| adam_g | NobodyCam, i was hitting that quite a bit yesterday after i had managed to wedge the database into a bad state | 00:00 |
|---|---|---|
| NobodyCam | :) | 00:01 |
| NobodyCam | virsh list | 00:04 |
| NobodyCam | gah ww | 00:04 |
| *** rloo has quit IRC | 00:20 | |
| *** krtaylor has joined #openstack-ironic | 00:20 | |
| * NobodyCam fires up another devtest after rebasing his elements dir | 00:24 | |
| *** matsuhashi has joined #openstack-ironic | 00:27 | |
| lifeless | NobodyCam: should be | 00:29 |
| NobodyCam | :) i needed to rebase images on master to the compute_manager option.... I think I check'd out the patch vs pick'in it | 00:30 |
| russell_h | adam_g: added a comment on that change, I think you need to merge the nested if into the one above it | 00:32 |
| adam_g | russell_h, sure | 00:32 |
| russell_h | adam_g: "if hasattr(e, '__name__') and e.__name__ == 'InstanceDeployFailure':" | 00:32 |
| russell_h | unless I'm really losing it | 00:32 |
| russell_h | or you could add a nested else | 00:32 |
| russell_h | I don't care, I just think it doesn't actually work as-is | 00:33 |
| adam_g | russell_h, that was >80 lines and wrapping it in ()'s looked ugly at the time :) | 00:33 |
| russell_h | ah | 00:33 |
| russell_h | :) | 00:33 |
| russell_h | there is nothing you can do to make python look pretty to me | 00:33 |
| * russell_h might be hanging out in the wrong spot | 00:34 | |
| adam_g | ha | 00:34 |
| JayF | Blueprint: Use Node.js for more faster eventing Author: russell_h | 00:34 |
| JayF | :P | 00:34 |
| jroll | s/node.js/go | 00:34 |
| lifeless | russell_h: adam_g: if getattr(e, '__name__') == 'InstanceDeployFailure': | 00:35 |
| adam_g | lifeless, goodcall | 00:36 |
| openstackgerrit | Adam Gandelman proposed a change to openstack/ironic: Raise unexpected exceptions during destroy() https://review.openstack.org/82948 | 00:42 |
| *** matsuhashi has quit IRC | 01:12 | |
| *** matsuhashi has joined #openstack-ironic | 01:14 | |
| *** eguz has quit IRC | 01:18 | |
| *** nosnos has joined #openstack-ironic | 01:27 | |
| comstud | lifeless: The no-DB scheduler is vaporware | 01:28 |
| comstud | as far as I'm concerned, anyway | 01:28 |
| lifeless | comstud: ah; so I can change the comment if you like, but that is why I said what I said | 01:28 |
| comstud | understood | 01:28 |
| *** krtaylor has quit IRC | 01:34 | |
| *** Haomeng has joined #openstack-ironic | 01:54 | |
| *** matsuhashi has quit IRC | 02:00 | |
| *** matsuhashi has joined #openstack-ironic | 02:08 | |
| *** hemna_ has quit IRC | 02:12 | |
| *** hemna_ has joined #openstack-ironic | 02:26 | |
| *** harlowja is now known as harlowja_away | 02:27 | |
| *** vkozhukalov has joined #openstack-ironic | 02:31 | |
| *** krtaylor has joined #openstack-ironic | 02:38 | |
| *** killer_p- has joined #openstack-ironic | 02:38 | |
| *** BadCub01 has quit IRC | 02:39 | |
| *** killer_p- is now known as killer_prince | 02:39 | |
| *** lazy_prince has quit IRC | 02:39 | |
| *** todd_dsm has joined #openstack-ironic | 03:03 | |
| *** todd_dsm has quit IRC | 03:08 | |
| *** matsuhashi has quit IRC | 03:15 | |
| *** nosnos has quit IRC | 03:28 | |
| *** blamar has quit IRC | 03:46 | |
| *** killer_prince is now known as lazy_prince | 03:53 | |
| *** blamar has joined #openstack-ironic | 03:56 | |
| *** blamar has quit IRC | 03:57 | |
| *** blamar has joined #openstack-ironic | 03:58 | |
| *** todd_dsm has joined #openstack-ironic | 04:10 | |
| *** blamar has quit IRC | 04:10 | |
| *** matsuhashi has joined #openstack-ironic | 04:25 | |
| *** killer_prince has joined #openstack-ironic | 04:27 | |
| *** eghobo has joined #openstack-ironic | 04:27 | |
| *** vkozhukalov has quit IRC | 04:28 | |
| *** nosnos has joined #openstack-ironic | 04:36 | |
| *** todd_dsm has quit IRC | 04:39 | |
| *** blamar has joined #openstack-ironic | 04:46 | |
| *** todd_dsm has joined #openstack-ironic | 04:49 | |
| *** todd_dsm has quit IRC | 04:49 | |
| *** lazy_prince has quit IRC | 05:28 | |
| *** Haomeng has quit IRC | 05:36 | |
| *** Haomeng has joined #openstack-ironic | 05:38 | |
| openstackgerrit | Russell Haering proposed a change to openstack/ironic: Re-use hash rings across requests https://review.openstack.org/82915 | 05:46 |
| *** lazy_prince has joined #openstack-ironic | 06:00 | |
| openstackgerrit | Jenkins proposed a change to openstack/ironic: Imported Translations from Transifex https://review.openstack.org/82993 | 06:07 |
| *** saju_m has joined #openstack-ironic | 06:19 | |
| *** Manishanker has joined #openstack-ironic | 06:23 | |
| *** mrda is now known as mrda_away | 06:31 | |
| *** matsuhashi has quit IRC | 06:33 | |
| *** matsuhashi has joined #openstack-ironic | 06:35 | |
| *** eghobo has quit IRC | 06:44 | |
| *** mdurnosvistov_ has joined #openstack-ironic | 06:49 | |
| *** vkozhukalov has joined #openstack-ironic | 07:02 | |
| *** stevehuang has joined #openstack-ironic | 07:02 | |
| *** matsuhas_ has joined #openstack-ironic | 07:03 | |
| *** matsuhas_ has quit IRC | 07:04 | |
| *** matsuhas_ has joined #openstack-ironic | 07:06 | |
| *** matsuhashi has quit IRC | 07:07 | |
| *** rameshg87 has joined #openstack-ironic | 07:12 | |
| *** saju_m has quit IRC | 07:13 | |
| *** stevehuang has quit IRC | 07:13 | |
| *** matsuhas_ has quit IRC | 07:59 | |
| *** matsuhashi has joined #openstack-ironic | 08:04 | |
| *** mdurnosvistov_ has quit IRC | 08:15 | |
| *** jistr has joined #openstack-ironic | 08:37 | |
| *** ifarkas has joined #openstack-ironic | 08:42 | |
| *** ndipanov_ has quit IRC | 08:44 | |
| dtantsur | Morning Ironic | 08:56 |
| agordeev | dtantsur: morning | 08:57 |
| *** ndipanov has joined #openstack-ironic | 09:00 | |
| *** max_lobur has joined #openstack-ironic | 09:03 | |
| *** athomas has joined #openstack-ironic | 09:10 | |
| *** martyntaylor has joined #openstack-ironic | 09:10 | |
| *** derekh has joined #openstack-ironic | 09:11 | |
| mdurnosvistov | Morning all! =) | 09:29 |
| *** lucasagomes has joined #openstack-ironic | 09:31 | |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Add worker threads limit to _check_deploy_timeouts task https://review.openstack.org/78651 | 09:38 |
| *** derekh has quit IRC | 09:42 | |
| *** derekh has joined #openstack-ironic | 09:42 | |
| agordeev | mdurnosvistov: morning! | 09:46 |
| *** romcheg has joined #openstack-ironic | 09:49 | |
| *** pradipta_away is now known as pradipta | 09:54 | |
| *** matsuhashi has quit IRC | 10:09 | |
| lucasagomes | devananda, about the reversible AES https://github.com/cloudkeep/barbican/wiki/Barbican-Getting-Started-Guide | 10:10 |
| *** romcheg has quit IRC | 10:36 | |
| *** matsuhashi has joined #openstack-ironic | 10:39 | |
| lifeless | devananda: so comstud took his -1 away... :) | 10:49 |
| lifeless | lucasagomes: 82637 is a dependency for tripleo deploys of ironic | 10:49 |
| lifeless | lucasagomes: you had some concerns about etherpad links; I think my reasoning makes sense in the review, but I'm around for a little more if you want to talk | 10:50 |
| lucasagomes | lifeless, will take a look | 10:50 |
| lucasagomes | lifeless, right I need to talk to u about something not related to that ^ tho | 10:50 |
