*** baoli has joined #openstack-ironic | 00:00 | |
*** bvandewa has quit IRC | 00:01 | |
jroll | devananda: JayF: sambetts|afk: lucas-dinner: heads up, the giant project-config refactor is landing shortly, we'll need to make sure our CI is doing what we expect tomorrow https://review.openstack.org/#/c/340567 | 00:04 |
---|---|---|
*** yuanying has quit IRC | 00:05 | |
*** yuanying has joined #openstack-ironic | 00:06 | |
*** baoli has quit IRC | 00:11 | |
*** hoangcx2 has quit IRC | 00:11 | |
*** aNupoisc has quit IRC | 00:12 | |
*** baoli has joined #openstack-ironic | 00:13 | |
*** bvandewa has joined #openstack-ironic | 00:14 | |
*** Nisha_away has quit IRC | 00:17 | |
*** yuanying has quit IRC | 00:19 | |
*** yuanying has joined #openstack-ironic | 00:20 | |
*** rama_y has joined #openstack-ironic | 00:21 | |
*** ahqa has quit IRC | 00:23 | |
*** baoli has quit IRC | 00:23 | |
*** bcornec1 has quit IRC | 00:31 | |
*** Goneri has quit IRC | 00:34 | |
*** rama_y has quit IRC | 00:35 | |
*** hoangcx has joined #openstack-ironic | 00:38 | |
*** sdake_ has quit IRC | 00:45 | |
*** sdake has joined #openstack-ironic | 00:47 | |
*** mtanin___ has quit IRC | 00:48 | |
*** Sukhdev has quit IRC | 00:52 | |
*** hoangcx2 has joined #openstack-ironic | 00:54 | |
*** yuanying has quit IRC | 00:58 | |
*** jrist has quit IRC | 00:59 | |
*** hoangcx2 has quit IRC | 01:01 | |
*** yuanying has joined #openstack-ironic | 01:02 | |
*** hoangcx2 has joined #openstack-ironic | 01:02 | |
*** sdake has quit IRC | 01:04 | |
*** jrist has joined #openstack-ironic | 01:11 | |
*** Goneri has joined #openstack-ironic | 01:13 | |
*** ijw has joined #openstack-ironic | 01:14 | |
*** hoangcx2 has quit IRC | 01:23 | |
*** hoangcx2 has joined #openstack-ironic | 01:24 | |
*** ijw has quit IRC | 01:25 | |
*** wajdi has joined #openstack-ironic | 01:29 | |
*** openstackgerrit has quit IRC | 01:33 | |
*** openstackgerrit has joined #openstack-ironic | 01:34 | |
*** hoangcx2 has quit IRC | 01:34 | |
*** yibo has quit IRC | 01:35 | |
*** hoangcx2 has joined #openstack-ironic | 01:36 | |
*** yibo has joined #openstack-ironic | 01:36 | |
*** baoli has joined #openstack-ironic | 01:41 | |
openstackgerrit | Naohiro Tamura proposed openstack/ironic: Overrides get_supported_power_states() for WOL Power https://review.openstack.org/293293 | 01:44 |
*** hoangcx2 has quit IRC | 01:51 | |
openstackgerrit | Merged openstack/ironic-python-agent: Fix resolv.conf in tinyipa image build https://review.openstack.org/346963 | 01:53 |
*** lintan has quit IRC | 01:57 | |
*** lintan has joined #openstack-ironic | 01:58 | |
*** bvandewa has quit IRC | 01:58 | |
*** harshs has joined #openstack-ironic | 02:04 | |
*** joprovost has quit IRC | 02:05 | |
*** spandhe has joined #openstack-ironic | 02:10 | |
*** sdake has joined #openstack-ironic | 02:12 | |
*** spandhe has quit IRC | 02:13 | |
*** baoli has quit IRC | 02:17 | |
*** baoli has joined #openstack-ironic | 02:32 | |
*** Goneri has quit IRC | 02:39 | |
*** harshs has quit IRC | 02:45 | |
*** baoli has quit IRC | 02:52 | |
*** sdake has quit IRC | 02:53 | |
*** yuanying has quit IRC | 02:54 | |
*** Lizhixin has joined #openstack-ironic | 02:54 | |
*** Lizhixin has left #openstack-ironic | 02:54 | |
*** harshs has joined #openstack-ironic | 02:55 | |
openstackgerrit | Naohiro Tamura proposed openstack/ironic: iRMC power driver for soft reboot and soft power off https://review.openstack.org/216743 | 03:04 |
*** baoli has joined #openstack-ironic | 03:10 | |
*** sdake has joined #openstack-ironic | 03:13 | |
*** baoli has quit IRC | 03:17 | |
*** rloo has quit IRC | 03:27 | |
*** lujinluo has quit IRC | 03:28 | |
openstackgerrit | Zhenguo Niu proposed openstack/ironic-inspector: Add API for listing all introspection statuses https://review.openstack.org/344921 | 03:39 |
*** Nisha_away has joined #openstack-ironic | 03:49 | |
*** yuanying has joined #openstack-ironic | 03:52 | |
*** harlowja_at_home has joined #openstack-ironic | 03:52 | |
*** links has joined #openstack-ironic | 03:54 | |
*** lujinluo has joined #openstack-ironic | 03:58 | |
*** harshs has quit IRC | 03:58 | |
*** harshs has joined #openstack-ironic | 04:07 | |
*** links has quit IRC | 04:09 | |
*** links has joined #openstack-ironic | 04:12 | |
*** harshs has quit IRC | 04:16 | |
openstackgerrit | Merged openstack/ironic-lib: Add parse_root_device_hints to utils.py https://review.openstack.org/346068 | 04:16 |
*** harlowja_at_home has quit IRC | 04:22 | |
*** bks has joined #openstack-ironic | 04:31 | |
*** vishwanathj has quit IRC | 04:32 | |
*** jtomasek has quit IRC | 04:41 | |
*** adu has joined #openstack-ironic | 04:48 | |
*** harshs has joined #openstack-ironic | 05:01 | |
*** moshele has joined #openstack-ironic | 05:08 | |
*** moshele has quit IRC | 05:19 | |
*** harshs has quit IRC | 05:22 | |
*** bks has quit IRC | 05:29 | |
*** deray has joined #openstack-ironic | 05:31 | |
*** ChubYann has quit IRC | 05:31 | |
deray | hello Ironicers !! long time no see .. :) | 05:33 |
*** Nisha_away has quit IRC | 05:35 | |
*** adu has quit IRC | 05:35 | |
*** spandhe has joined #openstack-ironic | 05:45 | |
*** edand has joined #openstack-ironic | 05:55 | |
*** edand has quit IRC | 06:09 | |
*** Nisha_away has joined #openstack-ironic | 06:11 | |
*** sdake has quit IRC | 06:16 | |
*** moshele has joined #openstack-ironic | 06:25 | |
*** spandhe has quit IRC | 06:26 | |
*** amitkqed has joined #openstack-ironic | 06:35 | |
*** amitkqed has quit IRC | 06:36 | |
*** amitkqed has joined #openstack-ironic | 06:37 | |
*** tesseract- has joined #openstack-ironic | 06:44 | |
*** sdake has joined #openstack-ironic | 06:45 | |
*** rcernin has joined #openstack-ironic | 06:47 | |
*** cppforlife_ has quit IRC | 06:50 | |
*** amit213 has quit IRC | 06:50 | |
*** ildikov has quit IRC | 06:50 | |
*** amit213 has joined #openstack-ironic | 06:51 | |
*** ildikov has joined #openstack-ironic | 06:51 | |
*** cppforlife_ has joined #openstack-ironic | 06:52 | |
*** mjura has joined #openstack-ironic | 06:58 | |
*** sdake has quit IRC | 07:01 | |
*** Nisha_away has quit IRC | 07:01 | |
openstackgerrit | Zhenguo Niu proposed openstack/ironic-inspector: Add API for listing all introspection statuses https://review.openstack.org/344921 | 07:01 |
*** lujinluo has quit IRC | 07:04 | |
*** bvandewa has joined #openstack-ironic | 07:07 | |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed openstack/ironic: [WIP] Testing latest u-c https://review.openstack.org/318440 | 07:07 |
*** livelace has joined #openstack-ironic | 07:09 | |
*** bvandewa has quit IRC | 07:15 | |
*** rbartal has joined #openstack-ironic | 07:17 | |
*** amotoki has joined #openstack-ironic | 07:19 | |
*** e0ne has joined #openstack-ironic | 07:20 | |
*** vdrok has quit IRC | 07:22 | |
*** vdrok has joined #openstack-ironic | 07:23 | |
*** lujinluo has joined #openstack-ironic | 07:29 | |
*** jpich has joined #openstack-ironic | 07:33 | |
*** adreznec has quit IRC | 07:35 | |
*** mtreinish has quit IRC | 07:36 | |
*** mtreinish has joined #openstack-ironic | 07:36 | |
*** linuxgeek has quit IRC | 07:37 | |
*** linuxgeek has joined #openstack-ironic | 07:38 | |
*** adreznec has joined #openstack-ironic | 07:39 | |
*** lennyb has quit IRC | 07:40 | |
*** lennyb has joined #openstack-ironic | 07:41 | |
*** ifarkas_afk is now known as ifarkas | 07:41 | |
*** e0ne has quit IRC | 07:48 | |
*** jtomasek has joined #openstack-ironic | 07:50 | |
*** milan has joined #openstack-ironic | 07:51 | |
vdrok | morning ironic! | 07:51 |
*** stendulker has joined #openstack-ironic | 07:53 | |
milan | morning vdrok! :) | 07:55 |
vdrok | good morning milan | 07:55 |
milan | and morning Ironic! :) | 07:55 |
*** zzzeek has quit IRC | 08:00 | |
*** bvandewa has joined #openstack-ironic | 08:00 | |
*** zzzeek has joined #openstack-ironic | 08:00 | |
openstackgerrit | Zhenguo Niu proposed openstack/ironic-python-agent: Parallel erase disk devices https://review.openstack.org/346723 | 08:02 |
milan | ifarkas: good morning :) inspector's smoke-test got completely green overnight, would you mind a review ;) https://review.openstack.org/#/c/336532/14 | 08:03 |
*** derekh has joined #openstack-ironic | 08:03 | |
ifarkas | milan, morning, sure thing ;-) | 08:03 |
ifarkas | morning all | 08:03 |
milan | ifarkas: thanks a lot! | 08:03 |
*** jmiu has quit IRC | 08:08 | |
*** phschwartz has quit IRC | 08:08 | |
*** phschwartz has joined #openstack-ironic | 08:08 | |
openstackgerrit | Davanum Srinivas (dims) proposed openstack/ironic: [WIP] Testing latest u-c https://review.openstack.org/318440 | 08:10 |
*** jmiu_ has joined #openstack-ironic | 08:10 | |
*** hoonetorg has quit IRC | 08:13 | |
*** lujinluo has quit IRC | 08:14 | |
*** hoonetorg has joined #openstack-ironic | 08:17 | |
*** lujinluo has joined #openstack-ironic | 08:21 | |
*** hshiina has joined #openstack-ironic | 08:22 | |
*** lucas-dinner is now known as lucasagomes | 08:24 | |
milan | morning lucasagomes :) | 08:28 |
lucasagomes | milan, morning | 08:28 |
*** Romanenko_K has joined #openstack-ironic | 08:30 | |
*** vmud213 has joined #openstack-ironic | 08:34 | |
vdrok | good morning ifarkas and lucasagomes | 08:35 |
lucasagomes | hi there :-) | 08:37 |
*** e0ne has joined #openstack-ironic | 08:40 | |
*** priteau has joined #openstack-ironic | 08:43 | |
vmud213 | morning all | 08:46 |
milan | morning vmud213 :) | 08:46 |
vmud213 | morning milan :) | 08:47 |
ifarkas | lucasagomes, hi, could you please review the drac raid patch? https://review.openstack.org/#/c/176319/ | 08:48 |
ifarkas | you probably going to hate me for its size :-) | 08:48 |
lucasagomes | ifarkas, ++! | 08:48 |
lucasagomes | happens :-) | 08:48 |
ifarkas | lucasagomes, awesome, thank you! | 08:48 |
vmud213 | morning lucasagomes, mgould|afk, vdrok and everyone | 08:48 |
vmud213 | greghaynes: Are you here? | 08:49 |
ifarkas | morning vdrok and vmud213 :-) | 08:49 |
vdrok | good morning vmud213 | 08:49 |
vmud213 | ifarkas, good morning | 08:49 |
deray | g'morning milan lucasagomes vdrok ifarkas vmud213 and all others | 08:52 |
*** yibo has quit IRC | 08:52 | |
vmud213 | deray o/ | 08:52 |
deray | vmud213, :) | 08:53 |
vdrok | hi deray :) | 08:53 |
*** Goneri has joined #openstack-ironic | 08:55 | |
deray | vdrok, hi :) | 08:55 |
*** bks has joined #openstack-ironic | 08:56 | |
deray | may be a low hanging fruit for all of you.. https://review.openstack.org/#/c/346683 please don't go by the size.. had to do some refactoring | 08:56 |
