| *** foexle has joined #openstack-ironic | 00:03 | |
| *** pcrews has joined #openstack-ironic | 00:16 | |
| *** sseago has joined #openstack-ironic | 00:35 | |
| *** pcrews has quit IRC | 00:41 | |
| *** foexle_ has joined #openstack-ironic | 01:10 | |
| *** jbjohnso has joined #openstack-ironic | 01:10 | |
| *** foexle has quit IRC | 01:13 | |
| *** jbjohnso has quit IRC | 01:37 | |
| *** jbjohnso has joined #openstack-ironic | 01:37 | |
| *** nosnos has joined #openstack-ironic | 01:49 | |
| *** jbjohnso has quit IRC | 01:59 | |
| *** nosnos has quit IRC | 02:09 | |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ironic: Implements send-data-to-ceilometer https://review.openstack.org/72538 | 02:27 |
|---|---|---|
| *** foexle_ has quit IRC | 02:28 | |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ironic: Implements send-data-to-ceilometer https://review.openstack.org/72538 | 02:30 |
| *** ramineni has joined #openstack-ironic | 03:11 | |
| *** bmahalakshmi has joined #openstack-ironic | 03:17 | |
| *** pcrews has joined #openstack-ironic | 03:29 | |
| *** pcrews has quit IRC | 03:36 | |
| *** notsogentle is now known as annegentle | 03:45 | |
| *** geekyogi has joined #openstack-ironic | 03:45 | |
| *** stendulker has joined #openstack-ironic | 03:54 | |
| *** stendulker has quit IRC | 04:02 | |
| *** eghobo has joined #openstack-ironic | 04:04 | |
| *** stendulker has joined #openstack-ironic | 04:16 | |
| *** eghobo has quit IRC | 04:18 | |
| *** eghobo has joined #openstack-ironic | 04:19 | |
| *** nosnos has joined #openstack-ironic | 04:26 | |
| *** stendulker has quit IRC | 04:34 | |
| *** Poornima has joined #openstack-ironic | 04:42 | |
| *** eghobo has quit IRC | 04:51 | |
| *** lazy_prince2 has joined #openstack-ironic | 05:09 | |
| *** k4n0 has joined #openstack-ironic | 05:13 | |
| *** lazy_prince2 has quit IRC | 05:19 | |
| *** bandicot has joined #openstack-ironic | 05:24 | |
| *** eghobo has joined #openstack-ironic | 05:28 | |
| *** shausy has joined #openstack-ironic | 05:42 | |
| *** rakesh_hs has joined #openstack-ironic | 05:46 | |
| *** romcheg has joined #openstack-ironic | 05:57 | |
| *** radsy has quit IRC | 05:58 | |
| *** romcheg has quit IRC | 06:00 | |
| *** geekyogi has quit IRC | 06:10 | |
| *** geekyogi has joined #openstack-ironic | 06:15 | |
| *** nosnos has quit IRC | 06:20 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 06:30 | |
| *** eguz has joined #openstack-ironic | 07:01 | |
| *** Mikhail_D_ltp has quit IRC | 07:02 | |
| *** eguz has quit IRC | 07:03 | |
| *** Mikhail_D_ltp has joined #openstack-ironic | 07:04 | |
| *** eghobo has quit IRC | 07:05 | |
| *** bmahalakshmi2 has joined #openstack-ironic | 07:18 | |
| *** bmahalakshmi has quit IRC | 07:19 | |
| *** jcoufal has joined #openstack-ironic | 07:20 | |
| *** jistr has joined #openstack-ironic | 07:20 | |
| *** ifarkas has joined #openstack-ironic | 07:32 | |
| *** bandicot has quit IRC | 07:36 | |
| *** mrda is now known as mrda-away | 07:40 | |
| GheRivero | morning all | 07:56 |
| dtantsur | Morning Ironic \o/ did you miss me? :) | 08:04 |
| *** derekh__ has joined #openstack-ironic | 08:13 | |
| *** ndipanov has joined #openstack-ironic | 08:14 | |
| *** sseago has quit IRC | 08:19 | |
| *** lucasagomes has joined #openstack-ironic | 08:28 | |
| *** yuriyz has joined #openstack-ironic | 08:34 | |
| yuriyz | morning Ironic | 08:35 |
| *** athomas has joined #openstack-ironic | 08:36 | |
| *** Poornima has quit IRC | 08:41 | |
| dtantsur | yuriyz, morning! | 08:43 |
| gilliard | morninf :) | 08:44 |
| dtantsur | gilliard, morning! | 08:44 |
| *** martyntaylor has joined #openstack-ironic | 08:45 | |
| *** romcheg has joined #openstack-ironic | 08:48 | |
| lucasagomes | morning all | 08:50 |
| dtantsur | lucasagomes, \o/ | 08:52 |
| openstackgerrit | Matthew Gilliard proposed a change to openstack/ironic: Make ComputeCapabilitiesFilter work with Ironic https://review.openstack.org/105802 | 08:52 |
| dtantsur | lucasagomes, please let me know what you think about comments to a discovery spec | 08:52 |
| lucasagomes | sure will take a look | 08:53 |
| dtantsur | lucasagomes, I got at least one request for a full-featured hardware management :D | 08:53 |
| lucasagomes | ew | 08:54 |
| dtantsur | lucasagomes, I'll do the last update today and will be asking you and devananda for fast-forwarding it. Otherwise we'll be arguing forever... | 08:57 |
| lucasagomes | sure yeah that needs to land quickly or we won't get anywhere | 08:57 |
| lifeless | lucasagomes: https://bugs.launchpad.net/nova/+bug/1341420 might scare you :) | 09:01 |
| lucasagomes | :( | 09:01 |
| * lucasagomes looks | 09:01 | |
| lifeless | lucasagomes: ran into it trying to get hp1.tripleo.org back online for CI | 09:02 |
| dtantsur | bbl | 09:05 |
| lucasagomes | ew 4 machines out of 45 it's almost 10 percent (8 something) | 09:08 |
