| *** wendar has joined #openstack-ironic | 00:01 | |
| *** lazy_prif has joined #openstack-ironic | 00:01 | |
| *** lazy_prif is now known as lazy_prince | 00:02 | |
| *** Isotopp_ has joined #openstack-ironic | 00:03 | |
| *** bigjools_ has joined #openstack-ironic | 00:03 | |
| *** ryanpetrello_ has joined #openstack-ironic | 00:05 | |
| *** stevebak` has joined #openstack-ironic | 00:06 | |
| *** dutsmoc has joined #openstack-ironic | 00:06 | |
| *** mgagne has joined #openstack-ironic | 00:08 | |
| *** russell_h_ has joined #openstack-ironic | 00:08 | |
| *** mgagne is now known as Guest72739 | 00:09 | |
| *** jroll|dupe has joined #openstack-ironic | 00:09 | |
| *** coolsvap_ has joined #openstack-ironic | 00:12 | |
| *** EmilienM_ has joined #openstack-ironic | 00:13 | |
| *** killer_prince has quit IRC | 00:13 | |
| *** coolsvap has quit IRC | 00:13 | |
| *** wendar_ has quit IRC | 00:13 | |
| *** russell_h has quit IRC | 00:13 | |
| *** victor_lowther__ has quit IRC | 00:13 | |
| *** Ng has quit IRC | 00:13 | |
| *** adam_g has quit IRC | 00:13 | |
| *** stevebaker has quit IRC | 00:13 | |
| *** ryanpetrello has quit IRC | 00:13 | |
| *** jroll has quit IRC | 00:13 | |
| *** bigjools has quit IRC | 00:13 | |
| *** comstud has quit IRC | 00:13 | |
| *** Guest6936 has quit IRC | 00:13 | |
| *** EmilienM has quit IRC | 00:13 | |
| *** Isotopp has quit IRC | 00:13 | |
| *** coolsvap_ is now known as coolsvap | 00:13 | |
| *** ryanpetrello_ is now known as ryanpetrello | 00:13 | |
| *** russell_h_ is now known as russell_h | 00:13 | |
| *** jroll|dupe is now known as jroll | 00:13 | |
| *** EmilienM_ is now known as EmilienM | 00:13 | |
| *** Ng_ has joined #openstack-ironic | 00:17 | |
| *** igordcard has joined #openstack-ironic | 00:17 | |
| *** adam_g has joined #openstack-ironic | 00:19 | |
| *** adam_g has quit IRC | 00:19 | |
| *** adam_g has joined #openstack-ironic | 00:19 | |
| *** yuanying has quit IRC | 00:22 | |
| *** HenryG has quit IRC | 00:37 | |
| *** mitz_ has joined #openstack-ironic | 01:05 | |
| *** yuanying has joined #openstack-ironic | 01:19 | |
| *** bigjools_ is now known as bigjools | 01:19 | |
| *** bigjools has quit IRC | 01:20 | |
| *** bigjools has joined #openstack-ironic | 01:20 | |
| *** pcrews has joined #openstack-ironic | 01:41 | |
| *** chenglch has joined #openstack-ironic | 01:47 | |
| *** yuanying has quit IRC | 01:52 | |
| *** pcrews has quit IRC | 01:53 | |
| *** Poornima_ has joined #openstack-ironic | 02:21 | |
| *** datajerk has quit IRC | 02:27 | |
| *** datajerk has joined #openstack-ironic | 02:27 | |
| *** yuanying has joined #openstack-ironic | 02:49 | |
| *** ramineni has joined #openstack-ironic | 02:56 | |
| *** jrist has joined #openstack-ironic | 03:09 | |
| openstackgerrit | Anusha Ramineni proposed a change to openstack/ironic-python-agent: This script adds support to build iso image from the given kernel and ramdisk. https://review.openstack.org/121433 | 03:15 |
|---|---|---|
| *** yuanying has quit IRC | 03:22 | |
| openstackgerrit | Anusha Ramineni proposed a change to openstack/ironic-python-agent: Add support to build iso image https://review.openstack.org/121433 | 03:40 |
| *** Poornima_ has quit IRC | 03:50 | |
| *** Poornima_ has joined #openstack-ironic | 04:15 | |
| *** yuanying has joined #openstack-ironic | 04:19 | |
| *** rushiagr_away is now known as rushiagr | 04:37 | |
| *** pensu has joined #openstack-ironic | 04:41 | |
| *** pensu has left #openstack-ironic | 04:41 | |
| *** lazy_prince is now known as killer_prince | 04:46 | |
| *** teju has joined #openstack-ironic | 04:46 | |
| teju | hi all.... ironic node "Provisioning State" is stuck in "active" and the bare metal node's state is "System running in UEFI".....ironic conductor logs : http://pastebin.com/DB5PPQ6v | 04:51 |
| *** yuanying has quit IRC | 04:52 | |
| *** jrist has quit IRC | 04:54 | |
| *** jrist has joined #openstack-ironic | 04:55 | |
| *** pradipta_away is now known as pradipta | 05:20 | |
| *** rushiagr is now known as rushiagr_away | 05:24 | |
| *** killer_prince has quit IRC | 05:42 | |
| *** lazy_prince has joined #openstack-ironic | 05:42 | |
| *** rakesh_hs has joined #openstack-ironic | 05:46 | |
| *** yuanying has joined #openstack-ironic | 05:49 | |
| *** k4n0 has joined #openstack-ironic | 05:57 | |
| openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/120701 | 06:09 |
| *** nikunj2512 has joined #openstack-ironic | 06:11 | |
| *** rushiagr_away is now known as rushiagr | 06:18 | |
| *** rushiagr is now known as rushiagr_away | 06:23 | |
| teju | hi all... can someone provide commands to build images for ironic bare metal deployment? | 06:25 |
| *** pensu has joined #openstack-ironic | 06:36 | |
| *** yuanying has quit IRC | 06:52 | |
| *** bluex-pl has joined #openstack-ironic | 07:22 | |
| *** pradipta is now known as pradipta_away | 07:25 | |
| *** pradipta_away is now known as pradipta | 07:28 | |
| *** ifarkas has joined #openstack-ironic | 07:28 | |
| *** andreykurilin_ has joined #openstack-ironic | 07:31 | |
| *** jistr has joined #openstack-ironic | 07:33 | |
| *** pradipta is now known as pradipta_away | 07:35 | |
| *** romcheg has joined #openstack-ironic | 07:38 | |
| *** andreykurilin_ has quit IRC | 07:49 | |
| *** yuanying has joined #openstack-ironic | 07:49 | |
| *** victor_lowther__ has joined #openstack-ironic | 07:59 | |
| *** Ng_ is now known as Ng | 08:02 | |
| *** coolsvap has quit IRC | 08:04 | |
| *** coolsvap has joined #openstack-ironic | 08:04 | |
| *** Ng has quit IRC | 08:04 | |
| *** Ng has joined #openstack-ironic | 08:04 | |
| *** chenglch has quit IRC | 08:05 | |
| *** chenglch has joined #openstack-ironic | 08:05 | |
| *** ramineni has quit IRC | 08:05 | |
| *** ramineni has joined #openstack-ironic | 08:05 | |
| *** Poornima_ has quit IRC | 08:05 | |
| *** Poornima_ has joined #openstack-ironic | 08:05 | |
| *** teju has quit IRC | 08:05 | |
| *** teju has joined #openstack-ironic | 08:05 | |
| *** rakesh_hs has quit IRC | 08:05 | |
| *** rakesh_hs has joined #openstack-ironic | 08:05 | |
| *** k4n0 has quit IRC | 08:05 | |
| *** k4n0 has joined #openstack-ironic | 08:05 | |
| *** nikunj2512 has quit IRC | 08:05 | |
| *** nikunj2512 has joined #openstack-ironic | 08:05 | |
| *** ifarkas has quit IRC | 08:06 | |
| *** ifarkas has joined #openstack-ironic | 08:06 | |
| *** jistr has quit IRC | 08:06 | |
| *** jistr has joined #openstack-ironic | 08:06 | |
| *** derekh has joined #openstack-ironic | 08:08 | |
| *** athomas has joined #openstack-ironic | 08:14 | |
| *** yuanying_ has joined #openstack-ironic | 08:17 | |
| *** ndipanov has joined #openstack-ironic | 08:19 | |
| *** yuanying has quit IRC | 08:20 | |
| *** yuanying_ has quit IRC | 08:23 | |
| *** lucasagomes has joined #openstack-ironic | 08:31 | |
| *** viktors|afk is now known as viktors | 08:31 | |
| *** athomas has quit IRC | 08:37 | |
| *** athomas has joined #openstack-ironic | 08:45 | |
| *** Poornima_ has quit IRC | 08:46 | |
| *** k4n0 has quit IRC | 08:48 | |
| *** Poornima_ has joined #openstack-ironic | 08:50 | |
| lucasagomes | is our gate broken? | 08:59 |
| lucasagomes | morning all :) | 08:59 |
| *** dtantsur has joined #openstack-ironic | 09:00 | |
| dtantsur | Morning Ironic | 09:02 |
| romcheg | Morning dtantsur, lucasagomes! | 09:02 |
| dtantsur | Got locked out of my server today, so no chat history for me :( anything interesting happened since Fri? | 09:02 |
| *** k4n0 has joined #openstack-ironic | 09:08 | |
| lucasagomes | dtantsur, uefi merged | 09:08 |
| lucasagomes | :) | 09:08 |
| chenglch | This is the error I get from a fresh devstack when I use nova boot . {"message": "consume_from_instance() takes exactly 2 arguments (3 given)} | 09:08 |
| dtantsur | \o/ | 09:08 |
| lucasagomes | chenglch, :/ from master? | 09:13 |
| chenglch | yes | 09:13 |
| dtantsur | `grep -r consume_from_instance ironic` is empty. something in Nova/ironic driver? | 09:14 |
| *** pelix1 has joined #openstack-ironic | 09:14 | |
| chenglch | This is the error from nova show http://paste.openstack.org/show/111654/ | 09:16 |
| dtantsur | this stack trace is rather useless. could you look for anything else in nova logs? | 09:17 |
| dtantsur | chenglch, https://review.openstack.org/#/c/121397/ | 09:18 |
| dtantsur | either apply this or rebuid your devstack | 09:18 |
| *** yuanying has joined #openstack-ironic | 09:19 | |
| *** pelix1 is now known as pelix | 09:24 | |
| chenglch | http://paste.openstack.org/show/111660/ | 09:29 |
| chenglch | consume_from_instance from nova/scheduler/ironic_host_manager.py only has 2 arguments | 09:30 |
| *** MattMan has joined #openstack-ironic | 09:41 | |
| *** rushiagr_away is now known as rushiagr | 09:42 | |
| chenglch | def consume_from_instance(self, instance): in ironic_host_manager.py def consume_from_instance(self, context, instance): in host_manager.py | 09:43 |
| pensu | chenglch: did you try the patch dtantsur give? Does it work? Somehow I am still getting the same error. | 09:48 |
| *** yuanying has quit IRC | 09:53 | |
| lucasagomes | alright the nova patch was approved | 10:04 |
| dtantsur | which one? | 10:09 |
| lucasagomes | the one fixing our gate | 10:22 |
| lucasagomes | https://review.openstack.org/#/c/121397/ | 10:22 |
| *** chenglch has quit IRC | 10:32 | |
| *** Krast has joined #openstack-ironic | 10:34 | |
| *** Krast has quit IRC | 10:35 | |
| *** HenryG has joined #openstack-ironic | 10:37 | |
| *** rushiagr is now known as rushiagr_away | 10:37 | |
| *** chenglch has joined #openstack-ironic | 10:39 | |
| chenglch | The patch work, nova boot works now | 10:39 |
| *** k4n0 has quit IRC | 10:42 | |
| *** pensu has quit IRC | 10:43 | |
| *** pensu has joined #openstack-ironic | 10:43 | |
| *** rushiagr_away is now known as rushiagr | 10:44 | |
| *** Poornima_ has quit IRC | 10:46 | |
| *** nikunj2512 has quit IRC | 10:46 | |
| *** Poornima_ has joined #openstack-ironic | 10:48 | |
| *** yuanying has joined #openstack-ironic | 10:50 | |
| *** bluex-pl has quit IRC | 10:50 | |
| *** bluex-pl has joined #openstack-ironic | 10:54 | |
| lucasagomes | chenglch, w00t :) | 10:55 |
| lucasagomes | chenglch, it's now approved in nova, soon will be fixed in master. Cheers for bringing it up here | 10:56 |
| *** rushiagr is now known as rushiagr_away | 10:56 | |
| *** lazy_prince has quit IRC | 10:58 | |
| *** ramineni has quit IRC | 10:58 | |
| *** k4n0 has joined #openstack-ironic | 11:04 | |
| *** lazy_prince has joined #openstack-ironic | 11:06 | |
| lazy_prince | hi all.. i am getting Cannot validate iSCSI deploy. The following parameters were not passed to ironic: ['root_gb', 'image_source']. do i need to provide image_source as part of the node-registration...? | 11:15 |
| *** yuanying has quit IRC | 11:22 | |
| *** pensu1 has joined #openstack-ironic | 11:23 | |
| *** pensu has quit IRC | 11:23 | |
| *** romcheg has quit IRC | 11:33 | |
