that logic | 18:12 |
| jroll | to map by something other than macs | 18:12 |
| *** geekyogi has quit IRC | 18:12 | |
| *** overlayer has quit IRC | 18:12 | |
| *** igordcard has quit IRC | 18:12 | |
| jroll | but that is *driver-specific* for us, not a generic thing | 18:12 |
| devananda | dtantsur: the more we discuss this, the more I agree with lucas' point -- we shouldn't make API changes just to work around "PXE ramdisk is hard to debug" | 18:12 |
| jroll | (unclear if that matters) | 18:12 |
| *** igordcard has joined #openstack-ironic | 18:13 | |
| jroll | I think that logic should be up to the driver... whether that is on the ramdisk side or ironic, as well | 18:13 |
| dtantsur | I hoped we can have at least some generic peaces... | 18:15 |
| *** bvivek has quit IRC | 18:15 | |
| jroll | I'm not opposed to that... but I think the generic piece should just look to the driver. maybe some default implementation... but then again that default implementation can be done the way lucas mentioned | 18:16 |
| jroll | idk | 18:16 |
| dtantsur | ok, I leave it up to IPA to define the discovery and abandon my spec, right? | 18:16 |
| jroll | not sure about that, either | 18:17 |
| * jroll thinks | 18:17 | |
| dtantsur | because while the complexity is ok for IPA, I won't dare to implement it for PXE | 18:18 |
| dtantsur | I don't than, will it be possible for J | 18:18 |
| NobodyCam | gah anyone know the correct rights for /etc/sudoers.d directory? | 18:20 |
| dtantsur | 0755? | 18:20 |
| jroll | sounds like that should only be readable by root | 18:21 |
| dtantsur | well, maybe actually 0700... | 18:21 |
| jroll | 0700 | 18:21 |
| jroll | yeah | 18:21 |
| NobodyCam | :) | 18:21 |
| NobodyCam | owned by root | 18:21 |
| -openstackstatus- NOTICE: python2.6 jobs are failing due to bug 1342262 "virtualenv>=1.9.1 not found" A fix is out but there are still nodes built on the old stale images | 18:22 | |
| *** ChanServ changes topic to "python2.6 jobs are failing due to bug 1342262 "virtualenv>=1.9.1 not found" A fix is out but there are still nodes built on the old stale images" | 18:22 | |
| Shrews | ooh, fun | 18:23 |
| NobodyCam | woo hoo can we depercate 2.6? | 18:23 |
| NobodyCam | oh wait I'm actually working on suse which IS 2.6 | 18:24 |
| jroll | hahaha | 18:24 |
| * Shrews deprecates NobodyCam | 18:25 | |
| NobodyCam | Shrews: jroll: up for a easy review? https://review.openstack.org/#/c/106905/ | 18:26 |
| Shrews | sure | 18:26 |
| Shrews | NobodyCam: so, why change the default? why can't tripleo just set the value in .conf? | 18:28 |
| jroll | Shrews: kind of a shitty default, though, don't you think? | 18:28 |
| NobodyCam | ya | 18:28 |
| jroll | s/tripleo/every deployer/ | 18:28 |
| NobodyCam | just too small | 18:28 |
| Shrews | well, yeah, i guess | 18:28 |
| NobodyCam | :-p | 18:28 |
| jroll | I'd say it's pretty unusable at 1G | 18:28 |
| Shrews | so how did we land on the *new* default? | 18:28 |
| jroll | also arbitrary :) | 18:28 |
| NobodyCam | if was 10G and TripleO was looking for 20 I might push back | 18:29 |
| NobodyCam | but 1 k really kinda small | 18:29 |
| NobodyCam | s/k/g/ | 18:29 |
| jroll | 1k is super small :P | 18:29 |
| Shrews | heh | 18:29 |
| *** rakesh_hs has quit IRC | 18:30 | |
| dtantsur | Well, folks, have to go anyway. You can express opinions on the spec by using -2/+2 :) | 18:30 |
| *** dtantsur is now known as dtantsur|afk | 18:30 | |
| NobodyCam | have a good night dtantsur|afk | 18:30 |
| jroll | see ya, dtantsur|afk | 18:30 |
| *** jbjohnso__ has joined #openstack-ironic | 18:35 | |
| * NobodyCam has to go pick the bubbie.. be back in a bit. | 18:37 | |
| *** Penick has quit IRC | 18:37 | |
| *** jbjohnso_ has quit IRC | 18:39 | |