*** priteau has quit IRC | 08:57 | |
*** hshiina has quit IRC | 08:57 | |
deray | i have started freshly focusing on https://review.openstack.org/#/c/294995/ and related stuff. was off from ironic for some time :-) | 08:58 |
deray | any code patch submitted for ``boot from cinder`` stuff? | 08:59 |
*** bvandewa has quit IRC | 09:01 | |
openstackgerrit | Tang Chen proposed openstack/ironic: Add python 3.x classifier https://review.openstack.org/347705 | 09:06 |
*** fragatina has joined #openstack-ironic | 09:11 | |
*** fragatina has quit IRC | 09:11 | |
*** fragatina has joined #openstack-ironic | 09:12 | |
lucasagomes | ifarkas, reviewed, +2'd :-) | 09:13 |
ifarkas | lucasagomes, great, thank you! will fix the comments! | 09:15 |
*** fragatina has quit IRC | 09:17 | |
*** Nisha_away has joined #openstack-ironic | 09:18 | |
lucasagomes | ifarkas, no worries about them | 09:19 |
lucasagomes | most are just comments/observations | 09:19 |
sambetts|afk | jroll, JayF: eeeee, /me scared he's broken everything XD | 09:22 |
*** sambetts|afk is now known as sambetts | 09:22 | |
*** fragatina has joined #openstack-ironic | 09:26 | |
sambetts | lucasagomes: any idea what happened to fail here? http://logs.openstack.org/58/272658/71/check/gate-tempest-dsvm-ironic-ipa-partition-pxe_ipmitool-tinyipa-nv/68cc2ae/logs/ in ir-cond it tries to call IPMItool and fails, its running with virtualbmc | 09:27 |
sambetts | but I don't see a failure in the virtualbmc logs | 09:28 |
lucasagomes | sambetts, looking | 09:28 |
*** mgould|afk is now known as mgould | 09:29 | |
mgould | morning vdrok milan ifarkas lucasagomes vmud213 deray sambetts and everyone else! | 09:29 |
lucasagomes | sambetts, hmmm looks like the pygmi Bmc didn't handle the request properly | 09:29 |
milan | morning mgould :) | 09:29 |
sambetts | lucasagomes: ah :/ | 09:30 |
openstackgerrit | Merged openstack/python-ironic-inspector-client: Add Python 3.5 tox env and detailed setup.cfg classifiers https://review.openstack.org/344355 | 09:30 |
lucasagomes | sambetts, will look | 09:30 |
milan | and morning sambetts :) | 09:30 |
milan | sambetts: would you mind checking https://review.openstack.org/#/c/336532/14? Inspector's smoke test ;) got some + es already | 09:31 |
*** fragatina has quit IRC | 09:31 | |
sambetts | milan: sure I'll take a look in a minute :) | 09:31 |
milan | sambetts: cool, thanks | 09:32 |
*** dansmith has quit IRC | 09:33 | |
* milan relocates | 09:33 | |
*** dansmith has joined #openstack-ironic | 09:35 | |
*** dansmith is now known as Guest41069 | 09:35 | |
*** milan has quit IRC | 09:36 | |
*** lujinluo has quit IRC | 09:38 | |
*** yibo has joined #openstack-ironic | 09:39 | |
*** ccamacho has joined #openstack-ironic | 09:44 | |
*** electrofelix has joined #openstack-ironic | 09:46 | |
*** mbound has joined #openstack-ironic | 09:46 | |
*** fumie has quit IRC | 09:47 | |
*** tangchen has quit IRC | 09:50 | |
*** tangchen_ has quit IRC | 09:50 | |
*** chlong_POffice has quit IRC | 09:57 | |
*** yibo has quit IRC | 10:00 | |
*** yibo has joined #openstack-ironic | 10:01 | |
*** yibo has quit IRC | 10:02 | |
*** yibo has joined #openstack-ironic | 10:03 | |
*** tangchen_ has joined #openstack-ironic | 10:04 | |
*** tangchen has joined #openstack-ironic | 10:04 | |
*** hwcomcn has joined #openstack-ironic | 10:06 | |
*** chlong_POffice has joined #openstack-ironic | 10:11 | |
*** edand has joined #openstack-ironic | 10:12 | |
*** hoangcx has quit IRC | 10:33 | |
*** mgould is now known as mgould|bbl | 10:37 | |
*** hoangcx has joined #openstack-ironic | 10:41 | |
openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic-python-agent: Add erase_devices_metadata cleaning step https://review.openstack.org/344164 | 10:42 |
openstackgerrit | Merged openstack/ironic-inspector: Add a simple smoke test to be run in the grenade gate https://review.openstack.org/336532 | 10:46 |
*** yuikotakadamori has joined #openstack-ironic | 11:00 | |
*** stendulker has quit IRC | 11:09 | |
*** lucasagomes is now known as lucas-hungry | 11:13 | |
openstackgerrit | Kyrylo Romanenko proposed openstack/python-ironicclient: Fix punctuation for OSC plugin help messages https://review.openstack.org/347765 | 11:18 |
*** hoangcx2 has joined #openstack-ironic | 11:21 | |
*** hoangcx has quit IRC | 11:21 | |
*** milan has joined #openstack-ironic | 11:22 | |
*** deray has quit IRC | 11:39 | |
*** deray has joined #openstack-ironic | 11:41 | |
*** Sukhdev has joined #openstack-ironic | 11:45 | |
openstackgerrit | Aparna proposed openstack/proliantutils: Raise exception when HPSSA controller not available https://review.openstack.org/347791 | 11:48 |
openstackgerrit | Teng Fei proposed openstack/ironic-inspector: Fixes Bug1606534 https://review.openstack.org/347795 | 11:54 |
TheJulia | shameless plug for the oneview team, they would <3 it if people review their dynamic node allocation change to their driver https://review.openstack.org/#/c/286192/ | 11:54 |
*** vmud213 has quit IRC | 11:56 | |
*** vmud213 has joined #openstack-ironic | 11:56 | |
openstackgerrit | Pavlo Shchelokovskyy proposed openstack/ironic: Update devstack plugin with new auth options https://review.openstack.org/297312 | 11:56 |
openstackgerrit | Pavlo Shchelokovskyy proposed openstack/ironic: Migrate to using keystoneauth Sessions https://review.openstack.org/236982 | 11:56 |
*** Nisha_away has quit IRC | 11:58 | |
*** thrash|g0ne is now known as thrash | 12:09 | |
*** vmud213 has quit IRC | 12:12 | |
*** vmud213 has joined #openstack-ironic | 12:12 | |
*** jjohnson2__ has joined #openstack-ironic | 12:15 | |
*** MattMan has quit IRC | 12:17 | |
*** lucas-hungry is now known as lucasagomes | 12:17 | |
*** MattMan has joined #openstack-ironic | 12:17 | |
*** chlong_POffice has quit IRC | 12:21 | |
*** trown|outtypewww is now known as trown | 12:23 | |
*** sdake has joined #openstack-ironic | 12:23 | |
*** hwcomcn has quit IRC | 12:25 | |
xavierr | good morning Ironic | 12:26 |
xavierr | morning and thank you TheJulia :) | 12:27 |
*** yuikotakadamori has quit IRC | 12:27 | |
*** chlong has joined #openstack-ironic | 12:29 | |
*** e0ne has quit IRC | 12:31 | |
jroll | sambetts: heh, does everything look okay? | 12:34 |
openstackgerrit | vinay kumar muddu proposed openstack/proliantutils: Use existing ilo object for each ris/ribcl call https://review.openstack.org/347805 | 12:35 |
openstackgerrit | Merged openstack/ironic: Change comment regarding network_interface https://review.openstack.org/347441 | 12:36 |
jroll | also, morning everyone | 12:39 |
*** vishwanathj has joined #openstack-ironic | 12:42 | |
*** ccamacho is now known as ccamacho|hungry | 12:45 | |
sambetts | jroll: seems to be, I checked the majority of jobs, and they all seem to be behaving as expected | 12:46 |
jroll | sambetts: awesome, thanks for doing that | 12:46 |
*** Nisha_away has joined #openstack-ironic | 12:48 | |
sambetts | jroll: there are a few more jobs I think we need to standardise the names with the new form, e.g. the postgres one etc but that can easily done in a follow up | 12:48 |
jroll | sambetts: right on, clark mentioned something about reducing the number of jobs, I want to talk to him more about his thoughts there | 12:49 |
jroll | (it is something I think we can do) | 12:49 |
*** baoli has joined #openstack-ironic | 12:49 | |
sambetts | number of generated jobs or number of jobs that are actually being run? | 12:50 |
*** baoli_ has joined #openstack-ironic | 12:50 | |
jroll | actual jobs | 12:51 |
jroll | sambetts: "its also probably worthwhile talking about whether or not we need to test all these combos with all these tests or maybe a bunch of it can be shoved into a functional type test then cover integration testing with a smaller set of jobs" | 12:51 |
jroll | sambetts: "we should keep iterating on that though because I think we can simplify further. eg ramdisk type, not sure we actually need to test all those combos if we have one that works" | 12:51 |
sambetts | ah right, interesting | 12:52 |
sambetts | I thought we had the stance its unsupported if not tested | 12:52 |
*** wajdi has quit IRC | 12:52 | |
sambetts | so we have to test all the ramdisk types otherwise they are classed as unsupported | 12:52 |
jroll | yeah, it's unclear exactly what he means, I haven't had the opportunity to chat about it | 12:52 |
thiagop | Good morning, Ironicville | 12:52 |
jroll | but if we could have a functional test that builds each type of ramdisk and just runs against the IPA API | 12:53 |
jroll | that might be sufficient | 12:53 |
sambetts | yeah that could work | 12:53 |
jroll | basically a couple things that might depend on the ramdisk build - inspection finding all the things, writing an image, writing configdrive, writing bootloader | 12:53 |
jroll | maybe run shred on a small file to make sure it's present in the ramdisk | 12:54 |
jroll | etc | 12:54 |
sambetts | yeah if we could test those on the ramdisks outside of a full deployment test that would definatly be nicer, and we could run that on IPA itself instead of having mutliple jobs on Ironic | 12:54 |
*** baoli has quit IRC | 12:55 | |
sambetts | technically I guess we don't need a full job for each ramdisk on Ironic now | 12:55 |
sambetts | because we run a full job on IPA commit anyway | 12:55 |
sambetts | for each ramdisk type | 12:55 |
jroll | I don't think we do test each ramdisk against ironic | 12:55 |
sambetts | ah, no, no we don't | 12:57 |
sambetts | we did during the change over to tinyipa but now we don't | 12:57 |
jroll | right, was a transitional thing | 12:57 |
openstackgerrit | Yuriy Zveryanskyy proposed openstack/ironic-specs: Add CRUD notifications to Ironic https://review.openstack.org/347242 | 13:04 |
*** deray_ has joined #openstack-ironic | 13:04 | |
*** links has quit IRC | 13:08 | |
*** deray has quit IRC | 13:08 | |
*** rloo has joined #openstack-ironic | 13:08 | |
*** mgould|bbl is now known as mgould | 13:10 | |
*** vnogin has left #openstack-ironic | 13:10 | |
*** ccamacho|hungry is now known as ccamacho | 13:28 | |
jroll | devananda: thanks for writing up that API stuff, I left some notes and added some more, let's keep iterating :) | 13:30 |
*** david-lyle_ has joined #openstack-ironic | 13:31 | |
*** david-lyle_ has quit IRC | 13:31 | |
*** david-lyle has quit IRC | 13:34 | |
jroll | hey cores, these patches could use some reviews to push the nova scheduling stuff forward https://review.openstack.org/#/c/345055/ https://review.openstack.org/#/c/345080/ | 13:39 |