| *** lazy_prince has quit IRC | 11:34 | |
| *** Poornima_ has quit IRC | 11:35 | |
| lucasagomes | lazy_prince, nop, nova will set those values for you | 11:36 |
| lucasagomes | see: nova/virt/ironic/patcher.py (in nova tree) | 11:37 |
| *** lucasagomes is now known as lucas-hungry | 11:38 | |
| *** rushiagr_away is now known as rushiagr | 11:42 | |
| *** pensu1 has quit IRC | 11:47 | |
| *** Madkiss has joined #openstack-ironic | 11:59 | |
| Madkiss | Hey folks | 12:00 |
| dtantsur | hi | 12:00 |
| *** romcheg has joined #openstack-ironic | 12:03 | |
| *** ramineni has joined #openstack-ironic | 12:07 | |
| *** pensu has joined #openstack-ironic | 12:08 | |
| *** rakesh_hs has quit IRC | 12:14 | |
| *** romcheg has quit IRC | 12:17 | |
| *** romcheg has joined #openstack-ironic | 12:17 | |
| *** yuanying has joined #openstack-ironic | 12:19 | |
| *** faizan_ has joined #openstack-ironic | 12:24 | |
| *** Poornima_ has joined #openstack-ironic | 12:25 | |
| *** faizan_ has left #openstack-ironic | 12:32 | |
| *** faizan has joined #openstack-ironic | 12:32 | |
| *** dtantsur is now known as dtantsur|brb | 12:34 | |
| *** lucas-hungry is now known as lucasagomes | 12:38 | |
| Madkiss | So, if I want to give ironic a test run right now, what would be a good combination for the tool chain to do that? | 12:50 |
| *** dprince has joined #openstack-ironic | 12:51 | |
| *** yuanying has quit IRC | 12:52 | |
| lucasagomes | Madkiss, http://docs.openstack.org/developer/ironic/dev/dev-quickstart.html | 12:54 |
| Madkiss | thanks | 12:56 |
| *** jcoufal has joined #openstack-ironic | 12:59 | |
| *** jistr has quit IRC | 13:00 | |
| lucasagomes | yw :) | 13:02 |
| *** jistr has joined #openstack-ironic | 13:03 | |
| lucasagomes | nova fix is merged o/ | 13:03 |
| *** lazy_prince has joined #openstack-ironic | 13:04 | |
| faizan | Hi lucasagomes, | 13:08 |
| faizan | lucasagomes, I found some code issue in pxe_ilo driver code, which got merged last week. | 13:10 |
| lucasagomes | faizan, right | 13:10 |
| lucasagomes | faizan, we can open a bug, tag to RC | 13:10 |
| faizan | lucasagomes, I raised a bug https://bugs.launchpad.net/ironic/+bug/1369545 | 13:10 |
| lucasagomes | and submit the code | 13:10 |
| lucasagomes | yup | 13:10 |
| *** k4n0 has quit IRC | 13:13 | |
| lucasagomes | faizan, tagged to rc1 the fix seems to be quite simple | 13:14 |
| faizan | lucasagomes, yes, thanks for tagging. I will raise a review soon. | 13:17 |
| lucasagomes | faizan, cool :) | 13:17 |
| lucasagomes | thanks you for fixing it | 13:18 |
| lucasagomes | thank* | 13:18 |
| *** rloo has joined #openstack-ironic | 13:18 | |
| lazy_prince | can someone help me with this..? http://paste.openstack.org/show/111731/ | 13:22 |
| Shrews | lucasagomes: based on the latest emails to the ML, do you think we should still merge https://review.openstack.org/121050? | 13:24 |
| lucasagomes | Shrews, hi there, I answered that thread | 13:24 |
| lucasagomes | I think so | 13:24 |
| lucasagomes | I'm +1 for that... but I don't, what you guys think? | 13:24 |
| lucasagomes | I mostly asked for clarification in some points that Mike made | 13:24 |
| lucasagomes | the random hash for dicts | 13:24 |
| Shrews | lucasagomes: oh, diff thread... lemme read | 13:24 |
| lucasagomes | I don't think that's true, the performance as well | 13:25 |
| lucasagomes | Shrews, yeah | 13:25 |
| lucasagomes | Shrews, feel free to revoke the +a if you think otherwise | 13:25 |
| *** romcheg has quit IRC | 13:28 | |
| Shrews | lucasagomes: i'm personally fine with leaving it. and can easily be undone later if there are problems. | 13:28 |
| lucasagomes | Shrews, yeah | 13:28 |
| lucasagomes | Shrews, in fact, the problem he pointed about the random hash seed | 13:28 |
| lucasagomes | I think not generating the pyc even help with that, because then the pyc will be generated everything we load the modules | 13:29 |
| lucasagomes | (but that PYTHONHASHSEED wans't disabled in our tox ini, ofc) | 13:29 |
| lucasagomes | if that* | 13:29 |
| lucasagomes | (I mean we are not caching the pyc's in the filesystem) | 13:30 |
| * lucasagomes english level is going down hah needs more coffee (tough monday) | 13:31 | |
| *** teju has left #openstack-ironic | 13:45 | |
| *** r-daneel__ has joined #openstack-ironic | 13:48 | |
| *** yuanying has joined #openstack-ironic | 13:49 | |
| *** romcheg has joined #openstack-ironic | 13:56 | |
| *** ramineni has quit IRC | 13:58 | |
| *** jroll has quit IRC | 14:00 | |
| *** jroll has joined #openstack-ironic | 14:00 | |
| *** chenglch has quit IRC | 14:03 | |
| *** pcrews has joined #openstack-ironic | 14:10 | |
| jroll | morning ironic :) | 14:11 |
| * jroll fixes the gate on the whiteboard | 14:11 | |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Avoid recursing between keepalive and raw_command https://review.openstack.org/121568 | 14:13 |
| openstackgerrit | Syed Ismail Faizan Barmawer proposed a change to openstack/ironic: Fix code error in pxe_ilo driver https://review.openstack.org/121577 | 14:17 |
| *** yuanying_ has joined #openstack-ironic | 14:17 | |
| jroll | lazy_prince: hey, what driver are you using there? | 14:17 |
| lazy_prince | jroll: pxe_ipmitool | 14:18 |
| jroll | huh. | 14:18 |
| jroll | is that the node that failed to provision because those params weren't there? | 14:18 |
| lazy_prince | well.. i see this error when tear_down is called.. | 14:19 |
| *** chuckC has quit IRC | 14:19 | |
| lazy_prince | it is consistent in my env.. not sue why.. so thought of asking if my nodes are really missing any properties.. | 14:20 |
| jroll | tear_down is called on an active node or an errored node? | 14:20 |
| *** yuanying has quit IRC | 14:20 | |
| lazy_prince | errored node.. | 14:20 |
| jroll | aha | 14:20 |
| jroll | so, I thought we had solved this, might be a regression | 14:20 |
| * jroll looks at a thing | 14:20 | |
| lazy_prince | aha.. or is the patch not merged yet..? | 14:20 |
| lucasagomes | jroll, yo morning | 14:20 |
| lazy_prince | i can give the patch a try if needed.. | 14:21 |
| jroll | lazy_prince: I thought we fixed it months ago,s o | 14:21 |
| jroll | s/,s o/, so/ | 14:21 |
| *** rwsu has joined #openstack-ironic | 14:21 | |
| * jroll investigates | 14:21 | |
| lazy_prince | but i have the code pulled on friday.. | 14:22 |
| jroll | right... I'm saying we may have broken it again | 14:22 |
| *** yuanying_ has quit IRC | 14:22 | |
| lazy_prince | agreed.. :) | 14:22 |
| jroll | aha, hm | 14:22 |
| faizan | Hi Lucasagomes, I posted a patch for pxe_ilo bug https://review.openstack.org/121577 | 14:23 |
| lucasagomes | faizan, ack, thanks! | 14:23 |
| lucasagomes | will take a look later | 14:23 |
| lazy_prince | do you want me to file a bug for this..? | 14:23 |
| jroll | lazy_prince: yes, please do | 14:24 |
| faizan | lucasagomes, can you please review this patch when you get time? | 14:25 |
| NobodyCam | good mornign Ironic | 14:26 |
| NobodyCam | lucasagomes: THANK YOU!!!! | 14:26 |
| jroll | lazy_prince: the broken code is pretty obvious, just not sure how it got that way... it was in a refactoring, I think | 14:26 |
| lazy_prince | hmm.. | 14:27 |
| * jroll says good morning to an excited NobodyCam | 14:27 | |
| NobodyCam | morning jroll :) | 14:28 |
| jroll | what did lucas do? :) | 14:28 |
| NobodyCam | he pushed up a patch that I said I would do. but then I dropped off line | 14:28 |
| NobodyCam | :-p | 14:28 |
| jroll | oh ha | 14:29 |
| NobodyCam | so he rocks and I suck! | 14:29 |
| NobodyCam | :-p | 14:29 |
| *** faizan has quit IRC | 14:29 | |
| lucasagomes | NobodyCam, morning! | 14:31 |
| *** dtantsur|brb is now known as dtantsur | 14:31 | |
| lucasagomes | NobodyCam, what? hah | 14:31 |
| lucasagomes | oh | 14:31 |
| lucasagomes | no bothers, about the comment thing? | 14:31 |
| lucasagomes | relax | 14:31 |
| dtantsur | jroll, NobodyCam, morning | 14:32 |
| jroll | faiz<tab> :( | 14:33 |
| *** Isotopp_ is now known as Isotopp | 14:33 | |
| jroll | morning dtantsur :) | 14:33 |
| NobodyCam | :) mornign lucasagomes and dtantsur | 14:35 |
| NobodyCam | lucasagomes: I totaly forgot that I had a 8:00am apt to take the house in for breaks and new belts... | 14:35 |
| NobodyCam | so Thank you :) | 14:36 |
| lucasagomes | NobodyCam, no problem at all :) | 14:36 |
| NobodyCam | :) | 14:36 |
| lucasagomes | new breaks for the house that sounds cool haha | 14:37 |
| NobodyCam | lol well at least we can stop now :) hehehe. getting ready to head to the south | 14:37 |
| jroll | woohoo | 14:38 |
| lucasagomes | nice :D | 14:39 |
| *** ramineni has joined #openstack-ironic | 14:39 | |
| NobodyCam | yea :) its starting to get chilly up here in the north | 14:39 |
| * lucasagomes likes the cold | 14:40 | |
| NobodyCam | heheheh | 14:41 |
| Shrews | me too. difficult to snowboard w/o cold weather :) | 14:41 |
| NobodyCam | Shrews: thats what skate boards are for :) | 14:41 |
| * jroll needs to do more snowboarding this year | 14:41 | |
| NobodyCam | and morning Shrews | 14:42 |
| jroll | dtantsur: about the IPMIFailure thing | 14:42 |
| jroll | there's a fairly large comment already in PXEDeploy.deploy() | 14:43 |
| Shrews | jroll: i'm thinking the next mid-cycle should be in CO or UT ;) | 14:43 |
| jroll | about why this happens etc etc | 14:43 |
| jroll | Shrews++ | 14:43 |
| jroll | or even tahoe | 14:43 |
| dtantsur | jroll, then try..except should be there | 14:43 |
| jroll | dtantsur: it is | 14:43 |
| jroll | dtantsur: and it re-raises | 14:43 |
| dtantsur | then why we need this patch? | 14:43 |
| dtantsur | then it should not reraise :) | 14:43 |
| devananda | morning, all | 14:43 |
| jroll | err, hmm | 14:43 |
| dtantsur | devananda, morning | 14:43 |
| jroll | it should only re-raise if it's not uefi | 14:44 |
| jroll | the only expected failure is if it is uefi | 14:44 |
| jroll | so yeah, you're right | 14:44 |
| jroll | morning devananda | 14:44 |
| dtantsur | jroll, then let's put the check _there_ | 14:44 |
| lucasagomes | devananda, morning | 14:44 |
| NobodyCam | morning devananda | 14:44 |
| NobodyCam | Shrews: http://en.wikipedia.org/wiki/Durango,_Colorado | 14:44 |
| jroll | dtantsur: it is, I think that code is fine as-is, without faizan's patch | 14:44 |
| dtantsur | jroll, otherwise it's too easy to break things in the future by adding something after IPMI call | 14:44 |
| dtantsur | heh, now understood :) | 14:45 |
| jroll | dtantsur: jfyi https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/pxe.py#L333-342 | 14:45 |
| jroll | dtantsur: though, I suspect faizan filed this bug after running into this in real life, ipmi is weird, so maybe it's a real thing | 14:46 |
| dtantsur | jroll, then we need to fix this code, not invent something really strange as to me | 14:47 |
| devananda | lucasagomes: on https://review.openstack.org/#/c/121397/ I agree with sdague - the nova unit tests didn't catch the problem, and should have | 14:47 |