| * lucasagomes grabs the link | 10:50 | |
| lucasagomes | lifeless, https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/deploy_utils.py#L83-L87 | 10:51 |
| lifeless | ruhroh | 10:51 |
| lucasagomes | this comment still makes sense? do we need to have that 1mb at the start of the disk? | 10:51 |
| lifeless | yes :) | 10:51 |
| lucasagomes | I saw it failing once I inverted the partition layout to have root to be always the last partition | 10:51 |
| lifeless | sadly | 10:51 |
| lucasagomes | ouch | 10:51 |
| lucasagomes | lifeless, http://fpaste.org/88703/ | 10:52 |
| lifeless | lets see if its changed in trusty | 10:52 |
| lucasagomes | ack | 10:52 |
| lucasagomes | cause I tested here without it | 10:52 |
| lucasagomes | and everything works | 10:52 |
| lifeless | /* rounding up wastes almost an entire cylinder - round down | 10:52 |
| lifeless | and leave it to compute_start_sect() to fix the difference */ | 10:52 |
| lifeless | lucasagomes: looks like your disk is too small ? | 10:54 |
| lifeless | lucasagomes: max allowable size of 1 ? :) | 10:54 |
| lucasagomes | it's not, but having that 1MB seems to cause some partition overlap there | 10:54 |
| lifeless | did you skip the leading comma in your patch? | 10:55 |
| lifeless | thats what makes it all work | 10:55 |
| lifeless | without the 1MB there is no room for e.g. grub | 10:55 |
| lifeless | and the partition table itself | 10:55 |
| lifeless | MBR can represent crazy stuff | 10:55 |
| lucasagomes | lifeless, right so disk partition softwares especially that ones with deal with mbr should preserve the 512 bytes at the beggining of the diisk | 10:56 |
| lifeless | so the reason for the 1MB | 10:56 |
| lucasagomes | yeah | 10:56 |
| lifeless | sfdisk preserves that 512 bytes | 10:56 |
| lifeless | but it doesn't shift the partition up | 10:56 |
| lifeless | it shrinks it | 10:56 |
| lucasagomes | ouch | 10:56 |
| lifeless | so when the partition size exactly matches the image size | 10:56 |
| lifeless | e.g. image size = 10GB | 10:56 |
| lifeless | flavor root = 10GB | 10:57 |
| lifeless | you end up with 10GB-512bytes partition | 10:57 |
| lifeless | and boom. | 10:57 |
| lucasagomes | yeah... well that's expected ^ | 10:57 |
| lifeless | thats what compute_start_sect does | 10:57 |
| lifeless | so by leaving a leading gap | 10:57 |
| lucasagomes | I mean, maybe ironic should be more smart and create the partition until the end of the disk | 10:57 |
| lifeless | 10GB image + 10GB flavor works. | 10:57 |
| lucasagomes | in case both values are equal | 10:57 |
| lifeless | as in the current code works | 10:57 |
| lifeless | using 0 there won't. | 10:58 |
| lifeless | can you pastebin or point me at gerrit for the patch that caused that fpaste error? | 10:58 |
| lucasagomes | oh it's in my env, but I can post it here | 10:58 |
| lucasagomes | 1 sec | 10:58 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Convert credentials for ipminative driver to binary strings https://review.openstack.org/82146 | 10:58 |
| lucasagomes | lifeless, http://paste.openstack.org/show/74327/ | 10:59 |
| lucasagomes | all I did was to invert the swap partition with the root partition | 10:59 |
| lucasagomes | so the root will be always the last partiton for both layouts ephemeral and normal | 11:00 |
| lucasagomes | and then boom :) crashed | 11:00 |
| lifeless | is root_mb perhaps 0 ? | 11:00 |
| lucasagomes | no, | properties | {u'memory_mb': u'512', u'cpu_arch': u'x86_64', u'local_gb': u'10', | | 11:01 |
| lucasagomes | | | u'cpus': u'1'} | 11:01 |
| lucasagomes | 10G | 11:01 |
| lucasagomes | and my disk has | 11:02 |
| lucasagomes | Disk /dev/vda: 16.1 GB, | 11:02 |
| lifeless | I meant swap_mb | 11:02 |
| lucasagomes | ah | 11:02 |
| lucasagomes | yeah | 11:02 |
| lucasagomes | swap is 0 | 11:02 |
| lifeless | 0 means 'use all disk' to sfdisk | 11:02 |
| lucasagomes | which will be converted to 1 | 11:02 |
| lucasagomes | ohhhh | 11:02 |
| lucasagomes | ah but no | 11:02 |
| lifeless | thats why we convert it to 1 | 11:02 |
| lucasagomes | internally we change it to 1mb | 11:02 |
| lucasagomes | yeah, well I can check if the value changed to 1 there | 11:03 |
| lucasagomes | 1 sec | 11:03 |
| lucasagomes | btu odd enough, once I change that 1mb from the beggining it works | 11:03 |
| lifeless | odd indeed | 11:03 |
| lucasagomes | lifeless, http://paste.openstack.org/show/74334/ (without the 1MB, worked) | 11:05 |
| lucasagomes | yeah well I will investigate it a bit more | 11:05 |
| lifeless | perhaps you could add 1M onto the first partition size | 11:08 |
| lifeless | but IIRC I tried that and the problem was cylinder rounding | 11:08 |
| lifeless | gnight! | 11:08 |
| lucasagomes | lifeless, night! | 11:09 |
| *** rameshg87 has left #openstack-ironic | 11:17 | |
| *** saju_m has joined #openstack-ironic | 11:20 | |
| openstackgerrit | Sergey Lukjanov proposed a change to openstack/ironic: Start using oslosphinx theme for docs https://review.openstack.org/83043 | 11:29 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Fix strings post landing for c63e1d9f6 https://review.openstack.org/83044 | 11:30 |
| *** romcheg has joined #openstack-ironic | 11:36 | |
| *** athomas has quit IRC | 11:59 | |
| *** Slower has quit IRC | 12:05 | |
| *** athomas has joined #openstack-ironic | 12:05 | |
| *** matsuhashi has quit IRC | 12:10 | |
| *** matsuhashi has joined #openstack-ironic | 12:11 | |
| *** matsuhas_ has joined #openstack-ironic | 12:13 | |
| *** matsuhashi has quit IRC | 12:13 | |
| *** lucasagomes is now known as lucas-hungry | 12:44 | |
| *** matty_dubs|gone is now known as matty_dubs | 12:46 | |
| *** matsuhas_ has quit IRC | 13:07 | |
| *** max_lobur1 has joined #openstack-ironic | 13:10 | |
| *** max_lobur has quit IRC | 13:10 | |
| *** rloo has joined #openstack-ironic | 13:12 | |
| *** pradipta is now known as pradipta_away | 13:12 | |
| *** nosnos has quit IRC | 13:35 | |
| devananda | morning, all | 13:43 |
| *** killer_prince has quit IRC | 13:46 | |
| *** lazy_prince is now known as killer_prince | 13:46 | |
| rloo | morning devananda | 13:47 |
| *** lucas-hungry is now known as lucasagomes | 13:50 | |
| lucasagomes | mornin devananda | 13:50 |
| lucasagomes | morning rloo | 13:50 |
| rloo | afternoon lucasagomes ;) | 13:51 |
| lucasagomes | :D | 13:51 |
| lucasagomes | devananda, ping, r u working on https://bugs.launchpad.net/ironic/+bug/1265587 ? | 13:51 |
| devananda | JayF: didn't you start on ^ ? | 13:52 |
| devananda | lucasagomes: i'm not | 13:52 |
| lucasagomes | devananda, right, if JayF is not working on it as well I will give it a go | 13:53 |
| lucasagomes | it's tagged as rc1 | 13:53 |
| devananda | yep. thanks | 13:53 |
| rloo | well, JayF did mention yesterday that he'd work on it. Maybe wait for a few hours to check with him? | 13:56 |
| NobodyCam | good morning Ironic | 14:04 |
| lucasagomes | rloo, yup, will wait | 14:05 |
| lucasagomes | morning NobodyCam | 14:05 |
| NobodyCam | morning lucasagomes :) | 14:05 |
| rloo | morning NobodyCam | 14:10 |
| NobodyCam | moring rloo | 14:10 |
| NobodyCam | morning even :) | 14:10 |
| *** jgrimm has joined #openstack-ironic | 14:11 | |