sambetts | anyone else finding review.openstack.org slow all of a sudden? | 13:41 |
*** ayoung has joined #openstack-ironic | 13:41 | |
*** wajdi has joined #openstack-ironic | 13:42 | |
jroll | yes | 13:43 |
jroll | lol | 13:43 |
rloo | jroll: will look at the resource_class stuff later today. when are you hoping for them to land, or er when do you think the nova folks would like it? | 13:45 |
jroll | rloo: thanks. nova is unopinionated on it, as long as there's a field they can read and pass to the resource tracker | 13:45 |
jroll | oh, timing, uh | 13:46 |
jroll | asap? :) | 13:46 |
rloo | jroll: so they need it in newton, and might want it sooner to test with :) | 13:46 |
jroll | rloo: realistically if you could look at this week, I'd appreciate it, I'd like it to land before end of next week | 13:46 |
rloo | jroll: i was going to say, that yesterday or asap were not options. | 13:46 |
sambetts | So with this addtion will we be able to ditch the need for a custom set of filters (e.g. the exact match etc) and custom Resource Tracker that updates the resources to zero for a node? | 13:47 |
jroll | rloo: feature freeze is in a month, so we need to give them enough time to land code using it before that | 13:47 |
*** harshs has joined #openstack-ironic | 13:47 | |
rloo | jroll: in a month? Yikes. ok, i'm on it... | 13:47 |
jroll | sambetts: in ocata both methods will still work, but ops can switch over at that time. we'll have to do the RT junk throughout ocata (so both are usable) and probably drop it in P | 13:48 |
*** Guest41069 is now known as dansmith | 13:48 | |
jroll | sambetts: that said, nova hasn't said when they'll drop the old way, I assume P but I'm not sure | 13:48 |
sambetts | I'm just trying to workout what having a defined single resource class per node gains us, don't we want it to be more like this node provides this set of resources, including any custom resource tags that a operator might want to put on it? | 13:50 |
*** spandhe has joined #openstack-ironic | 13:50 | |
jroll | rloo: thank you | 13:51 |
jroll | sambetts: we're getting there - the first step is quantitative resources, capabilities/etc will be via filters still | 13:51 |
-openstackstatus- NOTICE: Gerrit is being restarted now to relieve performance degradation | 13:52 | |
jroll | so first we say "this is a foo computer", where foo is known to be a computer with 128gb RAM and 20 CPU cores and a 1TB disk | 13:53 |
sambetts | jroll: why do we need a class per node then? over a single "baremetal" resource class, and then let the normal CPU, RAM and DISK resource scheduling handle the rest? | 13:54 |
jroll | sambetts: because that wouldn't actually do anything, right? it would just be run all nodes through the filter | 13:55 |
jroll | the normal CPU/disk/ram stuff is moving to this model | 13:55 |
jroll | where a "resource provider" is a compute node that provides some amount of CPU/disk/ram | 13:55 |
jroll | so the scheduler says, give me all resource providers that have 1024 RAM available | 13:56 |
jroll | that maps to a compute node | 13:56 |
sambetts | Right, I thought the point of baremetal having a unique resource class is so that it can be identified from VMs not from other BMs | 13:56 |
jroll | nope | 13:56 |
sambetts | because that is done using the other resources | 13:56 |
jroll | other resources? | 13:57 |
*** Fdaisuke has joined #openstack-ironic | 13:57 | |
sambetts | CPU, RAM, DISK etc | 13:57 |
jroll | baremetal resources don't fit into the nova model because they can't be subdivided | 13:57 |
sambetts | so an ironic node provides, exactly 1 baremetal resource, then rest of the scheduling happens using the other resources it provides | 13:58 |
jroll | idk how to describe this well without describing all of the scheduler changes | 13:58 |
sambetts | or are we suggesting in your patches, we only ever schedule on the resource_class and never using RAM/CPU/DISK once we on the new system | 14:00 |
jroll | sambetts: does it make it clearer if I say that the ramfilter etc are going away? | 14:00 |
jroll | yes, exactly | 14:00 |
*** joprovost has joined #openstack-ironic | 14:00 | |
sambetts | As I understand it they are going away, to be replaces by providers providing X RAM etc | 14:01 |
sambetts | or providing X CPUs | 14:01 |
sambetts | or 1 or 0 baremetals | 14:01 |
jroll | right | 14:01 |
jroll | a VM flavor will require e.g. 1 CPU and 1024 RAM | 14:02 |
sambetts | so why do we need unique class per node when the rest of the scheduling will still eventially happen based on the other resources a provider provides | 14:02 |
jroll | a BM flavor will require 1 baremetal-gold or whatever | 14:02 |
*** trown is now known as trown|brb | 14:02 | |
jroll | because baremetal resource providers won't provide RAM/CPU/disk | 14:02 |
sambetts | why? | 14:02 |
jroll | because those are assumed to be divisible | 14:03 |
jroll | and you'd end up with a 2gb flavor landing on a 4gb box, etc | 14:03 |
jroll | and then that box would still be available for scheduling another 2gb instance | 14:03 |
sambetts | no, because you add a "baremetal" resource class which each provider can only have 1 or 0 of | 14:03 |
sambetts | oh I think I see | 14:03 |
sambetts | hmm... | 14:04 |
jroll | hrm, that is an interesting thought | 14:04 |
*** vmud213 has quit IRC | 14:04 | |
jroll | if it required 2gb RAM *and* 1 baremetals | 14:04 |
sambetts | yes ^ that | 14:04 |
jroll | then it wouldn't have any baremetals left | 14:05 |
openstackgerrit | Kyrylo Romanenko proposed openstack/python-ironicclient: Add warning message for baremetal node unset command https://review.openstack.org/347856 | 14:05 |
jroll | you still have the too-much-ram problem | 14:05 |
*** trown|brb is now known as trown | 14:05 | |
sambetts | jroll: yeah, but that solved by the operator using differnt types of filters right? | 14:05 |
jroll | sambetts: like the ram filters that won't exist? :) | 14:06 |
sambetts | is all schduling customisation going away? | 14:06 |
jroll | no | 14:06 |
jroll | but we'd be forcing out of tree filters | 14:06 |
sambetts | ah ok, so we have to assume we can't change the matching any more | 14:06 |
jroll | afaik, anyway, but I know they're moving capabilities and such into this system, so maybe the filters are going away long term | 14:07 |
sambetts | hmm interesting problem that certain resources are defined as devisable | 14:07 |
jroll | all, really, I think | 14:08 |
sambetts | its a shame a resource provider can just define itself a undivisable :/ | 14:08 |
sambetts | as* | 14:08 |
thiagop | TheJulia: Tests are failing due to lack of applied_server_profile_uri on the devstack's node | 14:08 |
sambetts | can'T | 14:08 |
thiagop | TheJulia: I'm sending a patch right now to fix it on the devstack plugin (enroll_nodes function) | 14:08 |
jroll | sambetts: I assume that's a huge amount of complexity | 14:09 |
sambetts | jroll: probably :-P | 14:09 |
jroll | sambetts: you're welcome to reply to that email I sent and propose it and see where it breaks down :) | 14:09 |
jroll | honestly I'm just glad this is happening, and most of the work is being done for us | 14:09 |
*** aswadr_ has joined #openstack-ironic | 14:11 | |
*** harshs has quit IRC | 14:11 | |
TheJulia | thiagop: ugh, since devstack is enrolling the hardware :( | 14:11 |
thiagop | TheJulia: yup, it's saying that oneview is using the server (which is true by condition, which says that our refactoring is more reliable than the previous impl :D) | 14:12 |
*** bks has quit IRC | 14:13 | |
sambetts | jroll: I think the scheduling smaller flavors onto resource providers that are too big is something that I'm not sure we can solve without doing it the way you've suggested or having some custom filter (or whatever is going to replace them) to catch that case, I see the problem now :/ | 14:13 |
openstackgerrit | Thiago Paiva Brito proposed openstack/ironic: Add Dynamic Allocation feature for the OneView drivers https://review.openstack.org/286192 | 14:13 |
thiagop | TheJulia: ^ | 14:13 |
jroll | sambetts: ya, I meant propose the make a resource provider indivisible thing | 14:14 |
*** jcoufal has joined #openstack-ironic | 14:14 | |
thiagop | TheJulia: 30 is kabbalistic :D | 14:14 |
sambetts | jroll: which email do you think I should reply too? the midcycle update one? | 14:15 |
jroll | sambetts: yeah, that was my thought, or a new thing | 14:15 |
*** spandhe has quit IRC | 14:18 | |
lucasagomes | NobodyCam, random question, do you know anything about a be2iscsi driver for linux? Is it maintained by HPE ? (I understand you are not part of HPE anymore, but just checking...) | 14:24 |
*** anush_ has joined #openstack-ironic | 14:26 | |
thiagop | lucasagomes: TheJulia might help | 14:26 |
jroll | lucasagomes: you're still planning on using the specs_matcher thing in oslo.utils, right? just noticed it's released now :) | 14:27 |
lucasagomes | jroll, yes we are! Yesterday a patch refactoring bits of the root device hints to ironic-lib was merged as well | 14:28 |
lucasagomes | so I'm gonna work on that | 14:28 |
jroll | lucasagomes: awesome, I just pushed a patch to bump up the minimum in g-r | 14:28 |
lucasagomes | w00t cool | 14:28 |
jroll | let me know if you need an ironic-lib release | 14:28 |
lucasagomes | good to see this being unblocked | 14:28 |
lucasagomes | jroll, yeah that will help, but lemme start adding the code to ironic and ipa to see if I don't need anything else there | 14:29 |
jroll | lucasagomes: sounds good | 14:29 |
lucasagomes | thiagop, thanks :-) | 14:30 |
moshele | it seem that nova hypervisor-show is broken when hypervisor_type is ironic see https://bugs.launchpad.net/nova/+bug/1606941 I uploaded this fix https://review.openstack.org/#/c/347870/ to nova | 14:31 |
openstack | Launchpad bug 1606941 in OpenStack Compute (nova) "nova hypervisor-show is broken when hypervisor_type is ironic type" [Undecided,New] | 14:31 |
jroll | moshele: that's fun | 14:32 |
jroll | moshele: cpu_info is None, I guess? maybe we should just fix that instead (or does it not make sense to do so)? | 14:33 |
jroll | code comments say it's impossible, but maybe we should return P{ | 14:34 |
jroll | {}, rather | 14:34 |
*** PollyZ has joined #openstack-ironic | 14:34 | |
moshele | jroll: that also an option or we can do both | 14:35 |
jroll | moshele: yep | 14:35 |
* jroll comments | 14:35 | |
*** gabriel-bezerra has joined #openstack-ironic | 14:39 | |
moshele | jroll: thanks for the review will work on your comments :) | 14:39 |