| dtantsur | and I'm still waiting for details on why and how this bug happens | 14:47 |
| jroll | dtantsur: sure | 14:47 |
| lucasagomes | devananda, well... we had to change unittests as part of the fix | 14:47 |
| lucasagomes | devananda, I mean unittests would fail, but Ironic is non-voting in nova gate | 14:47 |
| devananda | lucasagomes: that's the problem | 14:47 |
| lucasagomes | so it failed there, but it was approved anyway | 14:48 |
| devananda | lucasagomes: no - the unit tests are in nova's tree and considered voting | 14:48 |
| jroll | yeah, I was surprised unit tests didn't fail | 14:48 |
| devananda | lucasagomes: the -tempest- tests are a separate job and non-voting | 14:48 |
| lucasagomes | oh | 14:48 |
| lucasagomes | maybe I overlooked then? | 14:48 |
| lucasagomes | hmm | 14:48 |
| lucasagomes | alright I will try to add a test to it | 14:48 |
| devananda | unit tests didn't fail because they mock calling the thing, rather than actually calling it | 14:48 |
| devananda | the mocked call was still passing 2 params, so it passed | 14:48 |
| lucasagomes | oh, that's bad | 14:49 |
| jroll | devananda: I don't see it being mocked, where is that? | 14:49 |
| jroll | I see self.host_manager = ironic_host_manager.IronicHostManager() | 14:49 |
| devananda | sorry, its early - i don't mean mocked as in the library | 14:50 |
| devananda | i mean, the unit tests are not checking that the actual nova code can instantiate and call all the classes of HostStates | 14:50 |
| devananda | there are unit tests which directly call host.consume_from_instance | 14:51 |
| devananda | eg, fake calls | 14:51 |
| devananda | note how both the original patch and the fix had to do this in several places | 14:51 |
| jroll | I'm not sure I understand, but https://review.openstack.org/#/c/121397/1/nova/tests/scheduler/test_ironic_host_manager.py | 14:51 |
| devananda | - host.consume_from_instance(instance) | 14:51 |
| devananda | + host.consume_from_instance(context, instance) | 14:51 |
| jroll | the tests had to be updated to pass | 14:51 |
| jroll | so why didn't they fail before? | 14:51 |
| devananda | because no unit tests outside of nova/tests/scheduler/test_ironic_host_manager.py is instantiating /nova/scheduler/ironic_host_manager.py's IronicNodeState | 14:52 |
| lucasagomes | I think it's because we are testing IronicManager directly in our tests... where we need a tests, not related to ironic | 14:52 |
| devananda | right | 14:52 |
| lucasagomes | to actually go and test all the possible hostmanagers in tree | 14:52 |
| devananda | we need a test that walks through all host managers, and checks that they all adhere to the common interface | 14:53 |
| lucasagomes | yeah | 14:53 |
| lucasagomes | I see now | 14:53 |
| devananda | akin to https://github.com/openstack/nova/blob/master/nova/tests/virt/test_driver.py#L63 | 14:53 |
| *** dtantsur_ has joined #openstack-ironic | 14:53 | |
| *** dtantsur has quit IRC | 14:53 | |
| *** dtantsur_ is now known as dtantsur | 14:53 | |
| lucasagomes | lemme take a look at it | 14:53 |
| *** ramineni has quit IRC | 14:54 | |
| jroll | I still don't understand why that needs to be outside of test_ironic_host_manager.py | 14:54 |
| jroll | host = ironic_host_manager.IronicNodeState("fakehost", "fakenode"); host.consume_from_instance('fake-context', instance) | 14:54 |
| jroll | consume_from_instance calls super(...).consume_from_instance() | 14:55 |
| devananda | jroll: that isn't testing whether the code which actually uses that object will call it that way | 14:55 |
| jroll | which presumably has 2 required parameters | 14:55 |
| lucasagomes | jroll, because the IronicHostManager is just an abstraction... let's say that conductor/manager.py calls instantiate the HostManager class specified by CONF.host_manager | 14:56 |
| jroll | right, I get that part | 14:56 |
| lucasagomes | and then call class.consume_from_instance(ctx, instance, new_parameter) | 14:56 |
| jroll | I get that we need tests for that | 14:56 |
| lucasagomes | if we test Ironic direct we won't detect that a new parameter was added there | 14:56 |
| jroll | I don't understand why what I posted up there passes | 14:56 |
| jroll | dunno, might just need more coffee | 14:56 |
| *** linggao has joined #openstack-ironic | 14:58 | |
| devananda | jroll: https://review.openstack.org/#/c/118391/22/nova/scheduler/filter_scheduler.py | 14:58 |
| *** lazy_prince has quit IRC | 14:58 | |
| devananda | jroll: because Nova doesn't call it directly -- the filter_scheduler is passed an object, which it assumes has a given interface. | 14:58 |
| devananda | jroll: the unit tests are asserting what that interface is only locally -- they're not comparing the -actual- call to the interface | 14:59 |
| *** bluex-pl has quit IRC | 14:59 | |
| devananda | jroll: and the unit test for the filter_scheduler is not passing in an ironic host manager | 14:59 |
| devananda | https://review.openstack.org/#/c/118391/22/nova/tests/scheduler/test_filter_scheduler.py | 15:00 |
| jroll | oh, I didn't bubble up enough | 15:00 |
| jroll | BaseBaremetalNodeState doesn't call supeR() | 15:00 |
| devananda | ah | 15:00 |
| * jroll is content, thanks for your patience :) | 15:00 | |
| devananda | ok - why doesn't that call super? | 15:01 |
| devananda | actually | 15:01 |
| jroll | right, that's a different question | 15:01 |
| jroll | oh, it makes sense, though | 15:01 |
| devananda | yea, that class is there to deduplicate code between bareemtal and irnoic | 15:01 |
| jroll | right | 15:01 |
| devananda | hah. so baremetal's hsot manager just broke too | 15:01 |
| devananda | and we didn't fix it | 15:02 |
| jroll | but that method doesn't call the super because the super behavior doesn't make sense for us | 15:02 |
| jroll | heh | 15:02 |
| jroll | baremetal is dead, long live ironic | 15:02 |
| NobodyCam | jroll: ++ | 15:02 |
| * jroll is a bad person :P | 15:02 | |
| devananda | ok - nope, it's fine. its subclass just passes | 15:03 |
| * devananda only cares because of the upgrade path | 15:03 | |
| *** Poornima_ has quit IRC | 15:04 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Avoid recursing between keepalive and raw_command https://review.openstack.org/121568 | 15:04 |
| NobodyCam | gah the corp vpn is so slow tody | 15:04 |
| *** ChuckC_ has joined #openstack-ironic | 15:04 | |
| *** ChuckC_ has quit IRC | 15:05 | |
| *** ChuckC_ has joined #openstack-ironic | 15:05 | |
| *** rushiagr is now known as rushiagr_away | 15:05 | |
| *** chuckC has joined #openstack-ironic | 15:07 | |
| devananda | speaking of upgrade path... adam_g, how's that going? | 15:07 |
| * devananda looks at the UEFI spec, marks it completed | 15:07 | |
| * devananda updates meeting wiki | 15:09 | |
| devananda | lucasagomes: mind if I cover rc status today? | 15:09 |
| lucasagomes | devananda, not at all | 15:09 |
| *** ChuckC_ has quit IRC | 15:10 | |
| *** bluex-pl has joined #openstack-ironic | 15:13 | |
| devananda | i'm going to add a section to the meeting for each driver | 15:17 |
| devananda | also, IBM has third party CI up, so I want to highlight that | 15:17 |
| NobodyCam | oh | 15:17 |
| NobodyCam | neat | 15:17 |
| *** igordcard has quit IRC | 15:18 | |
| *** yuanying has joined #openstack-ironic | 15:19 | |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Update ironic states and documentation https://review.openstack.org/118467 | 15:23 |
| *** ramineni has joined #openstack-ironic | 15:23 | |
| ramineni | jroll, jayF : Hi | 15:24 |
| ramineni | jroll, jayF : raised a review for the script https://review.openstack.org/#/c/121433/ , please have a look | 15:25 |
| *** viktors is now known as viktors|afk | 15:26 | |
| devananda | NobodyCam: grats on landing the baremetal->ironic proxy in nova, btw | 15:26 |
| NobodyCam | :-p | 15:26 |
| NobodyCam | I thank all who helped with that one :) | 15:27 |
| NobodyCam | I just updated the date of the next call :-p | 15:31 |
| jroll | ramineni: neat! one question: is any of that specific to the agent? or can that be used for any kernel/ramdisk? | 15:31 |
| devananda | NobodyCam: oh ty | 15:31 |
| NobodyCam | :-p | 15:31 |
| *** ChuckC_ has joined #openstack-ironic | 15:32 | |
| ramineni | jroll : can be used for any kernel and ramdisk | 15:32 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Update ironic states and documentation https://review.openstack.org/118467 | 15:32 |
| jroll | ramineni: hmm, then maybe there is a better place for this, idk. people using pxe/ilo driver would need this script, too, yes? curious what devananda thinks, I almost think this should be a DIB thing | 15:33 |
| ramineni | yes , this is used by ilo iscsi and iloipa | 15:35 |
| openstackgerrit | Jarrod Johnson proposed a change to stackforge/pyghmi: Avoid exception on close https://review.openstack.org/121602 | 15:37 |
| jroll | ramineni: right, so I think maybe IPA repo isn't the best place for that script | 15:37 |
| jroll | also curious to hear JayF's opinion | 15:37 |
| jroll | devananda: I updated https://etherpad.openstack.org/p/ipa-todos ; I think it's accurate now | 15:38 |
| devananda | jroll: great, ty | 15:38 |
| jroll | np | 15:38 |
| devananda | adam_g ran into an issue on friday, doing a sideways migration of a live baremetal instance -- the image ref isn't part of the nova_bm schema, so we aren't copying it to Ironic right now | 15:39 |
| devananda | resulting in an undeletable instance | 15:39 |
| devananda | because the PXE driver errors out if it can't find the image ref in node.instance_info | 15:39 |
| ramineni | jroll : ok | 15:39 |
| jroll | devananda: where's the image ref in nova-bm? instances table or something? | 15:40 |
| devananda | jroll: yea. our migration is only pulling from `nova_bm`, not `nova` | 15:40 |
| jroll | bah | 15:40 |
| devananda | yup | 15:40 |
| devananda | bah | 15:40 |
| jroll | doesn't seem like a hard problem to fix though, unless someone has a requirement that we don't touch the nova side of things | 15:41 |
| devananda | either we import the full `nova` models as well, or we teach the PXE driver how to handle missing info during deprovisioning | 15:41 |
| devananda | I think #2 is better, since it's actually fixing something | 15:41 |
| jroll | oh, lazy_prince ran into the same thing this morning | 15:41 |
| jroll | it actually fails in clean_up, right? | 15:41 |
| devananda | is there an open bug? | 15:41 |
| devananda | yes | 15:41 |
| jroll | I think that was actually a regression in the last couple weeks | 15:41 |
| jroll | he said he would open one, I didn't see it | 15:42 |
| devananda | ah - https://bugs.launchpad.net/ironic/+bug/1368984 | 15:42 |
| jroll | devananda: https://bugs.launchpad.net/ironic/+bug/1369587 | 15:42 |
| jroll | oh | 15:42 |
| jroll | same bug, different scope :) | 15:43 |
| devananda | yep | 15:43 |