| rloo | NobodyCam: understandable, you haven't gotten your coffee yet ;) | 14:11 |
| NobodyCam | :) actually its ready now... brb ...lol | 14:12 |
| *** Manishanker has quit IRC | 14:14 | |
| *** linggao has joined #openstack-ironic | 14:28 | |
| * Shrews settling on a post-morning-workout dirty chai.... mmmmm | 14:33 | |
| NobodyCam | :) blond rost with vanilla cream here :) | 14:35 |
| *** yuriyz has quit IRC | 14:40 | |
| NobodyCam | brb | 14:40 |
| *** branen_ has quit IRC | 14:40 | |
| devananda | dirty chai? | 14:51 |
| devananda | sounds tasty | 14:52 |
| Shrews | t'is indeed. chai + shot of espresso | 14:52 |
| devananda | so, i'm looking at https://launchpad.net/ironic/+milestone/icehouse-rc1 and really happy with the progress we made yesterday | 14:52 |
| NobodyCam | :) | 14:53 |
| NobodyCam | morning devananda | 14:53 |
| devananda | only two not committed yet | 14:53 |
| Shrews | yeah, well, working on it :-P | 14:53 |
| devananda | i'm going to give the bug list another pass, and am spinning up a fresh devstack locally | 14:54 |
| devananda | Shrews: we had ~10 open yesterday. that wasnt' a critique of the bug you're working on :) | 14:54 |
| Shrews | oh, suuuuure :-P | 14:54 |
| Shrews | NobodyCam, devananda: when you brought the nova virt driver into the ironic tree, was it just a copy of the existing baremetal driver or separate? | 14:59 |
| openstackgerrit | Aleksandr Gordeev proposed a change to openstack/ironic-python-agent: Get rid of modes. Introduce pluggable extensions https://review.openstack.org/83087 | 15:00 |
| openstackgerrit | A change was merged to openstack/python-ironicclient: Updated from global requirements https://review.openstack.org/75585 | 15:01 |
| NobodyCam | Shrews: it came from here : https://review.openstack.org/#/c/51328 | 15:01 |
| devananda | Shrews: i started the nova.virt.ironic driver from scratch back in december, it's been worked on by several folks (nobodycam, lucas, ...) since then, until we moved it into our tree | 15:01 |
| NobodyCam | undercloud | CREATE_IN_PROGRESS ... woo hoo | 15:03 |
| NobodyCam | thou I cann't get to that point with out a little hacking | 15:03 |
| Shrews | devananda: ok, gotcha. hmmm, wonder if i can use that unit test from 51328 if I copy the code back into nova for a test drive | 15:04 |
| Shrews | devananda: my only other thought for testing is to setup devstack, manually pull in my change to ironic, and drive enough api calls to test the parts i've modified | 15:06 |
| devananda | Shrews: "copy the code back into nova" -- should, in theory, work | 15:07 |
| devananda | Shrews: IMBC, but it'd be neat if that happened automatically in our unit tests | 15:07 |
| NobodyCam | Shrews: I do know we have broken some of the current tests | 15:09 |
| Shrews | devananda: you're definitely crazy.... but we do need something | 15:09 |
| Shrews | :) | 15:09 |
| NobodyCam | :-p | 15:09 |
| Shrews | i can't even fathom how that might work, atm | 15:10 |
| Shrews | it'd almost be easier to modify nova's tests to look for external drivers, copy them in, and run the tests | 15:11 |
| NobodyCam | wounder if we could use someting like the current dib element that installes ironic in side the nova venv? | 15:11 |
| *** dwalleck has joined #openstack-ironic | 15:11 | |
| NobodyCam | installs even | 15:12 |
| NobodyCam | Shrews: https://github.com/openstack/tripleo-image-elements/blob/master/elements/nova-ironic/post-install.d/81-nova-ironic-driver | 15:12 |
| *** stevehuang has joined #openstack-ironic | 15:14 | |
| Shrews | seems a bit much for unit testing. functional testing, maybe. | 15:16 |
| Shrews | i'm wondering if we can somehow copy/fake the nova ComputeManager.... | 15:16 |
| *** lsmola_ has quit IRC | 15:19 | |
| NobodyCam | Shrews: lots of mocking ?? :-p | 15:19 |
| * NobodyCam ducks | 15:20 | |
| Shrews | heh, yeah. just spitballing ideas | 15:21 |
| *** dwalleck has quit IRC | 15:28 | |
| lucasagomes | lifeless, our disk partitioning function is completely broken | 15:30 |
| lucasagomes | lifeless, if you do a deploy with root size = 1GB and ask for 1GB of swap the partitioning breaks | 15:30 |
| lucasagomes | (even if the disk supports it) | 15:31 |
| lucasagomes | lifeless, check it out: http://paste.openstack.org/show/74363/ | 15:32 |
| lucasagomes | devananda, ^ | 15:32 |
| lucasagomes | it's only working because somehow having 1MB of swap works, the sfdisk doesn't complain | 15:32 |
| * lucasagomes files a bug | 15:33 | |
| *** lsmola_ has joined #openstack-ironic | 15:34 | |
| *** vkozhukalov has quit IRC | 15:34 | |
| *** ifarkas has quit IRC | 15:36 | |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Encapsulate Ironic client retry logic https://review.openstack.org/83105 | 15:37 |
| jroll | agordeev: thanks for the patch :) lgtm at a glance, I'll check it out in detail later this morning | 15:37 |
| Shrews | 2 hour limit on parking about up. back online in a bit | 15:39 |
| *** ifarkas has joined #openstack-ironic | 15:40 | |
| *** saju_m has quit IRC | 15:43 | |
| russell_h | agordeev: jroll: this looks like about what I had in mind for killing modes too | 15:44 |
| russell_h | agordeev: thanks! | 15:44 |
| *** ifarkas has quit IRC | 15:48 | |
| *** romcheg has quit IRC | 15:49 | |
| rloo | jroll: when you have a minute, wrt 81645, rabbit_virtual_host -- what did you want it to be set to? default is '/'. | 15:51 |
| *** matty_dubs is now known as matty_dubs|lunch | 15:58 | |
| *** killer_prince has quit IRC | 15:59 | |
| *** killer_prince has joined #openstack-ironic | 15:59 | |
| *** eghobo has joined #openstack-ironic | 16:04 | |
| *** eghobo has quit IRC | 16:04 | |
| lucasagomes | NobodyCam, you have an enviroment handy there? | 16:05 |
| *** eghobo has joined #openstack-ironic | 16:05 | |
| *** eghobo has quit IRC | 16:05 | |
| *** eghobo has joined #openstack-ironic | 16:06 | |
| *** lsmola_ has quit IRC | 16:06 | |
| devananda | JoshNang: hi! are you still working on https://bugs.launchpad.net/ironic/+bug/1277555 ? | 16:09 |
| devananda | JoshNang: it looks likeyou posted a fix then abandoned it, but i dont see any other fix for this bug | 16:09 |
| devananda | JoshNang: I've unassigned that bug as you abandoned the patch sets targeting it | 16:15 |
| *** killer_prince has quit IRC | 16:17 | |
| *** lazy_prince has joined #openstack-ironic | 16:18 | |
| *** lazy_prince is now known as killer_prince | 16:18 | |
| *** lsmola_ has joined #openstack-ironic | 16:20 | |
| JoshNang | devananda: ahh! I must have lost that in my email. I can take a stab at your comments today. | 16:20 |
| devananda | lucasagomes: https://bugs.launchpad.net/ironic/+bug/1288007 is assigned to you and InProgress with no patch set. maybe just an issue with jenkins... can you update? | 16:20 |
| devananda | JoshNang: great! thanks | 16:20 |
| lucasagomes | devananda, oh I think the patch is abandoned | 16:21 |
| lucasagomes | lemme revive it | 16:21 |
| JayF | devananda: lucasagomes: I've started at all on that bug, but haven't made appreciable progress. If one of you want to pick it back up, feel free, otherwise I'll likely work on it again this afternoon. | 16:21 |
| lucasagomes | JayF, go ahead then | 16:22 |