*** siva_krish has joined #openstack-ironic | 14:40 | |
*** xiexs has quit IRC | 14:43 | |
rloo | jroll: qq, do we want to allow users to specify the resource_class when creating a node ('ironic node-create --resource_class ...')? | 14:45 |
lucasagomes | moshele, oh this can break some scripts that I know which uses hypervisor-show to figure out whether the nodes in Ironic have been acknowledged by the nova-scheduler already | 14:45 |
rloo | jroll: yes. i answered my own question. | 14:45 |
sambetts | rloo, jroll: do we have a default for that | 14:45 |
sambetts | ? | 14:45 |
rloo | jroll: forget it. i see you did it. | 14:45 |
rloo | sambetts: it is null. | 14:45 |
rloo | sambetts: did you read the spec? :) | 14:46 |
sambetts | yes, just can't rememeber all the things | 14:46 |
*** michauds has joined #openstack-ironic | 14:46 | |
rloo | sambetts: i can relate to that! | 14:46 |
sambetts | :-p | 14:46 |
rloo | sambetts: if it is null, it apparently should work as it works today (it == nova scheduling) | 14:47 |
*** Romanenko_K has quit IRC | 14:47 | |
sambetts | what do we do once the current schdularing goes away? We'll need a default then right? But I guess thats far enough in the future we don't need to worry about it? | 14:47 |
rloo | sambetts: well, someone should be worrying about it, but that someone isn't me. | 14:48 |
sambetts | rloo: hehe ;) | 14:48 |
sambetts | jroll, rloo: I've sent an email out re: indivisible resource providers | 14:49 |
sambetts | its a hard problem to discribe so I hope it make sense | 14:49 |
rloo | sambetts: btw, i was thinking a little bit about node.network_interface. i think the problem is that we want 1. way for user to specify/override *_interface at node level; 2. way to get the actual *_interface being used by a node | 14:50 |
rloo | sambetts: and node.*_interface would/might lead folks to think that is the interface being used. | 14:50 |
*** rook_ is now known as rook | 14:50 | |
sambetts | rloo: if node.*_interface is set that is the interface that is being used | 14:51 |
rloo | sambetts: so am wondering whether we should have another set of node.<prefix>_*_interface. to indicate that is what is being set/overridden... | 14:51 |
rloo | sambetts: the problem is the value of None... :( | 14:52 |
sambetts | right, which I don't want | 14:52 |
sambetts | I want us to populate the DB on node-create | 14:52 |
sambetts | no Nones | 14:52 |
rloo | sambetts: so how does the user change that later? | 14:52 |
sambetts | node-udpate | 14:52 |
rloo | sambetts: what if they want the default hw-type value? or the global config value? | 14:53 |
sambetts | they can get the default hw-type value from /drivers and set that, and I don't think there should be any global config values | 14:54 |
rloo | sambetts: don't say that. we already added a global config value! | 14:54 |
rloo | sambetts: ok, clearly it needs discussion. will need to wait for dmitry i think. and others. | 14:55 |
sambetts | for 1 weird interface ;) I will fight it much much harder for the other interfaces, also did you see my email a week or so back RE: interface defaults and how I think it should all interact so its standard with the across all *_interfaces | 14:57 |
*** rbartal has quit IRC | 14:57 | |
jroll | sambetts: thanks for the mail | 15:00 |
*** jistr is now known as jistr|call | 15:00 | |
jroll | sambetts: thinking default config could still exist, if the user doesn't pass it on node-create that's what is written | 15:01 |
*** Guest_39743 has joined #openstack-ironic | 15:01 | |
*** jaybeale has joined #openstack-ironic | 15:01 | |
sambetts | jroll: for resource_class? | 15:01 |
Guest_39743 | Allah is doing | 15:01 |
Guest_39743 | sun is not doing Allah is doing | 15:01 |
Guest_39743 | moon is not doing Allah is doing | 15:01 |
Guest_39743 | stars are not doing Allah is doing | 15:01 |
* jroll pokes infra | 15:01 | |
Guest_39743 | planets are not doing Allah is doing | 15:01 |
*** deray_ has quit IRC | 15:02 | |
mgould | how odd | 15:02 |
jroll | sambetts: for network interfaces | 15:02 |
robcresswell | Just kick them | 15:02 |
*** harshs has joined #openstack-ironic | 15:02 | |
Guest_39743 | galaxies are not doing Allah is doing | 15:02 |
jroll | robcresswell: I don't have ops, I bugged fungi | 15:02 |
*** fungi has joined #openstack-ironic | 15:02 | |
Guest_39743 | oceans are not doing Allah is doing | 15:02 |
Guest_39743 | mountains are not doing Allah is doing | 15:02 |
*** ChanServ sets mode: +b Guest_39743!*@* | 15:02 | |
*** Guest_39743 was kicked by ChanServ (Banned: spamming is off-topic) | 15:02 | |
mgould | seems like a very low-conversion-rate approach to evangelism... | 15:02 |
robcresswell | Indeed :) | 15:03 |
jroll | heh | 15:03 |
jroll | ¯\_(ツ)_/¯ | 15:03 |
rloo | i hope allah is not happy with our guest | 15:03 |
sambetts | jroll: I want it to all interact using the enabled_*_interfaces config options | 15:04 |
rloo | sambetts: i saw your email but needed some time to think about it and haven't yet thought about it. | 15:04 |
rloo | sambetts: we should definitely get this hammered out before newton ends. | 15:04 |
jroll | sambetts: I understand what you want, others want the config, but the real issue here is using NULLs to imply defaults | 15:04 |
fungi | jroll: i've added operator permissions for you in chanserv's access list. looks like devananda never passed down his to any successors | 15:04 |
jroll | fungi: indeed. thanks :) | 15:04 |
fungi | don't mention it | 15:04 |
rloo | jroll: i reviewed all the ironic-related resource_class patches. Enjoy :) | 15:05 |
jroll | rloo: thank you! | 15:05 |
rloo | jroll: yw | 15:06 |
*** thrash is now known as thrash|biab | 15:06 | |
sambetts | jroll: whats the status on the node tag stuff? | 15:08 |
rloo | sambetts: there are code patches that need to be reviewed I think | 15:09 |
jroll | sambetts: spec is in, I see people working on code but I haven't reviewed | 15:09 |
rloo | sambetts: https://review.openstack.org/#/q/topic:bug/1526266 | 15:09 |
sambetts | ah, I was just reading rloo comment RE: making resource_class more generic, and then was thinking what if we just added every tag a node has onto its resource type as a boolean resource class | 15:10 |
sambetts | s/resource type/resouce prvider/ | 15:10 |
rloo | sambetts: i asked about using tags. the resource class apparently can take only one value. | 15:10 |
rloo | sambetts: but good for you to pursue it :) | 15:11 |
sambetts | rloo: I'm suggesting we treat each tag as a separate resource class, with a possble value of zero or one, but perhaps I've misunderstood somewhere XD | 15:12 |
rloo | sambetts: but what do you do for a node that has more than one tag? | 15:13 |
rloo | sambetts: ironic-virt driver can only set the node's resource_class value to one thing. | 15:13 |
*** mtanino has joined #openstack-ironic | 15:14 | |
rloo | sambetts: that is my understanding | 15:14 |
sambetts | My understanding was that a resource provider can provide mutiple different resources defined by resource classes, so if a node had multiple tags, the resource provider would provide resources: tagone: 1 tagtwo: 1 tagthree: 1 RAM: 512 | 15:16 |
sambetts | but I might have missed something somewhere | 15:16 |
*** pcaruana has quit IRC | 15:17 | |
*** sabeen has joined #openstack-ironic | 15:18 | |
jroll | what sort of multiple resources are you thinking sambetts | 15:18 |
rloo | sambetts: i think then, for that node, the resource provider would show three resources and say that there are 3 nodes available when there is only one :-( | 15:18 |
NobodyCam | Good Morning Ironicers | 15:19 |
NobodyCam | anyone want to take a look at 325599 (keystone policy support) before I hit it with a +A ? | 15:20 |
sambetts | rloo: I don't think so? a resource provider just provides pools of resources, if you defined a flavor a requiring: tagone AND tagtwo and tagthree then schduling an instance using that flavor onto that resource provider would use up all those resources making unavaible to be scheduled onto again | 15:21 |
rloo | sambetts: i can't comment cuz i don't know the details of how it will work in nova. | 15:21 |
sambetts | jroll: just thinking about instead of having a single resource_class per node, using tags instead | 15:22 |
*** milan has quit IRC | 15:22 | |
jroll | sambetts: the resource_class simply defines what sort of computer it is | 15:22 |
sambetts | oh so resource_class is a specifc thing ? I thought we were adding a pool of resources to our resource provider with max value 1 which we just happened to call resource_class | 15:24 |
*** rcernin has quit IRC | 15:24 | |
jroll | no, each node is a resource_provider | 15:24 |
sambetts | right | 15:24 |
jroll | that provides exactly 1 or 0 of a specific resource_class | 15:25 |
jroll | I see what you mean, they could also provide 1 of a second resource_class | 15:25 |
sambetts | uyes ^ | 15:25 |
jroll | but I'm not sure what that gains? | 15:25 |
sambetts | just that we're introducing whats effectivly a "tag" for a node while also working on adding tags' | 15:26 |
sambetts | just thinking about how it all interacts with the existing ecosystem and planned work is all | 15:28 |
*** mjura has quit IRC | 15:28 | |
jroll | eh, I feel like it's special as it indicates to the placement service what sort of node it is | 15:28 |
jroll | sambetts: keep in mind the placement thing may be it's own service one day, and we'll be putting data there directly | 15:29 |
JayF | s/may/will/ at least from what the nova folks said | 15:29 |
sambetts | yeah :/ | 15:30 |
*** Goneri has quit IRC | 15:30 | |
sambetts | o/ JayF | 15:30 |
JayF | o/ | 15:30 |
JayF | sambetts: I pointed fungi at the TinyIPA-doesnt-have-any-security-in-the-build bug | 15:30 |
*** harshs has quit IRC | 15:31 | |
jroll | sambetts: the other thing is I don't expect tags to be done in the next week or two, and we need to get this done :) | 15:32 |
JayF | sambetts: jroll / devananda and I talked about this field maybe being changed out for a specific kind of tag in the future, but that the simpler path leads us to something working sooner | 15:35 |
fungi | jroll: sambetts: yeah, i think the https performance concerns are probably bunk, but also irrelevant if they can just be convinced to sign their builds with a key we can trust | 15:35 |
*** jistr|call is now known as jistr | 15:35 | |
fungi | https performance, even at high volumes, was a concern for site operators a decade ago. in modern environments tls handling accounts for only a very small slice of the overall work being done by most webservers. but whatevs | 15:36 |