| devananda | updated https://bugs.launchpad.net/ironic/+bug/1368984 and promoted to critical | 15:46 |
| devananda | that, IMO, really blocks our release | 15:46 |
| devananda | and given lazy_prince's bug, it sounsd like Nova is tripping on it some times too | 15:46 |
| jroll | agree | 15:47 |
| jroll | the code it's failing on is only there to clean up the image cache | 15:47 |
| devananda | if there's no cached image, then who cares if it doesn't have that info? | 15:48 |
| NobodyCam | devananda: +1 | 15:48 |
| devananda | fix should be some try: except: pass bits in the right place | 15:48 |
| jroll | right | 15:48 |
| jroll | devananda: also, I'm fairly certain this was in j3 | 15:49 |
| devananda | jroll: i actually dont know why this hasn't been a bug for months | 15:50 |
| jroll | eh? I think it's a regression related to the refactor to land ilo/pxe | 15:50 |
| jroll | or some other cleanup code around the same time | 15:51 |
| devananda | oh | 15:51 |
| devananda | could be | 15:51 |
| jroll | because we used to have the same problem when calling deploy.validate() before tear_down | 15:51 |
| jroll | we fixed that by just calling power.validate() | 15:52 |
| NobodyCam | jroll: thats the issue I recalled | 15:52 |
| jroll | right, so now clean_up ends up validating instance_info, it's unfortunate | 15:52 |
| jroll | I thought we were catching that exception, dunno | 15:53 |
| * jroll looks | 15:53 | |
| jroll | might just make a patch real quick | 15:53 |
| JayF | ramineni: that script doesn't even use DIB, does it? | 15:53 |
| *** yuanying has quit IRC | 15:53 | |
| ramineni | JayF: no , it doesn't , it's a normal shell script | 15:56 |
| JayF | ramineni: Are you intending to use the ramdisk and kernel that the CoreOS scripts create? | 15:56 |
| JayF | ramineni: or something DIB creates (that doesn't exist yet) | 15:57 |
| ramineni | JayF : yes , for ilo-ipa driver , we use ramdisk/kernel CoreOS scripts create | 15:58 |
| JayF | ramineni: then I'd suggest just putting that script in imagebuild/coreos and adding a an "iso" target to the Makefile. | 15:58 |
| JayF | ramineni: that uses your script to do the whole build steps and create it | 15:58 |
| JayF | ramineni: since you're not utilizing DIB at all | 15:59 |
| JayF | jroll: ^ wdyt? | 15:59 |
| jroll | I'm ok with that, specifically | 16:00 |
| jroll | but | 16:00 |
| jroll | seems like this script will also be used by ilo/pxe deployers | 16:00 |
| jroll | do we want them to have to pull IPA down to do so? | 16:00 |
| JayF | Oh, that's a really good point | 16:00 |
| jroll | or duplicate code between here and DIB or whatever | 16:00 |
| jroll | that's my primary concern | 16:00 |
| JayF | Hmm. | 16:00 |
| JayF | It seems like a script that turns an arbitrary ramdisk and kernel into an iso is generally useful | 16:01 |
| JayF | but I don't know where it should go, unless the script itself should be integrated into DIB... but personally I like the simplicity of it being separate | 16:01 |
| *** dtantsur is now known as dtantsur|bbl | 16:01 | |
| jroll | right... | 16:02 |
| jroll | maybe ironic/tools | 16:02 |
| *** eghobo has joined #openstack-ironic | 16:02 | |
| *** rushiagr_away is now known as rushiagr | 16:05 | |
| NobodyCam | are the colors correct on the Items for graduation spread sheet | 16:08 |
| *** marcoemorais has joined #openstack-ironic | 16:10 | |
| *** killer_prince has joined #openstack-ironic | 16:11 | |
| *** killer_prince is now known as lazy_prince | 16:11 | |
| *** teju has joined #openstack-ironic | 16:13 | |
| openstackgerrit | A change was merged to stackforge/pyghmi: Avoid exception on close https://review.openstack.org/121602 | 16:14 |
| *** jcoufal has quit IRC | 16:15 | |
| devananda | lucasagomes: not sure if I already pinged you on this or not .. Sept 18th is the requested cut-off date for a Juno final release of our client | 16:15 |
| devananda | lucasagomes: so any patches that need to land in the client, eg. t oexpose features we added in Juno -- we need to try to land them now | 16:15 |
| lucasagomes | devananda, lemme look at the cli queue | 16:16 |
| lucasagomes | devananda, we merged a couple of good patches last week | 16:16 |
| devananda | this is to give downstream packaging (deb/fedora) time to freeze all their dependencies | 16:16 |
| lucasagomes | for listing resources with details etc | 16:16 |
| devananda | and test the RC builds of the service packages | 16:16 |
| *** ndipanov is now known as ndipanov_gone | 16:17 | |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Allow clean_up with missing image ref https://review.openstack.org/121615 | 16:17 |
| jroll | devananda: ^ | 16:17 |
| *** jcoufal has joined #openstack-ironic | 16:18 | |
| lucasagomes | devananda, alright there's one patch fixing a bug, it has a +2 already I will review it | 16:18 |
| jroll | devananda: haven't written unit tests yet, but I think this will fix the issue alright | 16:18 |
| lucasagomes | devananda, then I think we are fine cutting a release | 16:18 |
| NobodyCam | jroll: that test seems a little lacking in code :-p | 16:19 |
| devananda | lucasagomes: awesome, ty | 16:19 |
| *** ifarkas has quit IRC | 16:19 | |
| NobodyCam | lol | 16:19 |
| jroll | :P | 16:19 |
| devananda | lucasagomes: oh. there's a fix from rakesh for OS_REGION_NAME that looks important | 16:20 |
| lucasagomes | devananda, yup that's the one | 16:20 |
| devananda | :) | 16:21 |
| jroll | added a WIP to that fix https://review.openstack.org/#/c/121615/ | 16:21 |
| ramineni | jroll : looks like no better place , if we dont want to duplicate the script , JayF: any better alternative you have in mind? | 16:21 |
| jroll | folks are welcome to pick this up | 16:21 |
| jroll | ramineni: I would suggest ironic/tools/ but idk how others feel about that | 16:22 |
| JayF | ramineni: I've been thinking about it and don't have much to suggest... | 16:22 |
| JayF | jroll: I dislike putting it in ironic/ repo, because then you have an IPA->Ironic build dep | 16:22 |
| JayF | jroll: assuming you're building an IPA ISO | 16:22 |
| jroll | guh | 16:22 |
| JayF | jroll: WTB ironic-common with common code between IPA and Ironic | 16:23 |
| jroll | that said, ironic is the most common place between DIB and IPA for this | 16:23 |
| jroll | ewwwwww | 16:23 |
| jroll | well, maybe that's not so bad | 16:23 |
| jroll | dunno | 16:23 |
| JayF | duplicating the code is also not an awful solution | 16:23 |
| JayF | just create a dib element that fully makes a pxe ramdisk, integrate the iso builder script into coreos build to add an iso make target | 16:23 |
| jroll | yeah, perhaps | 16:24 |
| JayF | I honestly think duplicating the code may be the lesser evil in this case | 16:24 |
| jroll | I do love the 'make iso' idea | 16:24 |
| devananda | JayF: duplicate it for now. if that becomes a problem at some point, I'm not opposed to considering a separate repo for shared things | 16:24 |
| JayF | ++ | 16:24 |
| *** jistr has quit IRC | 16:25 | |
| JayF | ramineni: so for now, for IPA, lets integrate your iso-builder script into imagebuild/coreos, and add a Makefile target for iso (make iso). I can add the ISO to the things we upload in our post job. | 16:25 |
| devananda | JayF: also, what about splitting the image buildign bits out of IPA's repo into their own ipa-image-build repo? | 16:25 |
| JayF | devananda: I don't hate that at all. That's more like the model we use downstream (we have our own image build repo that embeds hw utilities) | 16:26 |
| devananda | make IPA just the agent code, not the image bits, so that there is a contract and unidirectional dep | 16:26 |
| devananda | which also follows the model that ironic has with dib | 16:26 |
| JayF | devananda: but we still have the problem of ramineni having something that needs to both be in dib and the ipa builder (a script that makes an iso out of an arbitrary kernel and ramdisk) | 16:26 |
| lucasagomes | devananda, patch is in gate now, once that merges I think we should go and cut the release | 16:27 |
| devananda | lucasagomes: sweet | 16:27 |
| devananda | JayF: oh. gotcha. | 16:27 |
| NobodyCam | lucasagomes: awesome!!! | 16:27 |
| JayF | devananda: the other problem with splitting out the image builder is it'll make our post job more complex (to build the image) | 16:27 |
| jroll | JayF: ipa-builder could have a second script to wrap DIB or something | 16:27 |
| devananda | JayF: sounds like it's really a separate tool then | 16:27 |
| JayF | jroll: but at that point it's really more of ironic-images than ipa-builder | 16:27 |
| devananda | "take $think and make $ISO" | 16:27 |
| jroll | bah, true | 16:27 |
| JayF | devananda: yeah, that's why I'm stuck as to what to do with it. I don't want to say "toss this in it's own repo and we'll all depend on it" because it seems overkill for a script :) | 16:28 |
| jroll | devananda: I agree, but a repo for a single script seems overkill :P | 16:28 |
| devananda | ok - so ironic probably needs a repo for various image build tools | 16:28 |
| devananda | that would then wrap the bits in dib, ipa, and this one | 16:28 |
| NobodyCam | devananda: not just the tools directory in our current repo | 16:28 |
| JayF | devananda: would the dib elements to build the pxe ramdisk move there, too? | 16:28 |
| devananda | and an operator would look at ironic, choose a driver, and then choose the appropriate script to build images for that driver from this new repo | 16:28 |
| devananda | JayF: yes | 16:29 |
| jroll | ++ that ain't a bad idea | 16:29 |
| devananda | JayF: or a shim that calls them | 16:29 |
| JayF | +9001 | 16:29 |
| JayF | well, I don't like a shim | 16:29 |
| JayF | I like the actual elements living there :) | 16:29 |
| devananda | JayF: I agree. I'd rather the actual things live there. will make gating easier | 16:29 |
| JayF | The only thing then is to figure out the proper coordination to keep the IPA builder job going | 16:29 |
| devananda | because then we can model the dependencies more linearly | 16:29 |
| JayF | I assume it's possible in infra to have one project trigger another project to build, I just don't know how yet | 16:30 |
| devananda | post jobs | 16:30 |
| JayF | This would be a great topic for the meeting fwiw | 16:30 |
| devananda | JayF: I have to run out for a bit, meetings n crap, but please add it to https://docs.google.com/spreadsheets/d/1XBKdeDeGfaRYaThjIIoYRwe_zPensECnxsKUuqdoVmQ/edit#gid=0 | 16:30 |
| jroll | wow, it's 9:30 already | 16:30 |
| JayF | Post job is how it's done now :) what the job looks like is more wbat I'm curious about | 16:30 |
| JayF | will-do | 16:30 |
| devananda | I'll be back for our meeting | 16:30 |
| * jroll goes afk | 16:30 | |
| *** dlaube has joined #openstack-ironic | 16:30 | |
| NobodyCam | I can a seperate repo leading to more corss project dependency issues, at least some | 16:30 |
| JayF | devananda: can I add proposed summit idea to that spreadsheet? | 16:31 |
| jroll | we need a whiteboard | 16:31 |