| lucasagomes | JayF, I'm looking at another problem with the disk partitioning now | 16:22 |
| lucasagomes | thanks | 16:22 |
| devananda | lucasagomes: do you feel https://bugs.launchpad.net/ironic/+bug/1297871 is worth holding up the RC for? | 16:22 |
| *** romcheg has joined #openstack-ironic | 16:23 | |
| devananda | or https://bugs.launchpad.net/ironic/+bug/1288007 for that matter | 16:23 |
| lucasagomes | devananda, so, the reason why I'm looking at this disk partitoning is related to that bug | 16:23 |
| lucasagomes | I want to get rid of the swap partition first | 16:23 |
| lucasagomes | because just by changing from mb to gb | 16:23 |
| lucasagomes | we will end up eating 1gb from the disk for the swap | 16:23 |
| lucasagomes | oh, or not really... if we set swap_mb = 1 | 16:24 |
| lucasagomes | hmm /me confused | 16:24 |
| devananda | lucasagomes: no, you're right. for testing in VMs, forcing swap to be measured in GB, but not providing support for swap_gb=0, means we eat up more disk | 16:25 |
| devananda | lucasagomes: so those two issues are related | 16:25 |
| lucasagomes | devananda, yeah but we can work around in the code | 16:25 |
| lucasagomes | devananda, we convert it from gb to mb | 16:25 |
| devananda | lucasagomes: i'd add support for no swap then change swap_mb -> swap_gb | 16:25 |
| lucasagomes | devananda, so if it's 0 we can do swap_mb = 1 | 16:26 |
| lucasagomes | devananda, ack | 16:26 |
| lucasagomes | I will work on that | 16:26 |
| NobodyCam | lucasagomes: not atm | 16:26 |
| devananda | lucasagomes: but mapping 0 => 1 is a really bad hack, and users would be very confused. "Hey, I created a flavor with 0 swap, why is there still a swap partition?" | 16:26 |
| rloo | in devstack, nova.conf, [ironic], I have api_endpoint=http://...:6358/v1 | 16:26 |
| rloo | but ironic endpoint is 6385 | 16:27 |
| devananda | lucasagomes: sounds like good things to fix but not RC-blockers | 16:27 |
| devananda | rloo: that sounds like a typo | 16:27 |
| lucasagomes | devananda, +1, yeah I changed the importance of https://bugs.launchpad.net/ironic/+bug/1273785 today cause of that | 16:27 |
| devananda | lucasagomes: cool, thx | 16:27 |
| rloo | so is devstack setup actually using that api_endpoint? | 16:27 |
| lucasagomes | devananda, ok but will we land it while in feature freeze? | 16:27 |
| lucasagomes | or we are going to hold it until juno opens? | 16:28 |
| devananda | lucasagomes: juno will open as soon as we tag an RC | 16:28 |
| devananda | lucasagomes: so it could be next few days or early next week | 16:28 |
| lucasagomes | devananda, ah, alright | 16:28 |
| lucasagomes | sounds good | 16:28 |
| devananda | hm, as a bit of context for why we're doing this RC thing | 16:29 |
| devananda | we fork and open development early | 16:29 |
| devananda | so juno gets more attention before the summit | 16:29 |
| devananda | but | 16:29 |
| devananda | the cost is that back-porting fixes to the RC branch(es) will be harder | 16:30 |
| devananda | so we should try to get as many of the "critical" fixes for icehouse in before we tag the RC | 16:30 |
| lucasagomes | gotcha | 16:30 |
| devananda | there will be a separate process for back-porting bug fixes after the tag is made | 16:30 |
| lucasagomes | I see, that sounds good, and I can imagine that other people depends on that | 16:31 |
| lucasagomes | for e.g people documenting stuff | 16:31 |
| devananda | with a smaller review team (basically me at the moment, i should add some people to it!) | 16:31 |
| lucasagomes | heh +1 | 16:31 |
| devananda | but that team is necessarily small | 16:31 |
| devananda | because of the impact of any code landing on the RC branches | 16:32 |
| devananda | lucasagomes: not just on doc ppl. also down-stream packagers | 16:32 |
| *** epim has joined #openstack-ironic | 16:34 | |
| lucasagomes | yeah | 16:35 |
| lucasagomes | devananda, there's a nice post from thierry that I saw while ago about the feature freeze | 16:36 |
| lucasagomes | devananda, http://fnords.wordpress.com/2014/03/06/why-we-do-feature-freeze/ | 16:36 |
| rloo | devananda. wrt the 6358 typo in devstack -- should i just submit a review to fix that, and/or are we accumulating changes somewhere? I'm not sure that this is a high priority. | 16:37 |
| devananda | rloo: that sounds like a bug in devstack, but I dont see how it could be working (and clearly it is for many people) | 16:41 |
| devananda | adam_g: ^ ? | 16:42 |
| rloo | devananda. yeah, I also wonder. | 16:43 |
| rloo | devananda: my guess is that the driver is getting the ironic endpoint from keystone | 16:46 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Nova Ironic driver to set pxe_swap_mb in Ironic https://review.openstack.org/83131 | 16:48 |
| devananda | rloo: ah, i see. that option is only used if [ironic] admin_auth_token is set | 16:48 |
| devananda | rloo: so yes, it's getting the API endpoint from keystone | 16:48 |
| devananda | rloo: still, it is worth fixing that in devstack, and it should be trivial and easy for them to accept | 16:49 |
| rloo | devananda. ok, will do. just didn't want to give devstack more work to do with one little change :-) | 16:49 |
| rloo | devananda. and I don't plan on testing it. shhhh... | 16:50 |
| devananda | heh | 16:50 |
| NobodyCam | bad rloo :-P | 16:50 |
| rloo | NobodyCam. you weren't supposed to listen ;) | 16:51 |
| *** vkozhukalov has joined #openstack-ironic | 16:51 | |
| devananda | afk a bit | 16:51 |
| *** matty_dubs|lunch is now known as matty_dubs | 16:53 | |
| NobodyCam | hehehe | 16:54 |
| jroll | rloo: default of '/' is fine for rabbit_virtual_host, I just want to see it mentioned in there | 16:55 |
| rloo | jroll: hmm. there are other things that aren't mentioned cuz they are default. hard to know which are more important. let me think about it. | 16:56 |
| jroll | sure | 16:56 |
| jroll | it's just something I ran into while doing integration that would have been nice | 16:57 |
| jroll | nice to know* | 16:57 |
| *** harlowja_away is now known as harlowja | 16:58 | |
| *** gpizza has joined #openstack-ironic | 17:03 | |
| *** ndipanov has quit IRC | 17:07 | |
| *** jistr has quit IRC | 17:10 | |
| *** gpizza has quit IRC | 17:13 | |
| JoshNang | devananda: when you get back, I left some comments on that patchset I'd like some input on (no rush) https://review.openstack.org/#/c/73121/ | 17:14 |
| devananda | JoshNang: gah, sorry about that ... | 17:24 |
| JoshNang | devananda: no worries! i think both approaches have merits | 17:35 |
| *** romcheg has quit IRC | 17:35 | |
| JoshNang | but as long as you're ok with the conductor getting suppressed DB errors, I think db api is the place to catch them | 17:36 |
| russell_h | devananda: on this: https://review.openstack.org/#/c/82915/ | 17:41 |
| devananda | JoshNang: replied to https://review.openstack.org/#/c/73121/ | 17:41 |
| russell_h | devananda: I think its important that the API at least periodically refresh its hash rings | 17:41 |
| russell_h | I was trying not to do that because we'd been talking about the idea that a rebalance needs to be retriggered manually | 17:41 |
| devananda | russell_h: api keeping hash rings up to date ++ | 17:41 |