| *** Penick has joined #openstack-ironic | 18:40 | |
| *** jbjohnso__ has quit IRC | 18:42 | |
| *** lazy_prince is now known as killer_prince | 18:49 | |
| * devananda lunches | 18:58 | |
| *** derekh_ has joined #openstack-ironic | 18:59 | |
| *** anneonthebus has quit IRC | 18:59 | |
| *** jdob_ has joined #openstack-ironic | 19:00 | |
| *** Penick has quit IRC | 19:01 | |
| *** jdob_ has quit IRC | 19:01 | |
| *** jistr has joined #openstack-ironic | 19:06 | |
| *** Nisha has quit IRC | 19:07 | |
| Shrews | where art thou, openstackgerrit bot? | 19:20 |
| *** Mikhail_D_ltp has joined #openstack-ironic | 19:20 | |
| *** tatyana_ has joined #openstack-ironic | 19:22 | |
| *** tatyana has quit IRC | 19:23 | |
| *** tatyana_ is now known as tatyana | 19:23 | |
| * NobodyCam is back | 19:27 | |
| *** Penick has joined #openstack-ironic | 19:28 | |
| NobodyCam | Ty Shrews for 10695 | 19:28 |
| *** Penick has quit IRC | 19:28 | |
| NobodyCam | lifeless: just fyi 106905 landed! | 19:29 |
| *** Penick has joined #openstack-ironic | 19:32 | |
| * devananda is back | 19:38 | |
| NobodyCam | wb devananda | 19:38 |
| devananda | let's see if i can squash up a new nova driver patch | 19:39 |
| devananda | also, anyone want to review https://review.openstack.org/#/c/102695/8 ? | 19:39 |
| NobodyCam | oh there has been a todo for that for a while now | 19:39 |
| *** lucas-afk has quit IRC | 19:40 | |
| *** Mikhail_D_ltp has quit IRC | 19:41 | |
| NobodyCam | devananda: did you see http://logs.openstack.org/95/102695/8/check-tripleo/check-tripleo-ironic-undercloud-precise/d451bb0/logs/seed_logs/nova-compute.txt.gz#_2014-07-15_05_53_59_927 | 19:42 |
| devananda | looking | 19:42 |
| devananda | ooh | 19:42 |
| NobodyCam | that kinda looks real. | 19:42 |
| devananda | it does | 19:42 |
| devananda | interesting that it passed tempest tho | 19:42 |
| NobodyCam | at least I haven't seen it on the rechecks of late | 19:43 |
| devananda | yea, that could well be related to this change | 19:43 |
| devananda | or to another change that I saw proposed recently | 19:43 |
| NobodyCam | I can no bug recheck it | 19:43 |
| devananda | no... | 19:43 |
| NobodyCam | see what we get | 19:43 |
| NobodyCam | ok | 19:43 |
| NobodyCam | :-p | 19:44 |
| devananda | as a general rule, please avoid "recheck no bug" unless there's a really good reason | 19:44 |
| devananda | that goes for ALL prjoects. it just hurts openstack when we do that | 19:44 |
| NobodyCam | ack .. I do try and not no bug things | 19:44 |
| *** openstackgerrit has joined #openstack-ironic | 19:44 | |
| *** tatyana has quit IRC | 19:45 | |
| NobodyCam | brb | 19:49 |
| *** jcoufal has quit IRC | 19:52 | |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Reduce severity of generic discrete assert to 'Ok' https://review.openstack.org/107164 | 19:59 |
| *** jbjohnso has joined #openstack-ironic | 20:00 | |
| Shrews | some errors are just very strange though... and no logs | 20:03 |
| Shrews | e.g., https://review.openstack.org/105583 | 20:05 |
| *** annegentle has joined #openstack-ironic | 20:06 | |
| Shrews | i wouldn't even know where to begin to file a bug about that failure so that we can recheck on it | 20:06 |
| adam_g | Shrews, hmm | 20:09 |
| openstackgerrit | A change was merged to openstack/ironic: Make ComputeCapabilitiesFilter work with Ironic https://review.openstack.org/105802 | 20:15 |
| adam_g | Shrews, https://review.openstack.org/#/c/107165/ this may help | 20:16 |
| Shrews | adam_g: is that a backported change? | 20:18 |
| adam_g | Shrews, yeah just proposed it. we removed that at some point recentlyish, its possible something in infra changed that prevents the old behavior from working as it did? | 20:18 |