| jroll | to draw this dep graph | 16:31 |
| *** eghobo has quit IRC | 16:31 | |
| devananda | JayF: anyone can edit | 16:32 |
| * jroll really goes afk | 16:32 | |
| *** chuckC has quit IRC | 16:32 | |
| devananda | JayF: i'll bring it up in our meeting, but yes -- please do | 16:32 |
| JayF | aight | 16:32 |
| * devananda really goes afk too | 16:32 | |
| NobodyCam | with everyone going afk, /me thinks about a starbucks run | 16:33 |
| JayF | ramineni: to get this landed now; for IPA iso, I'd just integrate it with the imagebuild/coreos builder... same with the pxe ramdisk, I'd just integrate it where it lives now and make it work | 16:33 |
| JayF | ramineni: and we'll figure out a better permanent home for all of this | 16:33 |
| * lucasagomes lucas-afk | 16:33 | |
| lucasagomes | >.< nick :( | 16:33 |
| *** lucasagomes is now known as lucas-afk | 16:34 | |
| NobodyCam | lol | 16:34 |
| *** marcoemorais has quit IRC | 16:34 | |
| *** marcoemorais has joined #openstack-ironic | 16:34 | |
| ramineni | JayF : you meant , make iso idea right? | 16:34 |
| JayF | ramineni: aye | 16:34 |
| JayF | ramineni: just in the same folder as existing ramdisk :) | 16:35 |
| ramineni | JayF: sure , I will change that , and add iso target to make file | 16:35 |
| ramineni | JayF: thanks :) , will raise review again | 16:36 |
| *** eghobo has joined #openstack-ironic | 16:37 | |
| *** marcoemorais has quit IRC | 16:38 | |
| NobodyCam | brb making the StarBucks run | 16:38 |
| *** marcoemorais has joined #openstack-ironic | 16:38 | |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Add unit tests for "ironic node-delete" shell cmd https://review.openstack.org/120710 | 16:41 |
| *** ramineni has quit IRC | 16:46 | |
| openstackgerrit | A change was merged to openstack/ironic: Updated from global requirements https://review.openstack.org/120637 | 16:46 |
| *** yuanying has joined #openstack-ironic | 16:50 | |
| openstackgerrit | Mike Durnosvistov proposed a change to openstack/ironic: Check whether specified FS is supported https://review.openstack.org/98102 | 16:50 |
| *** ChuckC_ is now known as ChuckC | 16:58 | |
| *** derekh has quit IRC | 16:59 | |
| *** marcoemorais1 has joined #openstack-ironic | 16:59 | |
| *** teju has left #openstack-ironic | 16:59 | |
| * devananda is semi-back | 17:02 | |
| *** marcoemorais has quit IRC | 17:03 | |
| *** harlowja has joined #openstack-ironic | 17:04 | |
| ChuckC | adam_g: hi | 17:11 |
| *** athomas has quit IRC | 17:15 | |
| *** romcheg has quit IRC | 17:16 | |
| *** eghobo has quit IRC | 17:16 | |
| *** cdnchris has joined #openstack-ironic | 17:17 | |
| *** marcoemorais has joined #openstack-ironic | 17:18 | |
| *** marcoemorais1 has quit IRC | 17:18 | |
| devananda | JayF / jroll: are ya'll releasing IPA as a client, eg. with semver? or does that need to get tagged and packaged up as part of releasing Juno? | 17:20 |
| devananda | I really should have asked that a while ago :) | 17:20 |
| JayF | devananda: we were asking that while you were gone | 17:21 |
| JayF | devananda: I think we all pretty well decided it makes sense to cut a 'Juno' of IPA | 17:21 |
| *** cdnchris has left #openstack-ironic | 17:21 | |
| JayF | jroll: ^ a good topic for IPA subteam report | 17:21 |
| *** eghobo has joined #openstack-ironic | 17:22 | |
| *** yuanying has quit IRC | 17:23 | |
| *** dutsmoc is now known as comstud | 17:24 | |
| devananda | JayF: so that should perhaps be done by the openstack release mgr, ttx, except he's not aware of it | 17:25 |
| devananda | JayF: and IPA hasn't followed the release cadence | 17:25 |
| devananda | eg, feature freeze, gate tests, etc | 17:25 |
| JayF | We | 17:27 |
| JayF | *We've been trying to get gate tests, hehe | 17:27 |
| devananda | *nod* | 17:32 |
| devananda | Jayf: is there a build pulished on pypi somewhere? | 17:33 |
| devananda | I dont see anything | 17:33 |
| devananda | *published | 17:34 |
| JayF | No, we don't publish anything except the ramdisks on tarballs.openstack | 17:35 |
| JayF | or at least haven't yet | 17:35 |
| JayF | at one point I was against the idea of a pypi release of IPA... now I could see a user wanting to make their own custom image build depend on a given version, so I'm more OK with it in the context of that use case | 17:35 |
| JayF | it's just strange, IPA feels in some ways like iPxe or Syslinux | 17:35 |
| JayF | where the package installed gives you binary images to do what you need to with | 17:36 |
| JayF | rather than if you made an IPA package that installed the python locally on your system | 17:36 |
| JayF | if you get what I mean? | 17:36 |
| NobodyCam | devananda: thank you for the edits to https://wiki.openstack.org/wiki/Ironic/NovaBaremetalIronicMigration :) | 17:37 |
| *** penick has joined #openstack-ironic | 17:39 | |
| NobodyCam | I'm drawing a blank on the best way to fill out this section: Ensure service start-up | 17:40 |
| NobodyCam | check running / check port open / check log ? | 17:41 |
| JayF | Check logs for no WARNING or ERROR mssages? | 17:41 |
| JayF | what service are you talking about in that case specifically | 17:41 |
| * JayF hasn't opened that page | 17:41 | |
| NobodyCam | lol for this case it is the ironic-api / conductor | 17:41 |
| NobodyCam | servicces | 17:41 |
| JayF | Then how about an api call to /drivers | 17:41 |
| JayF | if it succeeds; api works | 17:41 |
| JayF | if conductor started; it'll be listed | 17:41 |
| JayF | and check logs for warnings and errors | 17:42 |
| NobodyCam | oh that would work :) | 17:42 |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Update README with a bit more info https://review.openstack.org/121637 | 17:42 |
| NobodyCam | rloo: is that white space for rst formating? | 17:50 |
| * devananda goes afk again for 20 minutes | 17:51 | |
| NobodyCam | :) | 17:52 |
| *** bluex-pl has quit IRC | 17:53 | |
| openstackgerrit | A change was merged to openstack/ironic: Don't write python bytecode while testing https://review.openstack.org/121050 | 17:53 |
| rloo | hi NobodyCam. Ouch, did I leave some white space? | 17:55 |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Update README with a bit more info https://review.openstack.org/121637 | 17:56 |
| rloo | NobodyCam: ^^ what white space? :-) | 17:56 |
| NobodyCam | morning rloo :) | 17:56 |
| NobodyCam | lol | 17:56 |
| rloo | bonjour NobodyCam | 17:56 |
| NobodyCam | marci bow coupe | 17:57 |
| rloo | NobodyCam: if you can think of something better to put in the README, let me know. I was lame. copy/pasted stuff from other clients. | 17:57 |
| NobodyCam | rloo: ack will take a better look in a few | 17:57 |
| rloo | NobodyCam: wrt https://wiki.openstack.org/wiki/Ironic/NovaBaremetalIronicMigration, are we at the 'polishing' stage, or need more info added? | 17:58 |
| *** penick has quit IRC | 18:00 | |
| NobodyCam | rloo: I'm adding detail to the ensure service startup section atm | 18:00 |
| rloo | NobodyCam: and how much detail do we need to provide? eg, do/should we point them to 'install ironic' docn? | 18:01 |
| *** penick has joined #openstack-ironic | 18:01 | |
| *** marcoemorais has quit IRC | 18:02 | |
| *** marcoemorais has joined #openstack-ironic | 18:02 | |
| *** igordcard has joined #openstack-ironic | 18:02 | |
| rloo | NobodyCam: maybe in the Assumptions part, we can mention assuming that they have baremetal running and ironic installed. | 18:02 |
| *** marcoemorais has quit IRC | 18:03 | |
| *** rushiagr is now known as rushiagr_away | 18:03 | |
| *** marcoemorais has joined #openstack-ironic | 18:03 | |
| *** cdnchris has joined #openstack-ironic | 18:05 | |
| NobodyCam | Humm I thouhgt I had somehting about that already | 18:07 |
| rloo | NobodyCam: after doing the 'ironic-nova-bm-migrate', will ironic node-list show the nodes migrated from BM? | 18:09 |
| JayF | jroll: heh | 18:12 |
| JayF | jroll: gmta? | 18:12 |
| *** romcheg has joined #openstack-ironic | 18:13 | |
| rloo | NobodyCam: wrt 'Ensure service start-up', if they had installed ironic, they should have ensured it worked then. | 18:15 |
| rloo | NobodyCam: http://docs.openstack.org/developer/ironic/deploy/install-guide.html doesn't give examples of testing that ironic is up and running :-( | 18:15 |
| *** cdnchris has quit IRC | 18:15 | |
| rloo | NobodyCam: it seems to me that in the migration doc, that if they migration script worked, they should be able to do an 'ironic node-list' and see the nodes that had been in baremetal. That might be a more useful test at that point in the doc. | 18:16 |
| *** mitz_ has quit IRC | 18:16 | |
| *** jistr has joined #openstack-ironic | 18:18 | |
| NobodyCam | rloo: yes that would be good to add | 18:18 |
| JayF | https://review.openstack.org/#/c/120660/4 and https://review.openstack.org/#/c/120661/4 are pretty trivial docstring reviews if someone wants to land some easy patches | 18:19 |
| *** yuanying has joined #openstack-ironic | 18:19 | |
| *** cdnchris has joined #openstack-ironic | 18:21 | |
| jroll | JayF: indeed :P | 18:25 |
| NobodyCam | rloo: hows that look RE: https://wiki.openstack.org/wiki/Ironic/NovaBaremetalIronicMigration#Ensure_service_start-up | 18:28 |
| rloo | sec just giving feedback on an easy patch... | 18:29 |
| jroll | NobodyCam: you might add that the "hosts" list depends on your conductor hostname | 18:31 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: ironic client to use os_region_name if provided https://review.openstack.org/120807 | 18:31 |
| rloo | NobodyCam: much better. So how much baby sitting do we do. Eg, do we mention that 'instance_uuid' will have a value if their bm node has an instance on it? (do we migrate those?) | 18:31 |
| *** openstackgerrit has quit IRC | 18:32 | |
| rloo | NobodyCam: do you know, are the create/update dates since they were added to ironic, or to bm? | 18:32 |
| jroll | that would be since added to ironic, normally | 18:33 |
| jroll | don't think we expose those to the api | 18:33 |
| *** openstackgerrit has joined #openstack-ironic | 18:33 | |
| rloo | jroll: i was wondering if bm had create/update dates, and if so, whether it made sense to 'migrate' those dates over. | 18:34 |
| ChuckC | adam_g: hi | 18:34 |
| rloo | jroll: esp if those dates had anything to do with dates in nova wrt any instances on those nodes. | 18:34 |
| jroll | rloo: yeah, /me looks | 18:34 |
| jroll | well | 18:34 |
| jroll | created_at is when the node was reg'd, updated_at is the last time the db row was updated | 18:35 |
| rloo | jroll: i suppose i should just look at the migration script | 18:35 |
| jroll | yeah, that's what I was going to do | 18:35 |
| rloo | ha ha. you first then ;) | 18:35 |
| jroll | rloo: https://github.com/openstack/ironic/blob/master/ironic/migrate_nova/migrate_db.py#L104-105 :) | 18:36 |
| rloo | cool! | 18:36 |
| rloo | so that is something we can mention in the migration doc | 18:36 |
| jroll | yep :) | 18:36 |