| devananda | russell_h: use a broadcast | 17:42 |
| devananda | i think I started to put a scaffolding together... I can try to dig up that code if needed | 17:42 |
| devananda | russell_h: but basically, use an rpc.fanout_cast | 17:42 |
| russell_h | devananda: do we do a broadcast every time a conductor comes up? | 17:42 |
| devananda | when the hash ring changes to tell everyone to rebalance | 17:42 |
| devananda | russell_h: nope. we should do taht during start and stop, tho | 17:42 |
| russell_h | so IMO this needs a little more thought | 17:43 |
| devananda | russell_h: this is what i meant when I said, the ring rebalance stuff needs work :) | 17:43 |
| russell_h | yeah :) | 17:43 |
| russell_h | I'd like to land what amounts to caching with an expiration now | 17:43 |
| russell_h | I see lack of any sort of caching there as a bug that is likely to have real impact on use of Ironic | 17:44 |
| russell_h | then give a lot more love to hash rings for juno | 17:44 |
| devananda | it'll impact API performance at scale, yes | 17:44 |
| devananda | but shouldn't affect functionality | 17:44 |
| russell_h | right | 17:44 |
| russell_h | it probably also depends how many drivers/conductors you have | 17:45 |
| devananda | russell_h: whereas cache-with-ttl that is not sync'd across API services will introduce non-deterministic behavior | 17:45 |
| russell_h | I see that as unavoidable | 17:46 |
| devananda | i think it's entirely avoidable if we do the rebalance logic correctly | 17:46 |
| russell_h | you can minimize it with a broadcast by making it likely every service will see an update really really soon | 17:46 |
| russell_h | but there is always going to be a window during which two services have different views of the hashring | 17:46 |
| russell_h | (unless you get really crazy with a multi-phased commit sort of thing) | 17:47 |
| russell_h | (which is a bad idea) | 17:47 |
| devananda | using fanout_cast will reduce it from minutes to seconds in most cases | 17:47 |
| devananda | using zookeeper (or another quorum mgr) will eliminate it | 17:47 |
| devananda | we need to avoid split-brain scenarios too | 17:47 |
| devananda | which fanout_cast will not accomplish | 17:48 |
| russell_h | even with zookeeper, clients can "lag" | 17:48 |
| russell_h | an individual client may will always have a view of the world that existed at some point in time, and it will always receive events in order, but its not guaranteed to be up to date | 17:48 |
| russell_h | how does fanout_cast avoid a split brain? | 17:49 |
| devananda | russell_h: i think it's premature to introduce something we know will add non-deterministic bugs just to remove what is essentially one db query today. | 17:49 |
| devananda | i meant, fanout_cast will not avoid split brain | 17:49 |
| russell_h | ah :) | 17:49 |
| russell_h | oh | 17:49 |
| russell_h | you said that | 17:50 |
| russell_h | devananda: so in my view, today it is eventually consistent | 17:50 |
| russell_h | if you add a conductor, APIs will recognize it "soon" | 17:50 |
| russell_h | other conductors will never recognize it | 17:50 |
| devananda | russell_h: today, all api servies have a synchronous view of the ring | 17:50 |
| russell_h | sure, for all requests that start after the conductor has registered | 17:51 |
| russell_h | the window of inconsistency is _very_ short | 17:51 |
| devananda | so at any time T, no matter which api service responds to it, which conductor it routes to is determined by ring(T) | 17:51 |
| devananda | whereas your patch will change that consistency | 17:51 |
| russell_h | sure, if you externally order: | 17:52 |
| devananda | what the conductor believes is mapped to it is actually /not/ important in the context of where API requests get routed | 17:52 |
| russell_h | 1. adding a conductor | 17:52 |
| russell_h | 2. doing a request | 17:52 |
| devananda | conductor will do what it's asked for | 17:52 |
| russell_h | you have deterministic behavior today | 17:52 |
| russell_h | for for requests that are in-flight it isn't deterministic | 17:52 |
| russell_h | so you really can't rely on deterministic behavior | 17:52 |
| russell_h | and, as far as I can tell, we don't | 17:52 |
| devananda | the conductor's view of the ring is important only for a) precaching things to optimize take over for another conductor, b) take over a running instance from another condcutor | 17:52 |
| devananda | which is not implemented yet | 17:53 |
| russell_h | right | 17:53 |
| devananda | "requests that are in-flight it isn't deterministic" please explain | 17:53 |
| russell_h | if I make 2 requests to the API | 17:53 |
| russell_h | and meanwhile someone adds a conductor | 17:53 |
| russell_h | the requests are for the same node | 17:54 |
| *** epim has quit IRC | 17:54 | |
| russell_h | it is possible for them to go to different conductors | 17:54 |
| devananda | yes | 17:54 |
| devananda | that's fine | 17:54 |
| russell_h | its even possible for the first request to go to the "new" conductor and the second one to go to the "old" one | 17:54 |
| devananda | ... how? | 17:54 |
| russell_h | (as long as by "first" and "second", we we're talking about when they're initiated - not if they're sequential) | 17:55 |
| devananda | so the window for that to happen is incredibly small | 17:55 |
| russell_h | right | 17:55 |
| devananda | it would require the API service's green threads to execute SQL queries in the reverse order | 17:56 |
| russell_h | yeah, or just 2 API services or whatever | 17:56 |
| devananda | thd1: SELECT ; thd2: SELECT ; thd2: <get response> ; thd1: <get response> | 17:56 |
| russell_h | the window is more-or-less the amount of time it takes the API to service a request | 17:56 |
| devananda | no | 17:56 |
| russell_h | if I have 2 API services | 17:56 |
| russell_h | and one request gets routed to each | 17:57 |
| devananda | it's the time it takes the API to start the transaction of the first query it makes after receiving teh API request | 17:57 |
| *** saju_m has joined #openstack-ironic | 17:57 | |
| russell_h | its entirely possible for one API to be doing a GC or something and not even accept the TCP connection until after the other one has serviced its request | 17:57 |
| devananda | right | 17:57 |
| devananda | russell_h: so that will be synchronized on the back end anyway | 17:57 |
| devananda | the small time window conflict here is not a problem | 17:57 |
| devananda | because conductors synchronize locking of nod resources | 17:58 |
| russell_h | exactly | 17:58 |
| russell_h | so I'm advocating that since we rely on that anyway | 17:58 |
| devananda | the issue is when they are far enough out of sync that action 1 is completed on conductor A and action 2 is set to conductor B | 17:58 |
| russell_h | we make the window somewhat larger | 17:58 |
| devananda | which implies a substantial time window | 17:58 |
| devananda | whcih is what you're proposing | 17:58 |
| devananda | :) | 17:58 |
| devananda | and why i object to it | 17:58 |
| russell_h | heh | 17:59 |
| devananda | the current error-margin is "time for a SQL query" and the current sync-window is "time for a SQL query" | 17:59 |