| Shrews | ah, i see the cherry pick comment now | 20:19 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add versioning to Agent decommission https://review.openstack.org/106859 | 20:20 |
| adam_g | Shrews, also FYI, poking tempest with all recent patches + in flight stuff and all but one compute API test passes from what we currently run (ie http://logs.openstack.org/27/103227/2/check/check-tempest-dsvm-ironic/0f5e4b1/logs/testr_results.html.gz) | 20:21 |
| Shrews | adam_g: what did you do for cinder? | 20:21 |
| adam_g | Shrews, oh, just set cinder = False. actually there were a couple of flags that needed toggle in addition to patches | 20:22 |
| adam_g | the one thats still failing (test_server_actions) should be fixable with a new flag added for console output support | 20:23 |
| Shrews | speaking of which.... devananda, did you see mtreinish's email response re: cinder+ironic? | 20:23 |
| Shrews | adam_g: all good news... but there were a bunch of server tests failing according to your etherpad | 20:24 |
| Shrews | what was the fix for those? | 20:25 |
| Shrews | tempest/api/compute/servers/ | 20:25 |
| devananda | Shrews: yes, havn't had time to respond. you want to? | 20:25 |
| Shrews | devananda: sounds like you and he need to reach a consensus. he wants cinder disabled, you don't. :( | 20:26 |
| adam_g | Shrews, let me update | 20:27 |
| devananda | Shrews: I'd love to disable it AND other things. talk with -infra. AFAIK clarkb expressly rejects that sort of non-co-gating | 20:27 |
| devananda | I was sort of parrotting infra's view, but I also agree with clark on it | 20:27 |
| devananda | even though selfishly i'd love our gate to be faster | 20:27 |
| Shrews | devananda: i'll ask him to weigh in on it then | 20:28 |
| adam_g | Shrews, oh i forgot to mention, im running with IRONIC_VM_COUNT=3, so scale is not an issue anymore, at least serially | 20:29 |
| Shrews | adam_g: ah. k | 20:31 |
| jbjohnso | hmm, I don't see my last review on pyghmi doing anything jenkins wise | 20:35 |
| Shrews | devananda: clark will respond tomorrow (in germany atm). he has since changed his viewpoint | 20:36 |
| devananda | Shrews: neat! | 20:36 |
| Shrews | indeed | 20:36 |
| jbjohnso | on a somewhat related note, I'm starting to get more in the territory where commentary on some hardware management and authentication for confluent if anyone is interested | 20:36 |
| Shrews | adam_g: i'll unabandon my 'disable cinder' review and fix it up per clark's suggestion | 20:37 |
| adam_g | Shrews, whats his suggestion? i cant keep track anymore :) | 20:38 |
| Shrews | adam_g: use features.yaml to disable it | 20:38 |
| adam_g | Shrews, ah, ok. what repo does that live in? i haven't looked at it yet | 20:38 |
| Shrews | adam_g: devstack-gate | 20:38 |
| adam_g | Shrews, cool | 20:38 |
| openstackgerrit | A change was merged to stackforge/pyghmi: Reduce severity of generic discrete assert to 'Ok' https://review.openstack.org/107164 | 20:43 |
| jbjohnso | just had to kick jenkins a bit | 20:45 |
| *** jistr has quit IRC | 20:55 | |
| Shrews | adam_g: got a sec? | 20:56 |
| adam_g | Shrews, ya | 20:57 |
| Shrews | adam_g: https://review.openstack.org/#/c/105278/2/tempest/api/compute/admin/test_hypervisor.py | 20:57 |
| Shrews | adam_g: Ken'ichi's comment | 20:58 |
| Shrews | adam_g: do you know how to check the hypervisor type? | 20:58 |
| adam_g | Shrews, lemme see | 20:58 |
| Shrews | _list_hypervisors() only returns the ID and hypervisor host UUID, afaict | 20:59 |
| adam_g | Shrews, you should be able to call a 'show hypervisor' on each id and check hypervsior_type | 21:00 |
| jbjohnso | I hear rumor of people saying ipminative is not working right? | 21:00 |
| adam_g | Shrews, not sure if thats what the test is doing | 21:00 |
| *** Penick has quit IRC | 21:01 | |