| jroll | though ironic might end up changing updated_at before the user does node-show | 18:37 |
| *** penick has quit IRC | 18:37 | |
| rloo | jroll: Hmm. Why is i_node.updated_at = n_node.created_at, and vice versa? | 18:37 |
| Shrews | rloo: umm, hrmm.... | 18:38 |
| Shrews | me thinks that to be erroneous | 18:39 |
| NobodyCam | was just about to sy because it a unfiled bug :-p | 18:39 |
| rloo | ok, who wants to file/fix? (I can, just thought I'd offer it to someone else) | 18:40 |
| Shrews | rloo: i'll grab it | 18:40 |
| rloo | low-hanging-fruit. c'mon, don't be shy! | 18:40 |
| rloo | and the winner is Shrews! Thank you! | 18:40 |
| NobodyCam | :) | 18:41 |
| adam_g | ChuckC, heya | 18:41 |
| ChuckC | adam_g: hey, I was wondering if I could add you to a neutron review for an ironic issue? | 18:42 |
| adam_g | ChuckC, sure | 18:42 |
| jroll | rloo: lolol, nice catch | 18:42 |
| jroll | ChuckC: I'm curious to see that review as well :) | 18:42 |
| rloo | jroll: ;) | 18:43 |
| jroll | rloo: I think we all wish we had your level of observant-ness | 18:43 |
| ChuckC | adam_g: it's about making mac_address updatable, so ironic servers can replace a NIC and the neutron port doesn't need to be deleted/recreated | 18:43 |
| ChuckC | jroll: no problem! | 18:43 |
| rloo | jroll: you can, it isn't hard. just make your mind a blank slate. no assumptions. ha ha. | 18:44 |
| jroll | ChuckC: feel free to just link it rather than adding me :P | 18:44 |
| *** lazy_prince is now known as killer_prince | 18:44 | |
| jroll | rloo: dunno, I read fast by default, it's hard to turn it off | 18:44 |
| jroll | killer_prince: hey, you should try https://review.openstack.org/#/c/121615/ | 18:44 |
| ChuckC | jroll: Ha! here you go: https://review.openstack.org/#/c/112129/2 | 18:46 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Fix timestamp column migration https://review.openstack.org/121651 | 18:47 |
| jroll | thanks :) | 18:49 |
| ChuckC | adam_g: thanks | 18:50 |
| *** rushiagr_away is now known as rushiagr | 18:51 | |
| *** yuanying has quit IRC | 18:53 | |
| *** dtantsur|bbl is now known as dtantsur | 18:56 | |
| devananda | back | 18:56 |
| *** lucas-afk is now known as lucasagomes | 18:56 | |
| lucasagomes | devananda, fyi patch is merged in the cli | 18:56 |
| devananda | sweet | 18:56 |
| NobodyCam | lol | 18:56 |
| lucasagomes | :) | 18:56 |
| *** penick has joined #openstack-ironic | 18:57 | |
| devananda | 2 min until meeting | 18:58 |
| NobodyCam | devananda: you running today? | 18:59 |
| devananda | sure | 18:59 |
| NobodyCam | :) | 18:59 |
| *** romcheg has quit IRC | 19:01 | |
| *** rushiagr is now known as rushiagr_away | 19:08 | |
| *** harlowja has quit IRC | 19:14 | |
| *** harlowja_ has joined #openstack-ironic | 19:14 | |
| *** marcoemorais has quit IRC | 19:18 | |
| *** marcoemorais has joined #openstack-ironic | 19:18 | |
| *** jistr has quit IRC | 19:18 | |
| *** jistr has joined #openstack-ironic | 19:19 | |
| *** cdnchris has quit IRC | 19:21 | |
| *** cdnchris has joined #openstack-ironic | 19:24 | |
| *** jistr has quit IRC | 19:25 | |
| *** dprince has quit IRC | 19:26 | |
| *** marcoemorais has quit IRC | 19:28 | |
| *** marcoemorais has joined #openstack-ironic | 19:28 | |
| *** GheRivero_pto is now known as GheRivero | 19:30 | |
| *** marcoemorais has quit IRC | 19:31 | |
| *** marcoemorais has joined #openstack-ironic | 19:32 | |
| *** cdnchris has quit IRC | 19:39 | |
| *** ChuckC has quit IRC | 19:41 | |
| *** pelix has quit IRC | 19:45 | |
| *** yuanying has joined #openstack-ironic | 19:50 | |
| *** cdnchris has joined #openstack-ironic | 19:54 | |
| *** ChuckC has joined #openstack-ironic | 19:58 | |
| * lucasagomes will crack open a bottle champagne if ironic graduates tomorrow! | 20:00 | |
| NobodyCam | lucasagomes: ++++++++++++++ | 20:01 |
| rloo | devananda, lucasagomes: two minor changes for CLI: https://review.openstack.org/#/c/121637/ & https://review.openstack.org/#/c/120953/ | 20:01 |
| jroll | lol | 20:01 |
| lucasagomes | rloo, will take a look | 20:01 |
| dtantsur | g'night folks :) | 20:01 |
| *** dtantsur is now known as dtantsur|afk | 20:02 | |
| JayF | night | 20:02 |
| rloo | lucasagomes, before I forget. I added a comment to 121050 -- it has merged but wonder if we should change | 20:02 |
| lucasagomes | the 2nd one you already -1'ed | 20:02 |
| lucasagomes | dtantsur|afk, night | 20:02 |
| lucasagomes | rloo, looks | 20:02 |
| rloo | lucasagomes: yeah. the code is ok, i just didn't like where he put the tests. | 20:02 |
| rloo | lucasagomes: i can +2 it if we want it in today. I was hoping he'd fix it tonight. or I guess I can fix it now? | 20:02 |
| * jroll bbiab | 20:03 | |
| lucasagomes | rloo, oh yeah it really changed ... hmmmm | 20:03 |
| devananda | rloo: or post a trivial patch on top that moves the tests | 20:03 |
| lucasagomes | rloo, I think he's from japan it may be very late there | 20:03 |
| rloo | lucasagomes: yeah, he's from japan. | 20:04 |
| lucasagomes | rloo, right, I will check that new nova approach tomorrow | 20:04 |
| lucasagomes | rloo, besically they are deleting the pyc files before run the tests :/ | 20:04 |
| lucasagomes | (prefer just telling python to just not generate it) | 20:04 |
| rloo | lucasagomes: yup; it seemed like maybe we should do a similar thing as nova there | 20:05 |
| lucasagomes | rloo, their approach looks more costy | 20:05 |
| lucasagomes | I will investigate it tomorrow | 20:05 |
| rloo | devananda: are you ok to wait til tomorrow to see if he fixes that? I'm inclined to take the least-amount-of-work way ;) | 20:05 |
| rloo | lucasagomes: take a look at the email thread. | 20:06 |
| lucasagomes | rloo, yeah I answer that | 20:06 |
| lucasagomes | put my 0.5c there | 20:06 |
| rloo | lucasagomes: they use cents in ireland? how much is that really worth? :-) | 20:06 |
| *** andreykurilin_ has joined #openstack-ironic | 20:07 | |
| lucasagomes | rloo, yup euro cents | 20:07 |
| rloo | lucasagomes: ah, useful in Paris then ;) | 20:07 |
| lucasagomes | exactly, that's very handy actually, not having to changing the currency etc | 20:08 |
| lucasagomes | devananda, https://review.openstack.org/#/c/120149/ if you have a time | 20:08 |
| lucasagomes | rloo, are you submitting a new patch set for #120953? | 20:11 |
| lucasagomes | just moving the code | 20:11 |
| rloo | lucasagomes: you want me to +2 and then submit a new patch? | 20:12 |
| lucasagomes | rloo, hmm I was thinking about just submit a new patch set for it?! | 20:12 |
| lucasagomes | seems less work | 20:12 |
| rloo | lucasagomes: can do that too ;) | 20:12 |
| lucasagomes | rloo, ack | 20:13 |
| openstackgerrit | Ruby Loo proposed a change to openstack/python-ironicclient: Add 'rebuild' option to node-set-provision-state https://review.openstack.org/120953 | 20:15 |
| lucasagomes | rloo, ta much! | 20:17 |
| rloo | lucasagomes: ;) | 20:17 |
| *** yuanying_ has joined #openstack-ironic | 20:17 | |
| lucasagomes | +2'ed | 20:17 |
| lucasagomes | and with that I will call it a day :) | 20:17 |
| lucasagomes | have a good night everybody! | 20:18 |
| * jroll +A | 20:18 | |
| jroll | night lucas :) | 20:18 |
| rloo | thx lucasagomes. Goodnight! | 20:18 |
| lucasagomes | night :) | 20:18 |
| *** lucasagomes is now known as lucas-zZz | 20:18 | |
| NobodyCam | night lucas-zZz :) | 20:18 |
| *** yuanying has quit IRC | 20:20 | |
| *** stevebak` is now known as stevebaker | 20:21 | |
| *** cdnchris has quit IRC | 20:22 | |
| *** cdnchris has joined #openstack-ironic | 20:28 | |
| NobodyCam | brb | 20:33 |
| * greghaynes waves | 20:33 | |
| greghaynes | Looks like there was some hash ring talk at the meeting - I agree with what I think is the consensus: Its a big change for this late in the cycle | 20:34 |
| Shrews | oh look, it's a greghaynes | 20:34 |
| greghaynes | :) | 20:34 |
| Shrews | devananda: ^^^ | 20:34 |
| devananda | hi! | 20:34 |
| greghaynes | Shrews: Yea, ty for pointing that out to me :) | 20:34 |
| Shrews | np | 20:35 |
| devananda | greghaynes: right. So let me frame the question this way: "is the PXE driver functionally broken at a scale beyond 1 conductor node"? | 20:35 |
| greghaynes | No idea, im not really versed in the PXE driver | 20:36 |
| greghaynes | Is there a different, specific bug for the PXE driver? I just know of https://bugs.launchpad.net/ironic/+bug/1355510 | 20:38 |
| jroll | I think that bug assumes that the PXE driver is the only driver | 20:38 |
| devananda | ah | 20:38 |
| devananda | so | 20:38 |
| devananda | "When using multiple conductor instances the HashRing should be updated over time as each conductor become active in order to allocate nodes evenly" | 20:38 |
| devananda | is not an accurate statement, really | 20:38 |
| devananda | it's glossed to the point of beign misleading, IMO | 20:39 |
| devananda | conductors don't actually allocate nodes | 20:39 |
| devananda | the HashRing instantiated in the API service is responsible for mapping nodes -> conductors, and then routing messages to the right place | 20:39 |
| devananda | a conductor receives the message and acts upon it -- but doesn't care about the HashRing at this point | 20:40 |
| devananda | it only cares if it can get the Task lock on that node | 20:40 |
| * devananda wonders if we need clearer dev docs on that process | 20:40 | |
| devananda | the place where the PXE driver in particular cares about this mapping | 20:40 |
| devananda | is that, when it provisions a node, it also updates the DHCP provider with a reference to that conductor's IP and the local TFTP files | 20:41 |
| devananda | if that conductor goes away, another conductor should, in theory, rebuild those TFTP files and update teh DHCP provider | 20:41 |
| devananda | that's where driver.deploy.take_over() comes in | 20:41 |
| devananda | and that should be called by a ring rebalance | 20:42 |
| devananda | Right - so I commented on the bug on 8/12 to this effect | 20:42 |
| devananda | but there's no response | 20:42 |
| devananda | greghaynes: and I dont see a patch listed on that bug? | 20:43 |
| devananda | oh, JayF also commented - and yes, you're correct. | 20:43 |
| greghaynes | ok, so that makes a lot more sense now. The patches I have been working on have not dealt with the periodic task bit | 20:44 |
| greghaynes | which is what will do the rebuild of tftp files | 20:44 |
| devananda | greghaynes: so I think there are two very distinct bugs here | 20:44 |
| devananda | 1. too many "buckets" move around the ring when the list of "targets" changes | 20:45 |
| devananda | 2. there's no periodic task on the conductors which syncs their local view of the hash and takes necessary actions when it changes | 20:45 |
| Shrews | devananda: https://bugs.launchpad.net/ironic/+bug/1365334 | 20:46 |
| devananda | right | 20:46 |
| devananda | there are two bugs :) | 20:46 |
| greghaynes | Agreed | 20:46 |