sambetts | jroll: good point | 15:36 |
*** harshs has joined #openstack-ironic | 15:37 | |
*** PollyZ has quit IRC | 15:38 | |
*** ifarkas is now known as ifarkas_away | 15:39 | |
jroll | fungi: totally agree :) | 15:41 |
*** harshs has quit IRC | 15:43 | |
Sukhdev | jroll : wanted to give you an updated - | 15:48 |
Sukhdev | jroll : I updated my test system with latest ironic code - all seems to working fine | 15:49 |
*** harshs has joined #openstack-ironic | 15:49 | |
Sukhdev | jroll : I will kick of some tests and monitor the progress - If I see anything, will let you know - as of now, things are looking pretty good | 15:50 |
Sukhdev | jroll : thought give you good news to start the day right :-) | 15:50 |
jroll | Sukhdev: good to hear, I suspected things would be good given we have CI running on this :) | 15:51 |
*** Nisha_away has quit IRC | 15:52 | |
*** PollyZ has joined #openstack-ironic | 15:52 | |
*** Nisha_away has joined #openstack-ironic | 15:55 | |
*** causten has quit IRC | 15:58 | |
*** causten has joined #openstack-ironic | 15:58 | |
sambetts | Sukhdev, jroll: \o/ | 15:59 |
*** harshs has quit IRC | 16:00 | |
*** e0ne has joined #openstack-ironic | 16:01 | |
*** trown is now known as trown|lunch | 16:02 | |
*** causten has quit IRC | 16:03 | |
*** causten has joined #openstack-ironic | 16:03 | |
*** causten has quit IRC | 16:04 | |
*** causten has joined #openstack-ironic | 16:05 | |
lucasagomes | jroll, I just noted that the spec matcher is missing some operators, e.g it has >= and <= but not < and > | 16:05 |
lucasagomes | for 's' prefixed, it does have s<=, s<, s>= and s> :-/ | 16:05 |
* lucasagomes puts a patch up | 16:05 | |
jroll | lucasagomes: I just ported what was there in nova | 16:06 |
lucasagomes | jroll, yeah | 16:06 |
lucasagomes | maybe we should extend it, will put a patch up | 16:06 |
lucasagomes | we can still use the current ones, it's just that it looks odd to not support > and < | 16:06 |
*** causten has quit IRC | 16:09 | |
*** harshs has joined #openstack-ironic | 16:09 | |
*** causten has joined #openstack-ironic | 16:10 | |
*** thrash|biab is now known as thrash | 16:11 | |
*** causten has quit IRC | 16:13 | |
*** causten has joined #openstack-ironic | 16:13 | |
*** causten_ has joined #openstack-ironic | 16:14 | |
*** causten has quit IRC | 16:18 | |
openstackgerrit | Vladyslav Drok proposed openstack/ironic: Add api-ref for new port fields https://review.openstack.org/325299 | 16:18 |
lucasagomes | JayF, around ? | 16:20 |
JayF | ayup | 16:20 |
lucasagomes | JayF, so i was looking at the code of erase_devices, it seems to convert the exceptions to IncompatibleHardwareMethodError https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/hardware.py#L733-L754 | 16:21 |
JayF | lucasagomes: well, only in some cases | 16:21 |
lucasagomes | JayF, which is handled by dispatch_to_managers() | 16:21 |
lucasagomes | JayF, and the erase_devices() itself does not raise anything https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/hardware.py#L314-L332 | 16:22 |
JayF | lucasagomes: that's the stanza that tries to do the ATA erase; if it can't ATA erase and the option to shred isn't enabled, it throws that, bascially saying "I can't erase this" | 16:22 |
*** jtomasek has quit IRC | 16:22 | |
JayF | lucasagomes: and we don't catch any exceptions in self._shred_block_device, so if that fails, it fails | 16:22 |
lucasagomes | JayF, so I got a bit confused, I thought that, at the end (erase_devices()) all exceptions have been captured and it would only return a dict with the outputs | 16:22 |
*** edand has quit IRC | 16:23 | |
lucasagomes | JayF, right, so it's better to just bubble the exception and not return a dict at all? | 16:23 |
JayF | lucasagomes: yeah, no, basically you end up with either 1) ATA erase didn't work, we raise IncompatibleHardwareMethodError, or 2) Shred didn't work, and whatever exception it raised bubbles up on it's own | 16:23 |
lucasagomes | ack, gotcha | 16:23 |
lucasagomes | yeah I was a bit confused by that code | 16:23 |
JayF | lucasagomes: (and fwiw, since GenericHardwareManager is the lowest prio, IncompatibleHardwareMethodError is almost guaranteed to fail out every time) | 16:24 |
JayF | lucasagomes: er, I mean, fail cleaning | 16:24 |
JayF | lucasagomes: in fact, I could argue you raising that in erase_metadata makes sense too, since you're just saying "this method can't do that" | 16:24 |
jroll | the exception bubbled up will be caught, btw, and wrapper in json | 16:24 |
jroll | wrapped* | 16:24 |
jroll | serialized? whatever :) | 16:24 |
JayF | lucasagomes: although since it's in the GHM whatever exception will have basically the same effect | 16:25 |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic: Add node.resource_class field https://review.openstack.org/345055 | 16:25 |
lucasagomes | jroll, JayF cool... Aright I will update the code | 16:26 |
lucasagomes | sorry for that :D | 16:26 |
JayF | np that's what review is for ;) | 16:26 |
*** PollyZ has quit IRC | 16:26 | |
*** causten has joined #openstack-ironic | 16:27 | |
*** ijw has joined #openstack-ironic | 16:27 | |
*** ijw has quit IRC | 16:29 | |
*** ijw has joined #openstack-ironic | 16:29 | |
*** causten_ has quit IRC | 16:30 | |
*** gfhellma has joined #openstack-ironic | 16:31 | |
lucasagomes | JayF, err sorry for insisting... but apparently _shred_block_device() also does handle the exceptions and raise a boolean to indicate whether it suceeded or not | 16:32 |
lucasagomes | JayF, https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/hardware.py#L756-L780 | 16:32 |
* lucasagomes digs more in the code | 16:32 | |
JayF | lucasagomes: that's a pretty serious bug if we don't fail cleaning on an inability to erase a disk | 16:33 |
JayF | egad | 16:33 |
JayF | lucasagomes: no, it's fine | 16:33 |
sambetts | fungi, jroll: would love to get https mirroring for tinyipa, I think their performace concerns come not from the server side but the tinycorelinux side, the distro is 10mb and thats the size they don't want to increase by including GPG and SSL in the base image, and tinycorelinux currently can run on very low end computers which is the speed/performance concern they are highlighting because on | 16:34 |
sambetts | those machines it would be slow | 16:34 |
sambetts | JayF: ^ | 16:34 |
JayF | lucasagomes: because if shred(): return; then the next stanza raises the IncompatibleHardwareMethodERror | 16:34 |
JayF | lucasagomes: in erase_block_device9) | 16:34 |
lucasagomes | JayF, right and that exception gets handled by dispatch_to_managers() | 16:35 |
fungi | sambetts: they wouldn't need to include gpg and ssl in their base image to sign their images and serve them via https, right? | 16:35 |
JayF | which goes on to the next manager (there is none, because GHM is the lowest priority), and then goes on to fail | 16:35 |
lucasagomes | that is called by erase_devices() (the actual cleaning function) | 16:35 |
JayF | that dispatch_to_managers, you can think of it almost as a giant try acccept loop | 16:36 |
lucasagomes | JayF, right, I will test it locally and see what it does... looks pretty confusing | 16:36 |
JayF | call method on highest prio manager, if it raises, go to next highest prio, etc | 16:36 |
JayF | I have a large amount of confidence in that code, and there are unit tests for it :) | 16:36 |
JayF | and even better unit tests in my downstream hardware manager monstrosities, lol | 16:36 |
sambetts | fungi: oh ... the base images yes, we were talking about the packages in the forum thread, and you'd need gpg and ssl so that the package manager could verify the packages | 16:36 |
sambetts | right? | 16:36 |
lucasagomes | JayF, hah cool, alright! | 16:38 |
* lucasagomes marks the patch as WIP | 16:38 | |
fungi | sambetts: i guess it depends on what the concern is. i thought the concern JayF expressed was over us not being able to validate the tinycore images, but maybe i misunderstood his concern | 16:38 |
lucasagomes | I really need to study more the cleaning code | 16:38 |
JayF | fungi: that's def my concern; that we're pulling packages down and building without any verification whatsoever | 16:39 |
sambetts | fungi, JayF: I think there are 2 concerns, we can't validate the base image, and the package management system runs unvalidated | 16:39 |
sambetts | s/validate/verify | 16:39 |
JayF | sambetts: fungi: I filed the bug because I saw https://github.com/openstack/ironic-python-agent/blob/master/imagebuild/tinyipa/build-tinyipa.sh#L36 (http download, not even a hash check) and went "gah!" | 16:39 |
fungi | sambetts: yep, so the package manager issue does make this trickier | 16:40 |
sambetts | the base images are definatly one we could/should persue though | 16:40 |
fungi | could we mount the image on a loopback and checksum the installed files, then compare those to a signed list? | 16:40 |
*** jpich has quit IRC | 16:41 | |
fungi | (to address the package manager concern) | 16:41 |
sambetts | where does the signed list come from/ | 16:41 |
sambetts | ? | 16:41 |
*** derekh has quit IRC | 16:42 | |
fungi | sambetts: that plan would rely on the disto providing checksum lists of files in their packages and then signing an index or something | 16:43 |
fungi | do they possibly provide a secure channel for mirroring packages? | 16:43 |
fungi | we could use that to build our own trusted index if they don't want to provide one | 16:44 |
sambetts | fungi: they already provide a checksum of the package file which is checked by the package management system, problem is all package repos are http not https | 16:44 |
sambetts | because they won't include ssl in their image | 16:44 |
jroll | are the checksums served securely? | 16:44 |
sambetts | nope | 16:44 |
*** Sukhdev has quit IRC | 16:44 | |
jroll | :( | 16:44 |
sambetts | they are in the same location as the file | 16:44 |
fungi | sambetts: yep, and therefore why i'm trying to figure out workarounds to getting those checksums in a secure manner so that we can validate after the fact | 16:45 |
JayF | jroll: if the checksum inside the package was gpg signed, that doesn't matter | 16:45 |
fungi | not necessarily from within tinycore linux, but externally via loopback mount | 16:45 |
*** mgould is now known as mgould|afk | 16:45 | |
JayF | jroll: because you could use the hashes inside that checksum list to validate the rest of the files were unmodified | 16:45 |
jroll | JayF: right | 16:45 |