| Shrews | adam_g: ah, yes. thx! | 21:01 |
| adam_g | np | 21:01 |
| *** Penick has joined #openstack-ironic | 21:02 | |
| *** jdob has quit IRC | 21:04 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Add driver-properties command https://review.openstack.org/76338 | 21:06 |
| *** Mikhail_D_wk1 has quit IRC | 21:07 | |
| *** stevebaker has quit IRC | 21:07 | |
| *** dguerri has quit IRC | 21:07 | |
| *** soren has quit IRC | 21:07 | |
| *** stevebaker has joined #openstack-ironic | 21:07 | |
| *** soren has joined #openstack-ironic | 21:07 | |
| *** dguerri has joined #openstack-ironic | 21:07 | |
| *** Mikhail_D_wk has joined #openstack-ironic | 21:08 | |
| *** amitpp has quit IRC | 21:13 | |
| *** igordcard has quit IRC | 21:16 | |
| *** jbjohnso has quit IRC | 21:20 | |
| NobodyCam | seeking other reviewer comments on https://review.openstack.org/#/c/103685 | 21:24 |
| jroll | NobodyCam: I still think validate() should take a method name, e.g. 'tear_down'. eventually we're going to want to validate something for tear_down | 21:25 |
| *** romcheg1 has quit IRC | 21:28 | |
| *** jgrimm has quit IRC | 21:31 | |
| NobodyCam | jroll: that sounds like it wold need a spec | 21:43 |
| *** mrda-away is now known as mrda | 21:46 | |
| mrda | Morning Ironic! | 21:46 |
| jroll | NobodyCam: I know... | 21:47 |
| jroll | NobodyCam: not my patch yo :) | 21:47 |
| NobodyCam | good morning mrda | 21:53 |
| mrda | NobodyCam: \o | 21:54 |
| Shrews | hi mrda | 21:54 |
| mrda | \o | 21:54 |
| devananda | mornin, mrda! | 22:01 |
| mrda | hey deva | 22:01 |
| NobodyCam | lol guess I should read the status message | 22:04 |
| NobodyCam | :-p | 22:04 |
| openstackgerrit | A change was merged to openstack/ironic: Fix wrong test fixture for Node.properties https://review.openstack.org/107031 | 22:04 |
| *** Penick has quit IRC | 22:05 | |
| *** igordcard has joined #openstack-ironic | 22:09 | |
| *** Penick has joined #openstack-ironic | 22:15 | |
| *** mdorman has quit IRC | 22:16 | |
| *** Penick has quit IRC | 22:16 | |
| Shrews | mrda: your metaclass change makes my head hurt | 22:20 |
| Shrews | :) | 22:20 |
| Shrews | actually, metaclasses in general make my head hurt | 22:20 |
| NobodyCam | Shrews: so would a doc string on a meta class be meta meta data? | 22:27 |
| Shrews | NobodyCam: DIAF | 22:27 |
| Shrews | :) | 22:28 |
| NobodyCam | lol | 22:28 |
| Shrews | ok. dinner time. see you all tomorrow | 22:29 |
| mrda | Shrews: Sorry about that | 22:34 |
| mrda | Shrews: devananda made me do the __metaclass__ thing. We only got there with lots of experimentation. | 22:35 |
| jroll | "made me"... | 22:35 |
| jroll | I knew everybody was scared of him | 22:35 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Use auth_token from keystonemiddleware https://review.openstack.org/107197 | 22:36 |
| mrda | jroll: well, when the ptl suggests an implementation, it makes sense to follow that direction ;) | 22:37 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Use auth_token from keystonemiddleware https://review.openstack.org/107197 | 22:40 |
| NobodyCam | jroll: looking at 100364... perfect candidate for lucas's new ManagementInterface | 22:41 |
| jroll | gahhhhhhhhhhhh | 22:42 |
| NobodyCam | lol | 22:42 |
| devananda | yanno i like it when ya'll prove me wrong ;) | 22:42 |
| jroll | oh that did land | 22:42 |
| * NobodyCam thinks it will land b4 the ManagementInterface spec | 22:42 | |
| NobodyCam | jroll: nope | 22:43 |
| NobodyCam | noot yet | 22:43 |
| jroll | huh? | 22:43 |
| jroll | NobodyCam: https://github.com/openstack/ironic/blob/master/ironic/drivers/base.py#L357 | 22:43 |
| devananda | mrda: Shrews: sorry. i'm not that fond of __metaclass__ either -- but it was the the way that I could think of to guard a class-object. | 22:43 |