| greghaynes | and functionally, they both depend on rebalancing | 20:47 |
| devananda | three | 20:47 |
| devananda | Shrews: https://bugs.launchpad.net/ironic/+bug/1279331 | 20:47 |
| Shrews | oy | 20:47 |
| * devananda targets https://bugs.launchpad.net/ironic/+bug/1279331 to RC1 | 20:47 | |
| devananda | - no periodic task to keep PXE up to date | 20:48 |
| jroll | really sick of random gate fails :/ | 20:48 |
| devananda | - hash ring isn't updated on condutor when other conductors join/leave | 20:48 |
| *** harlowja_ is now known as harlowja_away | 20:49 | |
| devananda | - hash ring mapping changes too much when anything joins/leaves | 20:49 |
| * NobodyCam updated https://wiki.openstack.org/wiki/Ironic/NovaBaremetalIronicMigration#Baremetal_section to attempt show remove better.. is it too much? | 20:49 | |
| Shrews | NobodyCam: i question the need for an example of "deleting" :) just my opinion though | 20:50 |
| rloo | NobodyCam: whoa! What happened to my eyes? | 20:50 |
| NobodyCam | too much then :-p | 20:50 |
| jroll | NobodyCam: lol, yeah, maybe show it in diff format or something? | 20:51 |
| *** pensu has quit IRC | 20:51 | |
| rloo | Shrews: is it cuz it takes too many lines? Could condense it with [Baremetal] ... | 20:51 |
| Shrews | rloo: "Remove the [baremetal] section" is pretty self-explanatory to me | 20:51 |
| rloo | NobodyCam: it isn't a big deal if they don't understand? if they don't delete it, things should still work anyway, right? Or are you worried they'll add that in? | 20:51 |
| devananda | NobodyCam: I ++Shrews -- just say "delete the [baremetal] section". No need to copy the whole thing into the wiki | 20:52 |
| NobodyCam | ack | 20:52 |
| greghaynes | So, a small thing I ran into - We dont want to make an API change when we add conductor_id to the db. Is there a sane way to add this property to the controllers/v1/node:Node class without adding a wsme property? The alternative is to put a bunch of logic in the NodesController that I think really belongs in the Node class itself... | 20:52 |
| *** linggao has quit IRC | 20:53 | |
| devananda | greghaynes: why add it to api/controllers/v1/node at all? | 20:53 |
| * greghaynes is still just learning this whole wsme ironic thing | 20:53 | |
| NobodyCam | ok better? | 20:53 |
| devananda | greghaynes: just add it to the db and object models, not the api model | 20:54 |
| greghaynes | devananda: Whenever a node is created via API that property needs to get filled out somehow, that and NodesController seemed like the two entry points for that | 20:54 |
| *** yuanying_ has quit IRC | 20:54 | |
| rloo | NobodyCam: so we can migrate the nodes from BM to Ironic. But if a BM node has an instance on it, then what? (2nd point in Assumptions...) | 20:54 |
| greghaynes | hrm, so what would actually perform the conductor_id lookup and fil out that property? | 20:54 |
| devananda | greghaynes: ironic/objects/node.py -- add "conductor_id: obj_utils.str_or_none | 20:54 |
| Shrews | NobodyCam: the example just doesn't make sense. i'd drop it entirely | 20:54 |
| *** cdnchris has quit IRC | 20:55 | |
| openstackgerrit | A change was merged to openstack/ironic: Fix timestamp column migration https://review.openstack.org/121651 | 20:55 |
| devananda | greghaynes: take a look at how chassis_id is handled within api/controllers/v1/node.py Node.__init__() | 20:56 |
| NobodyCam | Shrews: ok done. | 20:56 |
| devananda | greghaynes: it's hidden from the REST API but still handled internally | 20:56 |
| NobodyCam | :) | 20:56 |
| greghaynes | hrmm, ok | 20:57 |
| devananda | greghaynes: actually - we shouldn't handle this the same way as chassis_id -- there is already handling of internal attributes. | 20:58 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Update README with a bit more info https://review.openstack.org/121637 | 20:58 |
| devananda | for k in fields: if not hasattr(self, k): continue | 20:58 |
| greghaynes | yea, uuid looks like the inverse of this - its API only | 20:59 |
| greghaynes | er, chasis_uuid | 20:59 |
| *** eghobo has quit IRC | 21:00 | |
| devananda | greghaynes: right. this should be a private member only. eg, it goes in objects/node.py and db/sqla/models.py | 21:00 |
| devananda | and the API should just work (and not change) | 21:00 |
| jroll | /b 3 | 21:01 |
| jroll | >.> | 21:01 |
| greghaynes | yep, I just need to take care of the actually filling out that field on node creation. Easy way is to just put some logic in the NodeController that looks up the conductor_id on node create but seems less than idael | 21:01 |
| devananda | oh, i see what you mean | 21:02 |
| devananda | why not leave it NULL ? | 21:02 |
| greghaynes | and then just fill it out via periodic task? | 21:02 |
| devananda | a conductor shouldn't do anything when a node is first created | 21:02 |
| devananda | it should, however, start polling it within the regular periodic tasks | 21:03 |
| devananda | eg, sync_power_state | 21:03 |
| devananda | the point of that field is to assist with rebalancing | 21:03 |
| devananda | so | 21:03 |
| devananda | greghaynes: why not have the periodic task which handles rebalancing simply fill in for NULL values when it finds them? | 21:04 |
| openstackgerrit | A change was merged to openstack/ironic-python-agent: Remove PyPy from default tox runlist https://review.openstack.org/121195 | 21:04 |
| greghaynes | yep, SGTM | 21:04 |
| devananda | cool | 21:04 |
| greghaynes | ty | 21:04 |
| Shrews | devananda: greghaynes: what happens if a conductor fails before conductor_id gets assigned? could some nodes be left orphaned? | 21:05 |
| greghaynes | that should be fine - the periodic task is just checking for nodes that dont belong to an active conductor | 21:05 |
| greghaynes | so itll pick that node up next round | 21:06 |
| Shrews | when/where are nodes associated with a conductor then? provisioning time? | 21:07 |
| greghaynes | this isnt really much more than a cache of the association, to be clear | 21:08 |
| greghaynes | the mapping for API requests happens whenever rebalance is called on the hashring | 21:08 |
| *** linggao has joined #openstack-ironic | 21:13 | |
| devananda | greghaynes: yea, i think we need to be -really- clear that the db column is not an assignment, it's a cache of the association | 21:16 |
| devananda | which is a concern I have in the proposed implementation's wording | 21:17 |
| JayF | I'm not sure I see value in caching the association, is the for operator use? What action would I take based on that? | 21:17 |
| devananda | JayF: none. it's hidden from the API | 21:17 |
| devananda | JayF: it's an optimization for the rebalance case | 21:17 |
| JayF | I like the information there in theory, but if it's potentially out of date it makes me go hrm | 21:17 |
| JayF | hm. okay then | 21:17 |
| devananda | when a conductor checks "are there nodes that I need to take over" | 21:17 |
| devananda | w/o a cache: it walks the list of all nodes | 21:18 |
| devananda | w/ a cache: it walks the list of nodes only for known dead conductors | 21:18 |
| JayF | with a potentially invalidated cache, shouldn't it do that anyway? | 21:18 |
| devananda | eg, if there are 100 conductors, and 1 dies, on the next cycle, 99 conductors will walk 1% of the nodes | 21:18 |
| devananda | though | 21:19 |
| devananda | i was very tired when I discussed this with lifeless, and thinking about it now, I think this is really an overly pessimistic optimization | 21:19 |
| devananda | by that, i mean, the harm in having each conductor walk the full list isn't that great | 21:19 |
| JayF | I don't have many opinions about which way to do it from a technical standpoint; I just wanted to make sure it wasn't intended for operator consumption, because I'd consider that dubious | 21:19 |
| devananda | it is O(N) vs O(N/c) where N = # nodes and c = # conductors | 21:20 |
| JayF | and you miss some | 21:20 |
| JayF | potentially | 21:20 |
| devananda | for a non-trivial increase in complexity | 21:20 |
| JayF | right? | 21:20 |
| devananda | JayF: miss some? | 21:20 |
| JayF | You said | 21:20 |
| devananda | and O(N/c) is still O(N) | 21:20 |
| JayF | 14:16:55 <devananda> greghaynes: yea, i think we need to be -really- clear that the db column is not an assignment, it's a cache of the association | 21:20 |
| devananda | especially for small c | 21:21 |
| jroll | JayF: use case is "conductor dies, update the nodes it belonged to" | 21:21 |
| JayF | it's only a cache of the association | 21:21 |
| devananda | which is all known deployments today | 21:21 |
| JayF | is there ever a case when a node could change conductors and that column not be updated? | 21:21 |
| JayF | the way you said that ^ above made me suspicious there may be | 21:21 |
| devananda | JayF: there are two "walks" that need to happen during the periodic task | 21:21 |
| devananda | 1- conductor walks all nodes mapped to it, checks local state, checks DB and updates if needed | 21:22 |
| JayF | I know the API node will recalculate on the fly as well, so there's always a period of time when a conductor dies that the db doesn't reflect reality. My question is if it would ever not reflect reality in a way that'd cause a node to not be reassigned to a conductor | 21:22 |
| devananda | 2- conductor syncs hash ring, detects if it changed; IIF it changed, conductor fetches list of dead conductors, walks all nodes mapped to them. IFF some of those map to conductor, perform (1) on those nodes too | 21:22 |
| devananda | honestly | 21:23 |
| devananda | for simplicity | 21:23 |
| devananda | i think it's fine to just walk the whole list of nodes once | 21:23 |
| Shrews | devananda: is a lock needed to just query the node? if so, this could increase the number of node lock conflicts. limiting that might be worth it. | 21:25 |
| devananda | the more nodes in a deployment, the longer it'll take to walk the list -- and simply spacing the periodic task out further doesn't scale, because it means increasing the window where nodes may be unmapped/unmanaged | 21:25 |
| devananda | Shrews: no - querying doesn't lock | 21:25 |
| JayF | Shrews: devananda: Right now Ironic locks a node when an agent heartbeats, fwiw | 21:26 |
| JayF | I think we lock nodes on all vendor_passthru but imbw | 21:26 |
| devananda | JayF: yes. because it is a POST | 21:26 |
| Shrews | thankfully, we now have node lock retry :) | 21:27 |
| devananda | JayF: and because ConductorManager doesn't introspect it, doesn't know what the driver will do, and just passes it to the driver, so it must take a lock | 21:27 |
| devananda | hm. we should perhaps docuemnt somewhere that all API interaction with /vendor_passthru/ is serliazed in the driver IMPL | 21:28 |
| JayF | Should we promote the heartbeat endpoint to not be v-p in K? | 21:28 |
| *** linggao has quit IRC | 21:29 | |