sambetts | ideally tinycore would mirror their packages over both https and http then we could just change the repo we're pointing at after installing SSL | 16:46 |
fungi | basically, there are ways (with varying degrees of complexity depending on how much the distro wants to help out) of doing external validation of the result of installing those packages | 16:46 |
sambetts | fungi: but all of them involve needing a verified source to start with right? of either the packages themselves or something to verify the packages against | 16:48 |
* jroll side note, our client tests are silly | 16:49 | |
fungi | sambetts: yep, either we need a separate trusted channel to mirror those files over, or they would need to provide a signed list we could validate | 16:49 |
openstackgerrit | Jim Rollenhagen proposed openstack/python-ironicclient: Add support for node.resource_class https://review.openstack.org/345080 | 16:49 |
* jroll lunches | 16:49 | |
jroll | rloo: those patches should be on your happy list now :) | 16:49 |
rloo | jroll: ha ha. we'll see :) | 16:50 |
*** milan has joined #openstack-ironic | 16:50 | |
jroll | rloo: well, I think I addressed all your comments, at least | 16:50 |
sambetts | fungi: this is one of the tinycorelinux package repo mirrors btw: http://distro.ibiblio.org/tinycorelinux/7.x/x86_64/tcz/ | 16:50 |
rloo | jroll: happy is if i don't have to review :D | 16:51 |
sambetts | fungi: they are literally just a big http file dump | 16:51 |
jroll | rloo: well, maybe you'll be happy after this review then :D | 16:51 |
rloo | jroll: i suspect i will definitely be happier. it was pretty good to start with :) | 16:51 |
*** rpioso has joined #openstack-ironic | 16:54 | |
* lucasagomes is calling it a day | 16:55 | |
*** trown|lunch is now known as trown | 16:55 | |
lucasagomes | have a great evening all, talk tomorrow :-) | 16:55 |
*** lucasagomes is now known as lucas-dinner | 16:55 | |
thiagop | night lucas-dinner | 16:55 |
*** bvandewa has joined #openstack-ironic | 16:57 | |
*** e0ne has quit IRC | 16:58 | |
*** bvandewa has quit IRC | 16:59 | |
NobodyCam | night lucas-dinner | 16:59 |
NobodyCam | morning thiagop | 16:59 |
NobodyCam | :p | 16:59 |
*** livelace has quit IRC | 17:02 | |
sambetts | fungi: looking into the tinycorelinux package manager further, the underlying code uses wget which if given a https address will automatically work with no changes if SSL is enabled in the image, so if we could get them to host a https mirror of the packages, then we install SSL into tinyipa, and then we replace the mirror in the tinyipa image with the https address, it should just work | 17:04 |
*** moshele has quit IRC | 17:04 | |
sambetts | fungi: the hard bit then is getting a https mirror of the repo | 17:04 |
sambetts | jlvillal: meeting today? | 17:05 |
*** jcoufal_ has joined #openstack-ironic | 17:05 | |
rpioso | Is there a weekly QA meeting today? | 17:05 |
sambetts | krtaylor: around for meeting?? | 17:06 |
krtaylor | hehheh, I was just going to ask | 17:06 |
rpioso | sambetts: ty | 17:06 |
krtaylor | sure, I'll run one | 17:06 |
sambetts | kl :D | 17:06 |
* jroll had the same question, heh | 17:06 | |
*** mbound has quit IRC | 17:08 | |
*** jcoufal has quit IRC | 17:08 | |
fungi | sambetts: bittorrent? | 17:09 |
*** tesseract- has quit IRC | 17:13 | |
fungi | sambetts: it looks like distro.ibiblio.org supports rsync... | 17:13 |
JayF | fungi: is that something we (OpenStack infra, idk if I should "we" that, lol) would be willing to host? | 17:14 |
JayF | fungi: that'd be one hell of a cool blog post, how to secure TinyCoreLinux without breaking their use cases | 17:14 |
fungi | JayF: we already have mirror builders for all sorts of other stuff, and i believe are using rsync to generate our centos/epel mirrors | 17:14 |
JayF | fungi: my question is more; would we share the mirror with the world, or would it be just for us | 17:15 |
*** links has joined #openstack-ironic | 17:15 | |
*** sdake has quit IRC | 17:15 | |
sambetts | but if the orginal source isn't verfied the even a secure mirror wouldn't be secure right/ | 17:15 |
sambetts | ? | 17:15 |
fungi | sambetts: JayF: right, we have no idea how trustworthy the ibiblio "mirror" is for starters (how do they mirror their copy of tinycorelinux? or is that the official download location?) | 17:16 |
sambetts | fungi: I think http://tinycorelinux.net/7.x/x86_64/tcz/ is the offical repo | 17:16 |
fungi | and rsync itself is not necessarily a solution either as the bare rsync protocol (which i think is all ibiblio supports, no ssh) is also no better than http from an authenticity perspective | 17:17 |
fungi | though that deserves further research | 17:18 |
fungi | as i haven't tried either way | 17:18 |
xavierr | hey NobodyCam o/ | 17:18 |
NobodyCam | morning xavierr :) | 17:19 |
*** spandhe has joined #openstack-ironic | 17:19 | |
xavierr | NobodyCam: too busy? hehe | 17:19 |
fungi | looks like tinycorelinux.net may have ipv6 issues :/ | 17:19 |
fungi | or not. i think it's just unreachable for me | 17:20 |
NobodyCam | lol :) yep, but its a good thing | 17:20 |
thiagop | Hey NobodyCam | 17:21 |
*** causten has quit IRC | 17:21 | |
*** rama_y has joined #openstack-ironic | 17:21 | |
*** aNupoisc has joined #openstack-ironic | 17:23 | |
*** aNupoisc has left #openstack-ironic | 17:23 | |
fungi | weird. if i try to get to tinycorelinux.net from home or from rackspace dfw it times out, but if i try from rackspace iad it works | 17:25 |
xavierr | NobodyCam: if you want to see some magic, take a look on 286192 ;) | 17:25 |
*** electrofelix has quit IRC | 17:25 | |
sambetts | fungi: v.weird... | 17:25 |
*** aNupoisc has joined #openstack-ironic | 17:25 | |
fungi | sambetts: oh, no i think they may have an overzealous defense system in place. if i ping tinycorelinux.net and then try again, it's suddenly unreachable | 17:26 |
fungi | how absurd | 17:26 |
jroll | but TLS is a waste of cpu :) | 17:27 |
sambetts | haha | 17:27 |
fungi | yep, so much as a `ping -c1 tinycorelinux.net` and now `wget http://tinycorelinux.net/` is entirely unreachable | 17:28 |
NobodyCam | xavierr: oh | 17:28 |
fungi | i've repeated that experiment from several places now | 17:28 |
sambetts | fungi: wow... | 17:28 |
fungi | that sillyness alone makes me think continuing to pursue anything to do with those clowns is a waste of your effort | 17:29 |
sambetts | fungi: I can't even ping tinycorelinux.net, but it works in the browser | 17:30 |
sambetts | :/ | 17:30 |
fungi | automatically blacklisting any ip address that sends them a single icmp echo request, but not willing to secure their package management chain in any way | 17:30 |
fungi | sambetts: browsing via a proxy maybe? | 17:30 |
sambetts | fungi: nope | 17:30 |
fungi | wget works for me until i send an icmp echoreq and then wget stops working | 17:30 |
sambetts | how weird :/ | 17:32 |
fungi | seems to unblock me fairly quickly though. after a few minutes | 17:32 |
JayF | that's bananas | 17:33 |
JayF | that's one of the most bananas network security things I've ever heard | 17:33 |
fungi | it's not just icmp echo request either | 17:33 |
fungi | if i try to hit https://tinucorelinux.net/ then i get blocked and can't go to http://tinycorelinux.net/ from the same machine for several minutes | 17:34 |
fungi | which is how i first discovered this | 17:34 |
JayF | wow, that's crazy | 17:34 |
fungi | basically if you mistakenly think you might be able to browse their site over https, they don't want you coming anywhere near them | 17:35 |
JayF | That sounds like a horribly over-aggressive WAF | 17:35 |
fungi | doesn't it though | 17:36 |
jroll | random aside: decent little home lab if someone's been looking for one :) http://www.ebay.com/itm/191849466956 | 17:36 |
JayF | hit a port we don't have open, or a protocol (icmp) we don't allow, screw you for N minutes | 17:36 |
fungi | anyway, that's enough entertainment for me for one day | 17:36 |
fungi | jroll: that's quite a haul. but i'm over my server collecting addiction now and well into recovery | 17:37 |
jroll | fungi: indeed | 17:37 |
JayF | The real cost there isn't $1300 | 17:37 |
JayF | it's the $100/mo your power bill would go up | 17:37 |
JayF | lol | 17:37 |
fungi | $100/mo seems like a conservative estimate for juicing 39 hp dl380s | 17:38 |
*** harshs has quit IRC | 17:38 | |
JayF | it was more or less a stab in the dark, I'm not surprised | 17:38 |
jroll | super conservative | 17:38 |
jroll | JayF: that's why you have ironic so you can power them off over api | 17:39 |
fungi | i had maybe a dozen (albeit antiquated) risc platforms in my zoo of a rack, and was spending at least us$150/mo to power them (not counting the auxiliary air conditioning i had to run in that room) | 17:39 |
JayF | holy crap | 17:40 |
JayF | I mean, I'd love a good test lab. Not that much though. | 17:40 |
jroll | yeah, sounds about right | 17:40 |
JayF | not to mention the personal implications of having a bunch of loud servers in the house | 17:40 |
jroll | this is why you need a rental with power included :D | 17:40 |
JayF | will Ironic give me marriage counseling via api to help fix that problem, too /s :P | 17:40 |
fungi | right, when i was moving to the beach, i decided 1. that's an ongoing expense i can do without and put toward a downpayment for a house instead, 2. i really shouldn't waste space on this, and 3. wow i _so_ don't want to have to move all that _again_ | 17:41 |
* jroll hands JayF a basement | 17:41 | |
fungi | it's not just the servers, upses, network gear, rack, move-n-cool, et cetera that take up room either. it was also the closet full of spare parts i had to keep on hand to be able to maintain them in a working state | 17:43 |
TheJulia | JayF: Heh, it could be worse, your fiancee could support the idea of installing a rack in the basement of an RV and packing it full of gear....... | 17:43 |
jroll | LOL | 17:43 |
JayF | TheJulia: What is someone who wants their RV to get .5 miles to the gallon? | 17:43 |
JayF | TheJulia: presuming that's the answer to a jeopardy question | 17:44 |
JayF | lol | 17:44 |
JayF | generator powering all the servers + extra weight | 17:44 |
JayF | you'd need to pull a tanker trailer behind just to avoid stopping for gas every 20 miles | 17:44 |
*** ijw has quit IRC | 17:45 | |
* sambetts -> home | 17:45 | |
sambetts | night all! | 17:45 |
NobodyCam | night sambetts | 17:45 |
TheJulia | JayF: Hey, we get 10 MPG right now... ;) | 17:45 |