| devananda | i'm sure there are other ways to achieve the same result | 22:43 |
| NobodyCam | jroll: https://review.openstack.org/#/c/100218 | 22:44 |
| devananda | and I do think it reads well | 22:44 |
| NobodyCam | hummm | 22:44 |
| jroll | NobodyCam: I know that spec hasn't landed, but the implementation has, so | 22:44 |
| jroll | I think the implementation was proposed before ironic-specs existed | 22:44 |
| NobodyCam | devananda: looks like we should land https://review.openstack.org/#/c/100218 | 22:45 |
| mrda | devananda and Shrews: There's six.add_metaclass that I'll use instead | 22:45 |
| devananda | mrda: ++ | 22:45 |
| devananda | oh. /me pokes the spec | 22:46 |
| jroll | NobodyCam: if I put it in the ManagementInterface... I'll also need a spec to add it to the interface | 22:46 |
| * jroll cries | 22:46 | |
| NobodyCam | you have a bug | 22:46 |
| NobodyCam | :-p | 22:46 |
| * NobodyCam ducks | 22:46 | |
| *** igordcard has quit IRC | 22:47 | |
| jroll | lol | 22:47 |
| jroll | right | 22:47 |
| jroll | now I'm going to need a spec | 22:47 |
| JayF | Write a spec == "this isn't landing for a long time" | 22:47 |
| JayF | :( | 22:47 |
| NobodyCam | JayF: we have been much more active on the specs of late | 22:47 |
| *** derekh_ has quit IRC | 22:51 | |
| *** igordcard has joined #openstack-ironic | 23:00 | |
| devananda | so | 23:07 |
| devananda | this came up in the project meeting today | 23:07 |
| devananda | specs are pretty heavy handed | 23:07 |
| devananda | IMO they've been super helpful in the design process | 23:07 |
| devananda | but not a great tool to indicate what our plans are, with any sort of advanced notice | 23:08 |
| devananda | especially for smallish changes, the spec and code are often written together (or code-before-spec even) | 23:08 |
| JayF | They also, I don't think, get reviewed often enough. That's basically a problem with everything though, it's just now there's 2x as many reviews for a given feature | 23:08 |
| devananda | JayF: so, hopefully, if a spec is thoroughly reviewed and then approved BEFORE the code, | 23:08 |
| devananda | then the folks who reviewed the spec will have a much easier time revieing the code | 23:09 |
| devananda | and it should be approved faster | 23:09 |
| jroll | JayF: that may also just be an ironic problem (not to say it's not a problem) | 23:09 |
| JayF | Do we have statistics showing a lower average review time for code that had a spec written vs not? | 23:09 |
| JayF | I've heard that presented before and I'm not 100% convinced it's true, honestly | 23:09 |
| devananda | we haev statistics showing lower average # of revisions since we adopted specs | 23:09 |
| devananda | but correlation != causation | 23:09 |
| * devananda points to his recent email with review stats | 23:10 | |
| JayF | I've specifically made a point to review specs with my 'ironic review time' because I think I add more value there than on the code itself at times | 23:10 |
| JayF | and there's just a lot of specs that feel a little stuck in the mud | 23:10 |
| JayF | for lack of a more exact term | 23:10 |
| JayF | I think it's hard to say 'no' on a spec, in a way? | 23:11 |
| openstackgerrit | A change was merged to openstack/ironic-specs: New driver ManagementInterface https://review.openstack.org/100218 | 23:12 |
| NobodyCam | jroll: ^^^ :-p | 23:14 |
| NobodyCam | woo hoo | 23:14 |
| jroll | LOL | 23:14 |
| jroll | NobodyCam: can we just... land 100364 | 23:14 |
| jroll | I know the review number by heart | 23:14 |
| NobodyCam | lol | 23:14 |
| jroll | that's how annoying that review has been | 23:14 |
| NobodyCam | :( | 23:14 |
| devananda | JayF: why is saying "no" hard? | 23:15 |