| devananda | you're changing one without the other | 17:59 |
| devananda | the DB is currently the lock mechanism AND the sync mechanism. take one away, but not the other, and I think there's a substantial problem | 18:00 |
| devananda | (now, you could argue that DBs don't make good lock managers, and you'd be right, but they do make simple ones) | 18:00 |
| *** max_lobur has joined #openstack-ironic | 18:00 | |
| *** epim has joined #openstack-ironic | 18:00 | |
| *** athomas has quit IRC | 18:02 | |
| *** lucasagomes is now known as lucas-afk | 18:02 | |
| russell_h | the sync-window really is longer than that though | 18:02 |
| russell_h | its still short | 18:02 |
| russell_h | but its not like an entire request is serviced inside of one SQL transaction or something | 18:02 |
| devananda | no, but that doesn't matter | 18:03 |
| *** max_lobur1 has quit IRC | 18:03 | |
| devananda | maybe i'm confused | 18:03 |
| *** derekh has quit IRC | 18:04 | |
| russell_h | devananda: want me to write an email about it? | 18:04 |
| russell_h | I think I'm better at longer-form | 18:04 |
| devananda | russell_h: sure. but pls consider that I'm focused on Icehouse release candidates this week | 18:05 |
| openstackgerrit | Russell Haering proposed a change to openstack/ironic: Re-use hash rings across requests https://review.openstack.org/82915 | 18:05 |
| russell_h | devananda: for sure | 18:05 |
| devananda | russell_h: also to clarify, I think it's a good step towards what we should be doing in Juno | 18:05 |
| russell_h | ok | 18:06 |
| russell_h | but you're -1 to trying to optimize the API for icehouse? | 18:06 |
| russell_h | or -2 or whatever OS does | 18:06 |
| devananda | -2 for introducing what I see as a bug, in its current state | 18:09 |
| devananda | totally OK with continued work on it for Juno | 18:09 |
| russell_h | k | 18:09 |
| devananda | comment added | 18:12 |
| devananda | and ugh, my lunch appt just got moved up, so I'll be afk for a few hours | 18:13 |
| *** martyntaylor has left #openstack-ironic | 18:15 | |
| *** max_lobur1 has joined #openstack-ironic | 18:23 | |
| *** max_lobur has quit IRC | 18:26 | |
| *** vkozhukalov has quit IRC | 18:29 | |
| lifeless | lucas-afk: \o/ thats sad | 18:31 |
| lucas-afk | lifeless, heh it's also broken in nova bm | 18:34 |
| lucas-afk | lifeless, http://paste.openstack.org/show/74376/ | 18:34 |
| lucas-afk | lifeless, I will fix it in Ironic and back port to nova | 18:34 |
| lifeless | 2014-03-26 16:16:40.363 | | 18:37 |
| lifeless | 2014-03-26 16:16:40.363 | -# Optionally specify a list of memcached server(s) to use for | 18:37 |
| lifeless | 2014-03-26 16:16:40.364 | -# caching. If left undefined, tokens will instead be cached | 18:37 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Provide a new ComputeManager for Ironic https://review.openstack.org/82637 | 18:39 |
| openstackgerrit | lifeless proposed a change to openstack/ironic: Provide a new ComputeManager for Ironic https://review.openstack.org/82637 | 18:39 |
| lifeless | this sample conf stuff really shits me :) | 18:40 |
| comstud | hehe | 18:41 |
| lifeless | lucas-afk: random thought | 18:42 |
| lifeless | lucas-afk: not for nova-bm | 18:43 |
| lifeless | lucas-afk: but make swap partition == 0 -> no swap made at all ? | 18:43 |
| lucas-afk | lifeless, def, I have a bug about it assigned to me | 18:43 |
| lucas-afk | will be part of the fix | 18:43 |
| lucas-afk | lifeless, also changing swap_mb to swap_gb in ironic | 18:43 |
| lucas-afk | for consistency | 18:43 |
| lifeless | \o/ | 18:43 |
| lucas-afk | :) | 18:44 |
| openstackgerrit | Adam Gandelman proposed a change to openstack/ironic: Raise unexpected exceptions during destroy() https://review.openstack.org/82948 | 18:46 |
| NobodyCam | house got a mostly clean bill of health from the mechinic :) | 18:53 |
| *** max_lobur1 has quit IRC | 18:57 | |
| *** epim has quit IRC | 19:01 | |
| *** epim has joined #openstack-ironic | 19:10 | |
| JayF | lifeless: ++ I agree, there should be some method to not want a swap partition made | 19:28 |
| rloo | NobodyCam: do you know if ironic uses/needs the nova_bm DB table in nova? (baremetal uses that) | 19:41 |
| NobodyCam | it dose not.. .or shouldn't | 19:41 |
| NobodyCam | :) | 19:41 |
| NobodyCam | does even | 19:41 |
| rloo | NobodyCam: ok thx. in devstack, it looks like we/ironic is setting nova's ironic's section: sql_connection= ../nova_bm. | 19:42 |
| NobodyCam | oh that seems wrong. Is it creating the ironic DB? | 19:42 |
| rloo | NobodyCam: I didn't know if it should be setting it, and if it did, why it didn't also create the table. | 19:42 |
| rloo | NobodyCam. I'll check. I think I checked already but can't keep track... | 19:43 |
| NobodyCam | :) | 19:43 |
| rloo | NobodyCam: yup. in init_ironic, the ironic db is created. | 19:45 |
| adam_g | NobodyCam, what patches are in-flight to enable tripleO + Ironic? i'm going to start banging on it and keeping notes at https://etherpad.openstack.org/p/IronicAndTripleO | 19:45 |
| NobodyCam | adam_g: https://review.openstack.org/#/c/81959 https://review.openstack.org/#/c/81627 https://review.openstack.org/#/c/82637 | 19:46 |
| NobodyCam | I'm encountering a nova startup issue with seed vm atm | 19:47 |
| adam_g | NobodyCam, cool! might pick your brain when i get organized | 19:47 |
| NobodyCam | Just started to dig in to it. but initial look seems like the compute_manager fix for the undercloud is not getting set for seed.. I need to see if this is the actual issue or not | 19:48 |
| NobodyCam | adam_g: pick away | 19:48 |
| adam_g | NobodyCam, okay first! how do i get devtest running using specific patches from gerrit? or do i just get everything pulled down to the right place beforehand? | 19:51 |
| NobodyCam | ahh thats the trick... it's actually easy | 19:51 |
| NobodyCam | # install patches with dib | 19:51 |
| NobodyCam | DIB_REPOLOCATION_ironic=https://review.openstack.org/openstack/ironic | 19:51 |
| NobodyCam | DIB_REPOREF_ironic=refs/changes/37/82637/9 | 19:51 |
| NobodyCam | note ^^^ can be set per project/repo | 19:52 |
| adam_g | NobodyCam, ah nice | 19:52 |
| NobodyCam | for the devtest patches just apply them to before you start devtest | 19:53 |
| adam_g | right | 19:53 |
| NobodyCam | then set: export USE_IRONIC=1 | 19:53 |
| NobodyCam | and away you go | 19:53 |
| rloo | adam_g: maybe you can give some guidance on devstack changes. Just discovered that ironic doesn't need to set something in nova's config. Should I open a bug for that, or wait/accumulate a bunch of changes? The sql_connection isn't used so it doesn't matter if it is there right now, just need to remove it so it isn't confusing. | 19:54 |
| NobodyCam | need to reboot brick...brb | 19:56 |
| adam_g | rloo, maybe just keep a list going.. if its all trivial stuff it would probably be easier to fix in one patch | 19:56 |
| rloo | adam_g: ok. (Esp since I already submitted a bug/fix for devstack. don't want to be a PITA.) Do we have such a list anywhere? | 19:57 |
| adam_g | rloo, you can probably tack it on to https://etherpad.openstack.org/p/IronicAndDevstackAgain | 19:58 |