| rloo | NobodyCam: are you OK if I replace the 'Intialize the Ironic schema' part in the migration doc, with something about Installing Ironic via following http://docs.openstack.org/developer/ironic/deploy/install-guide.html#install-and-configure-prerequisites | 21:30 |
| jroll | devananda: I'd rather vendor_passthru left the lock up to the driver, tbh | 21:31 |
| devananda | jroll: ++ | 21:31 |
| jroll | devananda: I also want to work on a real policy thing for ironic | 21:31 |
| * jroll ducks | 21:32 | |
| devananda | o.0 | 21:32 |
| NobodyCam | rloo: yes, that doc covers it well I think | 21:32 |
| rloo | NobodyCam: thx. | 21:32 |
| jroll | devananda: different operator roles | 21:32 |
| * devananda throws a nerf dart at jroll, just for good measure | 21:32 | |
| jroll | devananda: e.g. monitoring account gets read-only | 21:32 |
| devananda | ah | 21:32 |
| devananda | jroll: ++ | 21:33 |
| jroll | devananda: e.g. some guy racking hardware gets access to POST /nodes | 21:33 |
| jroll | yeah | 21:33 |
| jroll | cool, let's make that happen in kilo | 21:33 |
| devananda | jroll: combine that with lifecycle notifications | 21:33 |
| devananda | and you have some neat things | 21:33 |
| jroll | just getting logs of you being on board :) | 21:33 |
| jroll | indeed | 21:33 |
| lifeless | devananda: uhm, I don't think the analysis above is quite right | 21:33 |
| lifeless | devananda: e.g. I don't think recording the conductor that owns a node in the node is a cache vs actually its semantic | 21:34 |
| jroll | devananda: both of these have come out of actually running an ironic install :P | 21:34 |
| devananda | lifeless: it's not an assingment, though | 21:34 |
| *** lucas-zZz has quit IRC | 21:43 | |
| devananda | lifeless: depending on how its implemented, it could be semantic -- it might mean "this is the last conductor which operated on this node" | 21:43 |
| devananda | lifeless: but implementing it to have that meaning => additional DB write for every other operation that a conductor does (really: changes in several parts of code to do that w/o needless db transactions) | 21:44 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add a doc note about the vendor_passthru endpoint https://review.openstack.org/121678 | 21:45 |
| *** cdnchris has joined #openstack-ironic | 21:47 | |
| *** andreykurilin_ has quit IRC | 21:48 | |
| *** andreykurilin_ has joined #openstack-ironic | 21:48 | |
| *** yuanying has joined #openstack-ironic | 21:51 | |
| lifeless | devananda: I don't understand why one would need to write when the value hasn't altered | 21:52 |
| lifeless | devananda: as for the O(N) vs O(N/c) - the first scales as the nodes managed increases, the second scales as the *nodes per conductor* increases. | 21:54 |
| lifeless | devananda: and we can reasonably expect the second one to be held roughly constant, which isn't true of the first. | 21:54 |
| devananda | lifeless: right, of course | 21:58 |
| *** cdnchris has quit IRC | 22:00 | |
| jroll | devananda: https://review.openstack.org/#/c/121678/1/doc/source/dev/drivers.rst is this about node vendor passthru, or driver vendor passthru? | 22:01 |
| jroll | (thinking the latter) | 22:01 |
| devananda | jroll: node | 22:02 |
| jroll | aha | 22:02 |
| jroll | devananda: want to add driver as well? | 22:02 |
| jroll | or later as a separate commit? | 22:02 |
| devananda | jroll: i can note it there. any thing you'd want to call out specially? | 22:03 |
| devananda | i think IPA is the only driver that implements it | 22:03 |
| jroll | devananda: same sort of info, I guess | 22:03 |
| jroll | devananda: making comments on this though, if you want to wait | 22:03 |
| jroll | e.g. this isn't syncronous | 22:03 |
| jroll | vendor_passthru is async | 22:03 |
| *** cdnchris has joined #openstack-ironic | 22:04 | |
| *** cdnchris has left #openstack-ironic | 22:04 | |
| devananda | jroll: ahh, it is now, you're right | 22:06 |
| devananda | that doesn't make any sense | 22:06 |
| jroll | what doesn't make any sense? | 22:06 |
| * devananda must have forgotten why we did that | 22:07 | |
| jroll | um | 22:07 |
| jroll | because the conductor handles it? | 22:07 |
| devananda | oh right - that's why folks want a GET | 22:07 |
| jroll | because we do the actual deploy in vendor passthru? | 22:07 |
| jroll | yeah | 22:07 |
| devananda | there's no means to return info from the driver today | 22:07 |
| jroll | right | 22:07 |
| devananda | I was thinking of older code. /me updates | 22:08 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding support for decommissioning https://review.openstack.org/104379 | 22:08 |
| *** penick has quit IRC | 22:13 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add a doc note about the vendor_passthru endpoint https://review.openstack.org/121678 | 22:14 |
| jroll | devananda: hrm, driver vendor passthru is not async :P | 22:15 |
| *** penick has joined #openstack-ironic | 22:15 | |
| devananda | gah! | 22:18 |
| *** yuanying has quit IRC | 22:24 | |
| *** harlowja_away is now known as harlowja_ | 22:24 | |
| *** jcoufal has quit IRC | 22:25 | |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding support for decommissioning https://review.openstack.org/104379 | 22:25 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add a doc note about the vendor_passthru endpoint https://review.openstack.org/121678 | 22:26 |
| devananda | jroll: let's try again :) | 22:26 |
| * jroll looks | 22:26 | |
| jroll | devananda: want me to nit on whitespace? :P | 22:27 |
| devananda | hah | 22:27 |
| * devananda needs to add RST rules to .vim | 22:27 | |
| jroll | d'oh | 22:28 |
| jroll | /v1/{NODE}/vendor_passthru/method={METHOD} | 22:28 |
| jroll | should be /v1/{NODE}/vendor_passthru?method={METHOD} | 22:28 |
| jroll | yes? | 22:28 |
| jroll | / vs ? | 22:28 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add a doc note about the vendor_passthru endpoint https://review.openstack.org/121678 | 22:28 |
| devananda | oh - right | 22:28 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add a doc note about the vendor_passthru endpoint https://review.openstack.org/121678 | 22:29 |
| jroll | devananda: and "vendor passthrough", should that be "vendor passthru" for consistency? | 22:29 |
| devananda | jroll: what if it just said "drivers may implement a passthrough API" ? | 22:29 |
| jroll | I mean, don't care either way | 22:29 |
| jroll | just asking :) | 22:30 |
| jroll | both are fine | 22:30 |
| jroll | devananda: +2 | 22:31 |
| devananda | tiny change to wording in one other spot, new version inc | 22:32 |
| devananda | jroll: thanks | 22:32 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Add a doc note about the vendor_passthru endpoint https://review.openstack.org/121678 | 22:32 |
| jroll | Looks fine | 22:33 |
| *** penick has quit IRC | 22:34 | |
| *** penick has joined #openstack-ironic | 22:37 | |
| NobodyCam | walkies time ... brb | 22:39 |
| *** ryanpetrello has quit IRC | 22:48 | |
| *** DavidHu_ has joined #openstack-ironic | 22:51 | |
| *** penick has quit IRC | 22:51 | |
| *** ryanpetrello has joined #openstack-ironic | 22:54 | |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Allow clean_up with missing image ref https://review.openstack.org/121615 | 22:59 |
| jroll | devananda: ^ now with tests! | 22:59 |
| * jroll thinks he should try it in devstack | 22:59 | |
| *** andreykurilin_ has quit IRC | 22:59 | |
| jroll | adam_g, killer_prince, y'all might also be interested ^ | 22:59 |
| adam_g | jroll, nice. was just testing something with that cherry-picked | 23:00 |
| jroll | adam_g: cool, actual code is the same with the exception of fixing the logging call | 23:00 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Add 'rebuild' option to node-set-provision-state https://review.openstack.org/120953 | 23:03 |
| * jroll takes a stck.sh break | 23:07 | |
| *** yuanying has joined #openstack-ironic | 23:21 | |
| *** yuanying has quit IRC | 23:23 | |
| *** yuanying has joined #openstack-ironic | 23:23 | |
| *** yuanying has quit IRC | 23:23 | |
| *** yuanying has joined #openstack-ironic | 23:24 | |
| NobodyCam | restck.sh jroll | 23:34 |
| NobodyCam | jroll: was curious what the "mock side_effect weirdness" was | 23:36 |
| *** igordcard has quit IRC | 23:37 | |
| jroll | lol | 23:39 |
| jroll | NobodyCam: so side_effect gets the arguments passed to the mock (in this case, node and task.context) | 23:40 |
| jroll | NobodyCam: so the exception gets instantiated with those, and yeah it fails hard | 23:40 |
| jroll | it's... weird | 23:40 |
| NobodyCam | oh odd | 23:41 |
| openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Fix Devstack docs for zsh users https://review.openstack.org/121712 | 23:44 |
| jroll | this is another odd one too :P | 23:44 |
| JayF | Why are you running zshell on your devstack box? | 23:44 |
| JayF | hahaha | 23:44 |
| jroll | because I sometimes develop there? | 23:45 |
| JayF | jroll: alternatively, image=/bin/bash -c $() | 23:45 |
| jroll | and I prefer to use zsh? | 23:45 |
| jroll | why am I using vim and not nano | 23:45 |
| jroll | mehhhhhhh | 23:45 |
| JayF | erm, image=$(/bin/bash -c "things") | 23:45 |
| * NobodyCam switches jroll's shell to c-shell | 23:45 | |
| JayF | c shell | 23:45 |
| JayF | c shell run | 23:45 |
| JayF | run c shell run | 23:46 |
| jroll | lol | 23:46 |
| JayF | shit I forgot a {} | 23:46 |
| NobodyCam | lol | 23:46 |
| NobodyCam | ahh http://pysh.sourceforge.net this looked fun http://pysh.sourceforge.net | 23:48 |
| NobodyCam | doh dbl paste | 23:48 |
| jroll | interesting | 23:48 |
| NobodyCam | lol gota luv the BIG BOLD and has not been updated in years | 23:49 |
| jroll | tested https://review.openstack.org/#/c/121615/ in devstack, totally works | 23:49 |
| * jroll double checks without nuking instance_info | 23:50 | |
| JayF | NobodyCam: so we were goofing around some before the RHC[E|T] a few years back, breaking machines as hilariously as possible and trying to get people to recover them | 23:50 |
| JayF | NobodyCam: we deleted all shells on a system for one of 'em, my buddy booted it with init=/usr/bin/python and used python to download and write out bash, symlink it to sh, and unbreak the system | 23:51 |
| NobodyCam | oh wow slick !!! | 23:51 |
| *** r-daneel__ has quit IRC | 23:52 | |
| NobodyCam | did you tell him you deleted the shells or did you make him try each one? | 23:53 |
| * JayF is working on KEYRING support for python-ironicclient | 23:53 | |
| JayF | NobodyCam: I think he had to try over and ovetr | 23:53 |
| NobodyCam | JayF: nice :) | 23:53 |
| openstackgerrit | Sergey Lupersolsky proposed a change to openstack/ironic: Add attempts to import necessary modules for IMPINative and iBoot fake drivers https://review.openstack.org/121715 | 23:55 |
| jroll | works in normal operation too | 23:56 |
| jroll | devananda, NobodyCam, let's land this thing: https://review.openstack.org/#/c/121615/ | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!