| devananda | also, i need to step afk. yay meetings | 23:15 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Use auth_token from keystonemiddleware https://review.openstack.org/107197 | 23:15 |
| devananda | and that ^ isn't working yet. but we will need to fix it soonish, as keystone_client.middleware is deprecated | 23:16 |
| devananda | bbiab | 23:16 |
| JayF | devananda: maybe it's just something for me-personally, or maybe generally about the scope of Ironic being unclear, but there's definately specs I've seen that I've wondered if they belonged in Ironic, but wasn't sure/confident enough to -1 it | 23:16 |
| *** harlowja is now known as harlowja_away | 23:16 | |
| devananda | JayF: please don't hesitate to raise such questions | 23:17 |
| devananda | on the spec, in the weekly meeting, anywhere | 23:17 |
| devananda | i rely on all of you to voice your opinions/concerns/etc | 23:18 |
| devananda | otherwise I dont know them :) | 23:18 |
| Shrews | i'm concerned that i don't have cake for my after dinner dessert | 23:18 |
| devananda | Shrews: that's horrible!!!! | 23:18 |
| JayF | No cake? Let Shrews eat bread. | 23:18 |
| Shrews | devananda: plz fix that. kthxbye | 23:18 |
| openstackgerrit | Michael Davies proposed a change to openstack/ironic: Ironic nova driver to cache ironic client calls https://review.openstack.org/102695 | 23:18 |
| devananda | Shrews: I will mail you a cake. as soon as I finish eating this one. | 23:18 |
| Shrews | ewwwww | 23:19 |
| devananda | another one | 23:19 |
| devananda | i hope you like vegan gluten-free cakes | 23:19 |
| Shrews | again, ewwwww | 23:19 |
| devananda | :) | 23:19 |
| mrda | It's my son's 13th birthday today - I will be eating cake | 23:19 |
| devananda | fine, more cake for me | 23:19 |
| *** Penick has joined #openstack-ironic | 23:20 | |
| Shrews | mrda: how does that help me and my cake deficiency? | 23:20 |
| * Shrews creates a spec for more cake | 23:20 | |
| NobodyCam | Shrews: http://sam557.deviantart.com/art/you-ve-got-virtual-cake-174768995 | 23:21 |
| Shrews | woot! | 23:21 |
| mrda | It doesn't, but it helps me with mine :) | 23:21 |
| * JayF tries to remember to compel his wife to make gluten-free goodies for devananda at the mid-cycle | 23:24 | |
| * NobodyCam steps afk for a few | 23:25 | |
| *** rloo has quit IRC | 23:26 | |
| NobodyCam | devananda: did you see I had two ironic meetup tickets? one for ironic and one for nova too | 23:26 |
| Shrews | NobodyCam: we need meetup tickets for the nova one? | 23:27 |
| mrda | Shrews: I don't think so | 23:27 |
| NobodyCam | no I got that b4 we had ironic tickets | 23:27 |
| Shrews | oh wait. nm. i'm thinking tripleo | 23:28 |
| NobodyCam | thats next weeke | 23:28 |
| NobodyCam | :) | 23:28 |
| Shrews | yeah, that's the only one i can attend | 23:28 |
| mrda | So NobodyCam, I think you should let mikal know this. Then he can give your nova seat to a nova devel - they're at capacity in their room | 23:28 |
| NobodyCam | oh and my flight out is also the last day of "lavander" festival (http://www.lavenderfestival.com) | 23:29 |
| mrda | ...because I'm assuming you'll be hanging with the rest of us ironics instead :) | 23:29 |
| NobodyCam | mikal: please note I have both a ironic and nova meetup ticket (or think I do) I do not actually need the nova one! please give to a nova dev!!! | 23:30 |
| NobodyCam | mrda: that work? | 23:30 |
| mikal | NobodyCam: thanks! | 23:31 |
| mikal | NobodyCam: I shall auction it to the lowest bidder | 23:31 |
| JayF | I think we'd also let a nova dev or two in the ironic room | 23:31 |
| JayF | then lock them in until the driver gets merged | 23:31 |
| JayF | that frees up a spot or two in the nova room :P | 23:31 |