| adam_g | rloo, whats the review for the other you submitted? | 19:58 |
| rloo | thanks adam_g! | 19:58 |
| rloo | adam_g: https://review.openstack.org/83137 | 19:58 |
| adam_g | rloo, oh, nice | 19:59 |
| rloo | adam_g: a nice low-hanging-fruit too but deva thought it was worth fixing sooner rather than later. doesn't break devstack now cuz the code is using keystone to get the api url. | 20:00 |
| *** mrda_away is now known as mrda | 20:01 | |
| adam_g | NobodyCam, do i need this one too? https://review.openstack.org/#/c/78467/15 | 20:07 |
| *** saju_m has quit IRC | 20:11 | |
| *** epim has quit IRC | 20:13 | |
| NobodyCam | adam_g: 81959 should have that set as a dependency | 20:14 |
| *** tatyana has joined #openstack-ironic | 20:21 | |
| *** hemna has joined #openstack-ironic | 20:22 | |
| *** saju_m has joined #openstack-ironic | 20:29 | |
| *** saju_m has quit IRC | 20:37 | |
| adam_g | NobodyCam, thoe DIB_REPOREF_* works for gerrit reviews? doesn't seem to be for me. pull sources doesn't look like it'd work either: git checkout $REF # for a branch or SHA1 | 20:43 |
| NobodyCam | adam_g: your tring to pull in 82637 | 20:44 |
| NobodyCam | ? | 20:44 |
| NobodyCam | this is what I set to pull that in: | 20:45 |
| NobodyCam | DIB_REPOLOCATION_ironic=https://review.openstack.org/openstack/ironic | 20:45 |
| NobodyCam | DIB_REPOREF_ironic=refs/changes/37/82637/9 | 20:45 |
| adam_g | NobodyCam, trying to pull in https://review.openstack.org/#/c/81627 to ~/.cache | 20:45 |
| openstackgerrit | A change was merged to openstack/ironic: Provide a new ComputeManager for Ironic https://review.openstack.org/82637 | 20:45 |
| adam_g | NobodyCam, trying to pull in tripleo-image-elements review change during initial bootstrapping of devtest (devtest_setup.sh) | 20:46 |
| NobodyCam | adam_g: allpy that to the tripleo-image-elements dir | 20:46 |
| NobodyCam | apply even | 20:46 |
| JayF | devananda: as part of the fix for https://bugs.launchpad.net/ironic/+bug/1265587, we're going to need to add some strings (when we return a 'clean' error vs the exact exception from the vendor), but in the earlier merge req. you said to avoid that during feature freeze. Why is it OK here? What is different? | 20:47 |
| NobodyCam | if your image-elemets is not on master branch it will not pull it | 20:47 |
| adam_g | NobodyCam, ah, ok | 20:50 |
| rloo | JayF: I'm guessing he means to avoid making string changes as much as possible. if it really needs to be changed, then it needs to be changed :-) | 20:56 |
| rloo | JayF: cuz it means more work for the translators. | 20:56 |
| JayF | Yeah I just like to know where the line is for 'really needs to be changed' ;) | 20:58 |
| rloo | JayF: change it and see what the reviewers say ;) | 20:59 |
| JayF | hah, that's exactly the type of subjective definition I was trying to avoid :D | 20:59 |
| rloo | JayF. The problem being that if you don't change and a reviewer that didn't know what Deva said, sees it, they may say, change it. So you most likely can't win ;) | 21:00 |
| JayF | That's mostly what happens with things that are subjectively defined. | 21:01 |
| rloo | JayF. OR I guess you can wait for deva and then mention 'deva said...' ;) | 21:01 |
| * NobodyCam still wants tlh translations | 21:01 | |
| *** epim has joined #openstack-ironic | 21:05 | |
| *** lifeless has quit IRC | 21:19 | |
| *** tatyana has quit IRC | 21:28 | |
| rloo | NobodyCam: for the ironic driver code, are you/we opening bugs for that? the help for admin_url is incorrect. | 21:28 |
| NobodyCam | rloo: just the help doc string | 21:30 |
| NobodyCam | ? | 21:30 |
| rloo | NobodyCam: yeah. help='Ironic public api endpoint.' | 21:31 |
| NobodyCam | I think I would just put up the patch | 21:31 |
| rloo | NobodyCam. I don't care if it is fixed soon, cuz no one will look til after it gets into nova I think. but don't want to forget to fix it. | 21:31 |
| NobodyCam | and we can land it right after rc is cut | 21:32 |
| rloo | NobodyCam. I'll just open a bug about it then. it is low hanging fruit for some newbie. | 21:32 |
| NobodyCam | :-p | 21:32 |
| NobodyCam | ack :) Ty rloo | 21:32 |
| Shrews | FYI: iterative testing using devstack is a slooooooow march to insanity | 21:34 |
| NobodyCam | Shrews: lol | 21:34 |
| * devananda is back | 21:35 | |
| NobodyCam | wb devananda :) | 21:35 |
| devananda | JayF: hm, so, you're right. we will need to change i18n strings for that bug | 21:36 |
| JayF | So what does that mean? | 21:37 |
| *** matty_dubs is now known as matty_dubs|gone | 21:46 | |
| devananda | JayF: there is not, afaik, an objective guideline for when a project should allow string changes post-FF. but I can point at this ML thraed | 21:47 |
| devananda | http://lists.openstack.org/pipermail/openstack-dev/2014-March/029868.html | 21:47 |
| JayF | devananda: what's your thought in this case? | 21:49 |
| devananda | which suggests fixing bugs in strinsg (like incorrect help messages) is OK, but should include a note to the i18n team | 21:49 |
| devananda | but adding new strings should be avoided | 21:50 |
| JayF | so jfdi, but make sure to ping someone on the i18n team to review it? | 21:52 |
| NobodyCam | so we can fix things like the issue rloo found. :) | 21:52 |
| devananda | NobodyCam: yes | 21:52 |
| rloo | IF we think we want it in for icehouse-rc1? | 21:53 |
| devananda | JayF: our position as an incubated (not integrated) project is that we have more leeway here, but also, are being watched to see how closely we can match the expectations of an integrated project | 21:53 |
| devananda | so it's tricky and not al lthat clear to me, frankly | 21:53 |
| JayF | Well I'm just doing the final steps of debugging the fix | 21:54 |
| JayF | I nearly had to start from scratch because rebasing what existed was pretty near impossible given how much those files had changed | 21:54 |
| JayF | so I'll push the patch, then ask someone in i18n for their opinion in the review, and see where it goes? | 21:54 |
| devananda | JayF: ++ | 21:56 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic: Encapsulate Ironic client retry logic https://review.openstack.org/83105 | 21:58 |
| *** dhellmann is now known as dhellmann_ | 22:03 | |
| *** blamar has quit IRC | 22:03 | |
| *** lynxman has quit IRC | 22:04 | |
| *** lynxman has joined #openstack-ironic | 22:04 | |
| rloo | devananda: when generating the sphinx doc, I see errors (not related to the doc I am adding). Do you think it is worth looking into before rc-1? | 22:05 |
| devananda | rloo: https://bugs.launchpad.net/ironic/+bug/1277282 :) | 22:07 |
| devananda | rloo: in short, no. but worth fixing eventually | 22:07 |
| devananda | rloo: I'd like to get the doc buld down to 0 errors, then turn it in in our check tests | 22:07 |
| rloo | devananda. ok, good I mean good, I don't need to look at them in the near future ;) | 22:07 |
| rloo | devananda: that makes sense. maybe we could put 'low-hanging-fruit' on it, but that might be cruel. | 22:08 |
| devananda | haha! graeet idea | 22:08 |
| *** lynxman has quit IRC | 22:11 | |
| Shrews | w00t. new retry code works, at least | 22:12 |
| *** lynxman has joined #openstack-ironic | 22:12 | |