*** sambetts is now known as sambetts|afk | 17:45 | |
sambetts|afk | o/ NobodyCam | 17:45 |
*** harshs has joined #openstack-ironic | 17:46 | |
NobodyCam | hey 10 mpg is great milage for a RV | 17:46 |
NobodyCam | :p | 17:46 |
TheJulia | Actually, thats the thing, I can't think of how to power all the extra gear... so we're pondering putting a rack someplace now :\ | 17:46 |
* TheJulia suspects a kegerator would have better milage than a small rack of servers | 17:46 | |
NobodyCam | hehehehe | 17:47 |
NobodyCam | party ar TheJulia's RV | 17:48 |
NobodyCam | ar==at | 17:48 |
*** causten has joined #openstack-ironic | 17:50 | |
*** trown is now known as trown|brb | 17:51 | |
* TheJulia has actually been pondering beer taps... and as such is a horrible person. | 17:51 | |
NobodyCam | not if there is also a cider tap... hehehehehe | 17:51 |
NobodyCam | :p | 17:51 |
* NobodyCam *ducks* | 17:52 | |
*** bvandewa has joined #openstack-ironic | 17:52 | |
*** bvandewa has quit IRC | 17:52 | |
*** spandhe has quit IRC | 17:53 | |
*** bvandewa has joined #openstack-ironic | 17:53 | |
*** spandhe has joined #openstack-ironic | 17:53 | |
*** spandhe has quit IRC | 17:54 | |
*** links has quit IRC | 17:54 | |
openstackgerrit | Aparna proposed openstack/proliantutils: Add hpssa support only in raid mode https://review.openstack.org/347978 | 18:01 |
*** harshs has quit IRC | 18:01 | |
*** TuanL has joined #openstack-ironic | 18:07 | |
*** ahqa has joined #openstack-ironic | 18:09 | |
*** TuanL has quit IRC | 18:11 | |
ahqa | Hi, I am deploying Ubutnu Trusty on server and using elements "vm" "cloud-init-datasources" and "enable-serial-console". After server is deployed, the IP is not updated from Config Drive, server is still having the old IP assigned by DHCP server. Everything else gets configured from Config Drive except the ip. I am using diskimage-builder to build the image. | 18:12 |
*** joprovost1 has joined #openstack-ironic | 18:12 | |
*** PollyZ has joined #openstack-ironic | 18:12 | |
*** joprovost has quit IRC | 18:13 | |
*** joprovost1 is now known as joprovost | 18:13 | |
*** baoli_ has quit IRC | 18:18 | |
cinerama | ahqa, not sure about that one, i'd check your logs on the server | 18:19 |
ahqa | cinerama: I will, thanks. | 18:23 |
openstackgerrit | Vladyslav Drok proposed openstack/ironic: Add api-ref for new port fields https://review.openstack.org/325299 | 18:24 |
*** ayoung has quit IRC | 18:25 | |
*** gfhellma has quit IRC | 18:28 | |
*** trown|brb is now known as trown | 18:28 | |
gmmaha | ahqa: did you add the ConfigDrive option for additional elements when creating the image? | 18:29 |
gmmaha | i think thats needed for cloud-init to work properly | 18:29 |
* gmmaha goes to find the doc | 18:30 | |
*** harshs has joined #openstack-ironic | 18:30 | |
*** joprovost1 has joined #openstack-ironic | 18:30 | |
*** joprovost has quit IRC | 18:31 | |
*** joprovost1 is now known as joprovost | 18:31 | |
ahqa | gmmaha: I don't have ConfigDrive element in my elements directory. | 18:32 |
gmmaha | ahqa: https://github.com/openstack/diskimage-builder/blob/master/elements/cloud-init-datasources/README.rst | 18:32 |
gmmaha | i assume you are using diskimage-builder to create the ubuntu image? | 18:32 |
gmmaha | ahqa: in my setup, i have this as mods to playbooks/inventory/group_vars/localhost to get cloud-init working http://paste.openstack.org/show/542681/ | 18:33 |
ahqa | Correct, I have already set DIB_CLOUD_INIT_DATASOURCES="ConfigDrive" in my cloud-init-datasources/install.d/ | 18:33 |
*** baoli has joined #openstack-ironic | 18:34 | |
gmmaha | ahqa: then i will check if the file '/etc/cloud/cloud.cfg.d/91-dib-cloud-init-datasources.cfg' has the right bits and check the system logs as cinerama suggested | 18:35 |
gmmaha | i wonder if the 'Openstack' option makes a difference | 18:35 |
*** spandhe has joined #openstack-ironic | 18:36 | |
ahqa | The file has "datasource_list: [ ConfigDrive, None ] | 18:37 |
JayF | sambetts|afk: jroll: Any objection to making the tinyipa 256 job the default if it's passing now? | 18:38 |
ahqa | gmmaha: I will try adding OpenStack option. | 18:39 |
TheJulia | OpenStack is for the metadata service | 18:40 |
jroll | JayF: if it's passing, sure | 18:40 |
gmmaha | TheJulia: isn't that the one which sets the IP of the machine and hostname etc? | 18:40 |
gmmaha | the metadata service? | 18:40 |
TheJulia | gmmaha: that is DHCP | 18:43 |
TheJulia | gmmaha: unless your trying to do it in a configdrive with a static address assignment. If your using cloud-init, you have to pre-create the appropriate configuration files to be placed in the appropriate locations in the config drives. | 18:44 |
*** ChubYann has joined #openstack-ironic | 18:45 | |
gmmaha | TheJulia: right, and i was under the assumption that ahqa is use static IPs. I thought bifrost's deploy-dynamic piece will do the needful as far as config files in the drives | 18:45 |
*** e0ne has joined #openstack-ironic | 18:45 | |
gmmaha | and cloud-init can consume them fine. (atleast it worked when i was deploying with trusty images of ubuntu) :) | 18:45 |
ahqa | I am using static IP which is assigned to servers. | 18:46 |
*** e0ne has quit IRC | 18:47 | |
TheJulia | via the inventory? | 18:47 |
ahqa | Previously I was using simple-init but there is a problem with latest build and simple-init is not working with ubuntu. | 18:47 |
ahqa | Yes. I have a python file and a json file. | 18:47 |
ahqa | Json file contains all the information. | 18:48 |
*** aNupoisc has quit IRC | 18:48 | |
TheJulia | okay, bifrost does not know how to do the needful in terms of other distributions network config files, the driver behind the use of simple-init is that it does mac->ip matching because initialization ordering can be different | 18:49 |
TheJulia | i.e. can't assume first interface you know about will be eth0 kind of thing | 18:49 |
TheJulia | ahqa: You could do static dhcp assignments, but truly sounds like a) minimum needful for other distros and b) simple-init needs to be fixed... | 18:50 |
*** causten_ has joined #openstack-ironic | 18:51 | |
ahqa | TheJulia: I will try to do it, thank you. | 18:52 |
gmmaha | TheJulia: i didn't know.. thanks for the info about static IPs and simple-init | 18:53 |
*** aNupoisc has joined #openstack-ironic | 18:55 | |
*** causten has quit IRC | 18:56 | |
TheJulia | FWIW, we do write an /etc/network/interfaces file for debian, I don't know if it will work with the latest ubuntu, and it might not work if you have multiple network interfaces. It was really put in place for cirros machines could have network connectivity set in CI. | 18:56 |
TheJulia | s/do/can/ | 18:56 |
NobodyCam | TheJulia: question. Is there a reason I have forgotten that we restart dnsmasq twice here: https://github.com/openstack/bifrost/blob/master/playbooks/roles/bifrost-ironic-install/tasks/start.yml#L45-L57 | 19:02 |
gmmaha | TheJulia: That would be cool. adding the ability to write /etc/network/interfaces for ubuntu and probably other distros. Not sure what that entails | 19:04 |
*** gfhellma has joined #openstack-ironic | 19:05 | |
TheJulia | gmmaha: problem is, beyond adding "-e write_interfaces_file=true" to your deployment command line, is getting the interfaces right :( As I mentioned earlier, interface ordering, so truthfully simple-init just really needs to be fixed | 19:06 |
gmmaha | TheJulia: right. that was the biggest problem i ran into when i was deploying machines ( not getting predictable NIC names) | 19:06 |
gmmaha | was hoping with Ubuntu 15.XX>= moving to systemd that might be less of a problem. But need a big sample data set to say one way or the other | 19:07 |
gmmaha | atleast we will no longer end with an 'ethX' | 19:07 |
gmmaha | hopefully | 19:07 |
TheJulia | NobodyCam: restarted != reloaded maybe | 19:08 |
NobodyCam | both have state=restarted | 19:09 |
TheJulia | NobodyCam: so we kill dnsmasq on line 38.... I feel like some ordering has been changed around and it just got duplicated... | 19:09 |
NobodyCam | kinda what I was thinking.. :) | 19:10 |
TheJulia | if I remember fairly explicitly, dnsmasq has no reload capability out of the box via it's init scripts/configs on many platforms | 19:10 |
NobodyCam | yea | 19:10 |
TheJulia | hence why we ended up with restarted... also because reload does not explicitly restart | 19:11 |
NobodyCam | but I think we only need to restart it once | 19:11 |
TheJulia | yeah, only once should be perfectly fine | 19:11 |
NobodyCam | :) | 19:11 |
TheJulia | I bet the second one got introduced with dhcp stuff which has been refactored | 19:11 |
* TheJulia hopes it is restarting... or has a direct hup signal in there somceplace | 19:11 | |
NobodyCam | lol and being refactored again | 19:11 |
TheJulia | yeah | 19:12 |
TheJulia | but yeah, the note shrews put in there is nice and detailed as to why | 19:13 |
NobodyCam | yea | 19:13 |
* NobodyCam kicks off another run and steps afk for a few minutes | 19:14 | |
openstackgerrit | Julia Kreger proposed openstack/ironic: [WIP] Add storage_interface DB field and object https://review.openstack.org/348005 | 19:14 |
openstackgerrit | Julia Kreger proposed openstack/ironic: [WIP] Add storage_interface to base driver class https://review.openstack.org/348006 | 19:14 |
openstackgerrit | Julia Kreger proposed openstack/ironic: [WIP] Add storage_interface DB field and object https://review.openstack.org/348005 | 19:20 |
*** michauds has quit IRC | 19:20 | |
*** harshs has quit IRC | 19:21 | |
*** michauds has joined #openstack-ironic | 19:22 | |
*** sdake has joined #openstack-ironic | 19:25 | |
*** jcoufal_ has quit IRC | 19:26 | |
*** PollyZ has quit IRC | 19:32 | |
openstackgerrit | Julia Kreger proposed openstack/ironic: Introduce VolumeConnector object https://review.openstack.org/214584 | 19:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic: Introduce new RPCs to support volume connector operation https://review.openstack.org/214585 | 19:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic: Introduce new RPCs to support volume target operations https://review.openstack.org/285220 | 19:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic: Introduce VolumeTarget object https://review.openstack.org/285219 | 19:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic: Add volume_targets table to save target information https://review.openstack.org/285218 | 19:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic: Add volume_connector table to save connector information https://review.openstack.org/200983 | 19:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic: [WIP] Add storage_interface DB field and object https://review.openstack.org/348005 | 19:35 |