| mikal | Ummm, not to be mean, but you'd have to _propose_ the driver first | 23:31 |
| mikal | Just sayin' | 23:31 |
| JayF | not me, personally | 23:32 |
| JayF | I'm further up the chain | 23:32 |
| JayF | we have a driver for /ironic/ which is waiting to be merged which depends on the ironic driver for /nova/ which is yet to be merged | 23:33 |
| JayF | so basically we're running pre-pre-pre-alpha software. or something like that. | 23:33 |
| JayF | :P | 23:33 |
| lifeless | JayF: why is your ironic driver blocked on the nova ironic driver? | 23:34 |
| lifeless | I thought we had/were removing all the ironic-driver data leaks from the nova driver. | 23:34 |
| JayF | lifeless: I'm in like 99% joking mode here :) | 23:35 |
| mrda | NobodyCam: sure | 23:35 |
| NobodyCam | mikal: would you like me to cancel the ticket? it's under my hp email. | 23:35 |
| JayF | lifeless: although honestly, I think our nova-driver for onmetal is patched compared to upstream, but idk what the diff is at this point | 23:35 |
| JayF | (jroll or comstud would, though) | 23:36 |
| jroll | oh god wat | 23:36 |
| jroll | wait | 23:36 |
| jroll | we're not blocked on nova driver | 23:36 |
| jroll | lifeless: ^ | 23:36 |
| jroll | well, kind of. "onmetal production" has changes to the nova driver. for configdrive, etc. | 23:37 |
| jroll | but the agent driver patch is not blocked on that... the only change for the agent driver patch is to add `or 'agent' in node.driver` here: https://github.com/openstack/ironic/blob/master/ironic/nova/virt/ironic/patcher.py#L39 | 23:38 |
| *** igordcard has quit IRC | 23:38 | |
| jroll | mikal: the driver is proposed? | 23:39 |
| mikal | NobodyCam: hold off on that... I think I need to queue it so someone on the wait list gets it when its released | 23:39 |
| NobodyCam | mikal: ack... pm the # over to ya use / abuse as you see fit | 23:39 |
| mikal | NobodyCam: could you reply to my eventbrite email and let me know that you can release it though so I remember? | 23:39 |
| jroll | mikal: https://review.openstack.org/#/q/status:open+project:openstack/nova+branch:master+topic:ironic,n,z | 23:40 |
| mikal | jroll: it needs to be split into phases, as discussed at the summit with deva | 23:41 |
| mikal | devananda is working on it I am sure, its just not done yet | 23:41 |
| jroll | mikal: right, it was split a week or two ago, idk what happened since | 23:41 |
| jroll | right | 23:41 |
| mikal | Oh, I get you | 23:42 |
| mikal | So there's three in the chain? | 23:42 |
| * mrda is glad we're all on the sam page again ;) | 23:42 | |
| mikal | Is that chain complete? | 23:42 |
| jroll | no idea, ask deva :P | 23:42 |
| jroll | rather, ask nova-core | 23:42 |
| mikal | Well, I was goign to work on this other blueprint | 23:42 |
| mikal | Its called "have a nap" | 23:42 |
| jroll | because I don't think deva would mind landing that way | 23:43 |
| jroll | +9000 for "have a nap" | 23:43 |
| mrda | mikal: you only just got out of bed. Back to work! :) | 23:43 |
| * mrda knows his timezones | 23:43 | |
| jroll | heh | 23:44 |
| mikal | mrda: yeah, its the two work calls in the middle of the night at the 5:45am meeting which are killing me | 23:44 |
| jroll | and I thought being 2 hours off of san antonio was bad... | 23:44 |
| * mrda withdraws his mikal cheekyness | 23:44 | |
| *** hemna has quit IRC | 23:44 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add versioning to Agent decommission https://review.openstack.org/106859 | 23:46 |
| * devananda returns from meetinglandia | 23:58 | |
| *** Penick has quit IRC | 23:58 | |
| devananda | mikal: i proposed the nova driver in three phases like weeks ago | 23:59 |
| mikal | LIES | 23:59 |
| mikal | Also, perhaps | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!