| JayF | devananda: I think you have to reopen 64711 before I can push my patchset to it | 22:13 |
| JayF | devananda: or would you rather me make a new change-id? | 22:13 |
| * JayF brb | 22:15 | |
| devananda | JayF: you said you nearly had to start from scratch. is there any point in keeping the rev history then? | 22:15 |
| JayF | I honestly have no real picture of when it's something new vs something old. I'll gladly make something new if that's desired? | 22:16 |
| JayF | the end result looks... incomplete so I think it'll need iteration | 22:17 |
| JayF | either that or parts of the bug were fixed as the code was moved around | 22:17 |
| JayF | I'll make it a new change though. And we'll see :) | 22:17 |
| openstackgerrit | Jay Faulkner proposed a change to openstack/ironic: Sanitize node.last_error message strings https://review.openstack.org/83231 | 22:18 |
| JayF | devananda: ^ would appreciate you taking a look at that, I think I missed a couple of things and will go back and check, but imbw or just doubting myself | 22:19 |
| devananda | looking | 22:22 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Install guide for Ironic https://review.openstack.org/81645 | 22:23 |
| NobodyCam | oh lots of white space on that one :) ... reading thru tho | 22:24 |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Install guide for Ironic https://review.openstack.org/81645 | 22:25 |
| rloo | NobodyCam: not so fast ;) | 22:25 |
| NobodyCam | lol | 22:25 |
| NobodyCam | hehehe | 22:25 |
| *** dhellmann_ is now known as dhellmann | 22:31 | |
| NobodyCam | :) # Available in Ubuntu 14.04 (trusty) / # http://packages.ubuntu.com/search?keywords=ironic | 22:31 |
| rloo | NobodyCam: yeah. | 22:33 |
| *** jgrimm has quit IRC | 22:39 | |
| NobodyCam | rloo: I see commit message says it is imcomplete should I mark on the review??? | 22:39 |
| *** dhellmann is now known as dhellmann_ | 22:41 | |
| rloo | NobodyCam: it is incomplete cuz I still have questions and am probably missing stuff. I need feedback! so yes, please mark. | 22:41 |
| NobodyCam | :) | 22:42 |
| *** blamar has joined #openstack-ironic | 22:42 | |
| rloo | I'm not actually sure *when* it will be complete but anyway... | 22:42 |
| rloo | will check later maybe. dinner time. ciao. | 22:42 |
| NobodyCam | enjoy :) | 22:42 |
| *** blamar has quit IRC | 22:53 | |
| openstackgerrit | Adam Gandelman proposed a change to openstack/python-ironicclient: Avoid traceback with insufficient auth credentials https://review.openstack.org/79494 | 22:58 |
| *** harlowja has quit IRC | 22:58 | |
| *** harlowja has joined #openstack-ironic | 23:03 | |
| adam_g | NobodyCam, weirdness | 23:22 |
| adam_g | http://paste.ubuntu.com/7159732/ | 23:24 |
| adam_g | (seed node) | 23:24 |
| * NobodyCam clicks | 23:24 | |
| NobodyCam | I agree WTF | 23:25 |
| adam_g | on the devtest side, i get as far as | 23:25 |
| adam_g | Waiting for seed node to configure br-ctlplane... | 23:25 |
| adam_g | + wait_for 30 10 ping -c 1 192.0.2.1 | 23:25 |
| adam_g | tried rebuilding seed, not much luck | 23:26 |
| NobodyCam | drat my brick shut off again | 23:27 |
| NobodyCam | adam_g: thats a odd place to stop .. the ping timed out? | 23:29 |
| adam_g | NobodyCam, well | 23:29 |
| adam_g | NobodyCam, it ends up waiting for the OVS control plane bridge ot come up in the seed, which it never did | 23:30 |
| adam_g | so if i ssh into seed via its other IP, and manually runos-apply-config (i think it was -apply-config), the OVS bridge eventually comes up but i noticed other scripts fail for issues like keystone-manage | 23:31 |
| adam_g | do i need to set anything other than NODE_ARCH to build amd64 images? | 23:31 |
| NobodyCam | adam_g: I don't think so.. your building ubuntu images? | 23:33 |
| adam_g | NobodyCam, yes, using a local mirror of some saucy cloud images | 23:33 |
| adam_g | i thought devtest was defaulting to i386 but maybe im wrong | 23:33 |
| NobodyCam | yes it should default to i386 | 23:34 |
| NobodyCam | adam_g: I have this ugly hack http://paste.openstack.org/show/CjMfKuAbgNUnoxJYGddu/ | 23:34 |
| NobodyCam | but I would expect the next wait to fail not the ping | 23:34 |
| adam_g | hmm | 23:35 |
| adam_g | NobodyCam, is my understanding correct: create seed -> create image stuffed /w elements -> boot -> os-collect-config at boot -> os-refresh-config -> OVS setup/etc ? | 23:38 |
| NobodyCam | yep and the config o-r-c comes from config.json in seed-stack-config element | 23:41 |
| NobodyCam | s/config o-r-c/config FOR o-r-c/ | 23:41 |
| devananda | adam_g: looking through a recent test run ( http://logs.openstack.org/31/83231/1/check/check-tempest-dsvm-virtual-ironic/ff61a33/logs/screen-ir-cond.txt.gz ) | 23:49 |
| devananda | adam_g: it looks like do_node_teardown was called before do_node_deploy | 23:49 |
| devananda | and after taht tempest probably failed on everything | 23:49 |
| devananda | dunno where you're at, but thought I'd share that observation | 23:49 |
| adam_g | devananda, huh? what context? | 23:51 |
| devananda | adam_g: non-sequitor | 23:51 |
| *** epim has quit IRC | 23:51 | |
| adam_g | devananda is this during a tempest run against devstack+ironic? | 23:51 |
| adam_g | like, tempest as-is today? | 23:51 |
| devananda | yes | 23:52 |
| devananda | from http://logs.openstack.org/31/83231/1/check/check-tempest-dsvm-virtual-ironic/ff61a33/ | 23:52 |
| adam_g | devananda, okay so as far as i can tell, the current scenario tests and anything else that spawns instances are not going to pass against tempest, at least not for a while | 23:53 |
| devananda | adam_g: "for a whille" -- why not? | 23:54 |
| devananda | i thought the plan was to pair down tests to the set that work | 23:54 |
| adam_g | devananda, one sec | 23:54 |
| devananda | so, i had it in mind taht pointing out which ones fail when might be helpful. if that's not the case, sorry for the unhelpful distraction :) | 23:55 |
| *** epim has joined #openstack-ironic | 23:55 | |
| adam_g | devananda, we need to deal with the whole compute driver feature matrix thing first | 23:56 |
| NobodyCam | :) its good to know whats failing | 23:56 |
| devananda | adam_g: so there's no way to test a subset then? | 23:56 |
| adam_g | devananda, if you look at http://logs.openstack.org/31/83231/1/check/check-tempest-dsvm-virtual-ironic/ff61a33/logs/testr_results.html.gz, its likely everything thats failing is failing due to missing features in the driver | 23:56 |
| devananda | *today | 23:56 |
| adam_g | actually thats not entirely true | 23:56 |
| devananda | if we can test a subset, we should identify that subset | 23:57 |
| adam_g | sorry one min, otp | 23:57 |
| devananda | i'm going to assume that we CAN test a subset because we HAVE TO | 23:58 |
| notq | heh | 23:58 |
| NobodyCam | :) we'll make it work! | 23:58 |
| devananda | graduation depends on functional and integration testing -- which is what tempest is for | 23:58 |
| adam_g | devananda, yeah | 23:59 |
| adam_g | https://review.openstack.org/#/c/81351/ | 23:59 |
| devananda | maybe we need to write a new temepst test suite and not use nova's. or something. | 23:59 |
| adam_g | er | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!