openstackgerrit | Julia Kreger proposed openstack/ironic: [WIP] Add storage_interface to base driver class https://review.openstack.org/348006 | 19:35 |
* TheJulia feels awful now | 19:35 | |
*** anush_ has quit IRC | 19:40 | |
*** sdake has quit IRC | 19:43 | |
jroll | O_O | 19:43 |
jroll | TheJulia: I hope you feel awful for writing a bunch of code and not physically awful | 19:43 |
TheJulia | jroll: rebasing the entire chain which I was trying to avoid :( | 19:44 |
TheJulia | I feel awful for that, and only that :) | 19:45 |
jroll | ah | 19:45 |
TheJulia | and then I had to go fix topics | 19:45 |
cinerama | i was like "oh i'll just see if anyone's reviewed...OH NO WHERE IS THE CODE" :) | 19:47 |
* TheJulia thinks it is time to run into town and do food shopping and stuff before starting on the next part of the changeset | 19:47 | |
*** sdake has joined #openstack-ironic | 19:53 | |
*** itooon has joined #openstack-ironic | 19:56 | |
*** PollyZ has joined #openstack-ironic | 20:02 | |
*** rpioso has quit IRC | 20:09 | |
*** jaybeale has quit IRC | 20:14 | |
*** tangchen has quit IRC | 20:17 | |
*** michauds has quit IRC | 20:18 | |
*** tangchen has joined #openstack-ironic | 20:19 | |
*** Nisha_away has quit IRC | 20:25 | |
*** michauds has joined #openstack-ironic | 20:27 | |
*** harshs has joined #openstack-ironic | 20:27 | |
*** rpioso has joined #openstack-ironic | 20:28 | |
openstackgerrit | Ruby Loo proposed openstack/ironic-python-agent: Use ironic_lib's execute() https://review.openstack.org/348023 | 20:30 |
*** aswadr_ has quit IRC | 20:31 | |
*** PollyZ has quit IRC | 20:33 | |
*** itooon has quit IRC | 20:35 | |
*** sdake has quit IRC | 20:46 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic: Add node.resource_class field https://review.openstack.org/345055 | 20:48 |
*** joprovost has quit IRC | 20:48 | |
*** PollyZ has joined #openstack-ironic | 20:52 | |
openstackgerrit | Jim Rollenhagen proposed openstack/python-ironicclient: Add support for node.resource_class https://review.openstack.org/345080 | 20:53 |
*** ayoung has joined #openstack-ironic | 20:55 | |
*** gfhellma1 has joined #openstack-ironic | 21:00 | |
*** gfhellma has quit IRC | 21:03 | |
jroll | devananda: around? looking at https://review.openstack.org/#/c/325599/16/ironic/common/policy.py | 21:08 |
jroll | devananda: can regular people assign whatever name they want to a tenant? | 21:08 |
*** clenimar has quit IRC | 21:08 | |
jroll | re: line 46 | 21:09 |
jroll | oh wait, we might not be using that | 21:09 |
* jroll reads on | 21:09 | |
*** jjohnson2__ has quit IRC | 21:11 | |
openstackgerrit | Clark Boylan proposed openstack/python-ironicclient: Fix coverage target https://review.openstack.org/348043 | 21:12 |
*** jaybeale has joined #openstack-ironic | 21:18 | |
openstackgerrit | Julia Kreger proposed openstack/ironic-specs: Boot from Volume - Reference Drivers https://review.openstack.org/294995 | 21:19 |
*** wajdi has quit IRC | 21:20 | |
*** rpioso has quit IRC | 21:21 | |
TheJulia | reminder for cores: as requested during the meeting, please review https://review.openstack.org/#/c/230924 as a new version of ironic-lib is needed to fix the functionality defect | 21:22 |
jroll | devananda: +2 on the policy stuff, lots of nits, can you follow up that stuff if I land it now? | 21:22 |
mrda | Morning Ironic | 21:26 |
xavierr | morning mrda | 21:26 |
jroll | \o mrda | 21:26 |
thiagop | morning mrda | 21:26 |
*** xavierr is now known as xavierr-afk | 21:26 | |
TheJulia | good morning mrda | 21:27 |
mrda | o/ | 21:28 |
*** jaybeale has quit IRC | 21:32 | |
*** causten_ has quit IRC | 21:35 | |
jroll | see y'all tomorrow | 21:38 |
TheJulia | goodnight jroll | 21:39 |
*** causten has joined #openstack-ironic | 21:40 | |
openstackgerrit | Jim Rollenhagen proposed openstack/ironic: Add node.resource_class field https://review.openstack.org/345055 | 21:44 |
*** spandhe has left #openstack-ironic | 21:50 | |
*** causten has quit IRC | 21:53 | |
*** causten has joined #openstack-ironic | 21:53 | |
*** causten has quit IRC | 21:57 | |
*** rama_y has quit IRC | 21:57 | |
*** causten_ has joined #openstack-ironic | 21:59 | |
*** causten__ has joined #openstack-ironic | 22:01 | |
*** rama_y has joined #openstack-ironic | 22:01 | |
*** mbound has joined #openstack-ironic | 22:03 | |
*** causten_ has quit IRC | 22:05 | |
*** harshs has quit IRC | 22:06 | |
*** mbound has quit IRC | 22:07 | |
*** causten__ has quit IRC | 22:09 | |
*** harshs has joined #openstack-ironic | 22:10 | |
*** causten has joined #openstack-ironic | 22:13 | |
*** causten_ has joined #openstack-ironic | 22:15 | |
*** causten has quit IRC | 22:18 | |
thiagop | I'm calling it a day | 22:19 |
thiagop | see you tomorrow | 22:19 |
*** thiagop has quit IRC | 22:19 | |
*** causten_ has quit IRC | 22:20 | |
*** causten has joined #openstack-ironic | 22:21 | |
openstackgerrit | Mario Villaplana proposed openstack/ironic: Add power state change notifications https://review.openstack.org/321865 | 22:25 |
openstackgerrit | Mario Villaplana proposed openstack/ironic: Add notification base classes and docs https://review.openstack.org/298461 | 22:25 |
*** causten has quit IRC | 22:25 | |
*** michauds has quit IRC | 22:30 | |
*** causten has joined #openstack-ironic | 22:36 | |
*** causten has quit IRC | 22:38 | |
*** causten has joined #openstack-ironic | 22:43 | |
*** trown is now known as trown|outtypewww | 22:49 | |
*** PollyZ has quit IRC | 23:00 | |
*** sabeen has quit IRC | 23:00 | |
*** siva_krish has quit IRC | 23:01 | |
*** siva_krish has joined #openstack-ironic | 23:02 | |
*** vishwanathj has quit IRC | 23:02 | |
*** causten has quit IRC | 23:02 | |
*** siva_krish has quit IRC | 23:03 | |
*** causten has joined #openstack-ironic | 23:03 | |
*** siva_krish has joined #openstack-ironic | 23:03 | |
openstackgerrit | Jay Faulkner proposed openstack/ironic: Pass agent metrics config via conductor https://review.openstack.org/301919 | 23:05 |
*** causten has quit IRC | 23:08 | |
*** siva_krish has quit IRC | 23:08 | |
*** sdake has joined #openstack-ironic | 23:17 | |
*** fumie has joined #openstack-ironic | 23:17 | |
*** hoangcx2 has quit IRC | 23:24 | |
*** hoangcx has joined #openstack-ironic | 23:24 | |
*** sdake has quit IRC | 23:25 | |
*** mbound has joined #openstack-ironic | 23:28 | |
*** gfhellma has joined #openstack-ironic | 23:30 | |
*** gfhellma1 has quit IRC | 23:31 | |
*** Sukhdev has joined #openstack-ironic | 23:36 | |
*** mtanino has quit IRC | 23:37 | |
*** harshs has quit IRC | 23:37 | |
*** chlong has quit IRC | 23:38 | |
*** gfhellma has quit IRC | 23:40 | |
*** harshs has joined #openstack-ironic | 23:40 | |
JayF | Should we start having dev docs built for ironic_lib? I'm thinking yes. | 23:42 |
JayF | I'm going to file a bug for that. | 23:42 |
morgabra | jroll: JayF: and friends: https://github.com/openstack/nova/blob/master/nova/virt/xenapi/driver.py#L320 | 23:43 |
morgabra | how should ironic interact with that you think? | 23:43 |
JayF | It's kinda weird, right? | 23:43 |
JayF | I wonder how the libvirt driver does it, as that's kinda considered the 'reference' nova driver | 23:44 |
morgabra | Well, presumably neutron exposes it somehow: https://wiki.openstack.org/wiki/Neutron/Metering/Bandwidth | 23:44 |
morgabra | even though that doesn't quite fit, there might be a mechanism for it | 23:44 |
morgabra | regardless, it breaks the abstraction if the virt driver talks to neutron | 23:45 |
morgabra | so it seems like an ironic endpoint for it is the only choice | 23:45 |
morgabra | (or just don't do it at all, lol) | 23:45 |
JayF | Oh wow. That's only implemented by the xenapi. | 23:45 |
JayF | Not in libvirt or anywhere else. | 23:45 |
morgabra | yeah.... | 23:45 |
JayF | https://github.com/openstack/nova/search?utf8=%E2%9C%93&q=get_all_bw_counters | 23:45 |
JayF | man, that's a great docstring too on the base class, lol | 23:46 |
morgabra | oh I know | 23:46 |
morgabra | heh | 23:46 |
JayF | it doesn't even describe what sort of thing it expects to be returned | 23:46 |
JayF | I'd suspect step 1 is finding that out, and pushing a docstring update to the base class + xenapi | 23:46 |
morgabra | {'instance-001' : { 12:34:56:78:90:12 : {'bw_in': 0, ....}} | 23:47 |
JayF | but like, does it want just what the pure counter says or what? | 23:47 |
morgabra | presumably, the counter being in bytes | 23:47 |
JayF | like does it start counting at 0 from instance creation? | 23:47 |
JayF | or does it take the first measurement, then do math for the rest | 23:47 |
morgabra | nah, you can read the manager code | 23:47 |
JayF | So basically you'd have to have something like | 23:48 |
JayF | a periodic task in Ironic to update some counter value on a node, but only for nodes with an instance uuid | 23:48 |
morgabra | first run for a given instance/mac combo, it saves the counter | 23:48 |
morgabra | and the rest is just updating it | 23:49 |
JayF | hm. okay | 23:49 |
morgabra | well, ironic could just passthrough to whatever is driving the network | 23:49 |
morgabra | do it live | 23:49 |
JayF | proxy apis haven't worked out so well for nova, lol | 23:49 |
morgabra | that's a whole other can-o-worms I think, getting neutron to expose that kind of data | 23:49 |
morgabra | but hey | 23:49 |
JayF | I think this is squarely mailing list fodder | 23:49 |
JayF | because it's cross 2, maybe 3 projects | 23:50 |
JayF | and I'd be very curious if anyone else is thinking abouts tuff in this place. | 23:50 |
morgabra | getting a counter for a specific interface in neutronland for a given port is also impossible for many types of nets/hardware | 23:51 |
JayF | sure; but it's impossible for some nova drivers too | 23:51 |
JayF | so I'd imagine we'd handle it the same way they do; returning nothing or raise NotImplemented if we can't do it for that node | 23:52 |
*** sabeen has joined #openstack-ironic | 23:52 | |
JayF | since Nova will then cache that the instance can't do b/w measuring | 23:52 |
JayF | jroll: WDYT about https://bugs.launchpad.net/ironic-lib/+bug/1607130 | 23:53 |
openstack | Launchpad bug 1607130 in ironic-lib "ironic-lib should provide developer documentation" [Undecided,New] | 23:53 |
*** adu has joined #openstack-ironic | 23:57 | |
*** hoangcx2 has joined #openstack-ironic | 23:58 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!