| lucasagomes | and as u said, they all have retried and failed at least 3 times before it's marked as failed | 09:08 |
| lifeless | lucasagomes: that was using 29/45 - so its 4/29 requests fail (> 10%) | 09:08 |
| lifeless | lucasagomes: doing 45 even more fail - like 15/45 | 09:09 |
| lucasagomes | :/ | 09:10 |
| lifeless | yeah | 09:10 |
| lifeless | I've raised the red flag on this within/HP - got some public cloud folk eyeballing it now | 09:10 |
| lucasagomes | if the retry forcebily tried to rescheduler in another host (if resources area available elsewhere) instead of insisting on the same | 09:10 |
| lucasagomes | would minimize it as well | 09:10 |
| lifeless | lucasagomes: nope - it does retry a different host | 09:10 |
| lifeless | it tracks which ones it tried | 09:11 |
| lucasagomes | ah | 09:11 |
| lifeless | all it needs is three hosts outstanding and not yet claimed, and the front three, and the failure can happen in < 1s | 09:11 |
| lucasagomes | i c urghhh | 09:11 |
| *** vinbs has joined #openstack-ironic | 09:12 | |
| lucasagomes | well bright side, CI is working and doing it's job capturing such problems | 09:12 |
| lucasagomes | that would be hardly seem else where on tests | 09:12 |
| lifeless | lucasagomes: actually I found this trying to *deploy* the CI infra :) | 09:12 |
| lifeless | lucasagomes: a little meta :> | 09:12 |
| lucasagomes | ah lol | 09:13 |
| lifeless | so sleep(5) seems to fix it | 09:14 |
| lifeless | which confirms the race aspect of it | 09:14 |
| lifeless | ;) | 09:14 |
| vinbs | Hello Ironic! | 09:16 |
| lucasagomes | heh fair | 09:16 |
| lucasagomes | vinbs, hi there | 09:16 |
| romcheg | Morning Ironic! | 09:20 |
| openstackgerrit | A change was merged to openstack/ironic: Update docstring for api...node.validate https://review.openstack.org/106092 | 09:20 |
| romcheg | lucasagomes: Have you ever hit ERROR: openstackclient.shell Exception raised: six>=1.6.0 when installing Icehouse with Grenade? | 09:21 |
| *** athomas has quit IRC | 09:23 | |
| romcheg | Indeed, when I try running openstackclient manually, I get the same error :( | 09:23 |
| *** lazy_prince has quit IRC | 09:23 | |
| lucasagomes | romcheg, ew not really, but my env is a bit old I haven't build a new one for a while | 09:24 |
| lucasagomes | :( | 09:24 |
| lucasagomes | romcheg, any bug open on the openstackclient? | 09:24 |
| romcheg | lucasagomes: so don't do than :) | 09:24 |
| lucasagomes | lol hah ok | 09:24 |
| romcheg | I'm looking for one | 09:25 |
| *** athomas has joined #openstack-ironic | 09:29 | |
| *** dtantsur has quit IRC | 09:31 | |
| *** dtantsur has joined #openstack-ironic | 09:34 | |
| *** lazy_prince has joined #openstack-ironic | 09:40 | |
| *** Alexei_987 has joined #openstack-ironic | 09:45 | |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic-specs: Generic Hardware Discovery Bits https://review.openstack.org/102565 | 09:46 |
| dtantsur | lucasagomes, ^^^ | 09:48 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic-specs: Generic Hardware Discovery Bits https://review.openstack.org/102565 | 09:49 |
| *** k4n0 has quit IRC | 09:57 | |
| *** pelix has joined #openstack-ironic | 10:03 | |
| soren | I find that I'm having to specify --nic net-id=bblahblah on my "nova boot" command line. Otherwise the instances get no network assigned. Does anyone else see the same thing? With no --nic options, Nova used to create an interface on each network visible to the given tenant. | 10:04 |
| soren | Oh. It does get assgined. Eventually. | 10:06 |
| soren | Never mind. | 10:06 |
| openstackgerrit | A change was merged to openstack/ironic: REST API 'limit' parameter to only accept positive values https://review.openstack.org/106093 | 10:08 |
| openstackgerrit | A change was merged to openstack/ironic: Updates Ironic Guide with deployment information https://review.openstack.org/100787 | 10:08 |
| vinbs | romcheg, I had come across that error when installing openstack client manually | 10:08 |
| openstackgerrit | A change was merged to openstack/ironic: Document 'POST /v1/.../vendor_passthru' https://review.openstack.org/105869 | 10:09 |
| openstackgerrit | A change was merged to openstack/ironic: Update webapi doc with link and console https://review.openstack.org/106102 | 10:09 |
| vinbs | romcheg, I had to upgrade pip to resolve that issue | 10:09 |
| romcheg | vinbs: I have a fresh pip but I will try that | 10:10 |
| vinbs | romcheg, what version of pip do you have? | 10:11 |
| *** k4n0 has joined #openstack-ironic | 10:11 | |
| romcheg | 1.5.6 | 10:11 |
| *** k4n0 has quit IRC | 10:35 | |
| *** martyntaylor has left #openstack-ironic | 10:35 | |
| *** bmahalakshmi2 has quit IRC | 10:49 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic-specs: New driver ManagementInterface https://review.openstack.org/100218 | 10:58 |
| *** ramineni has quit IRC | 11:03 | |
| *** killer_prince has joined #openstack-ironic | 11:04 | |
| *** lazy_prince has quit IRC | 11:07 | |
| *** killer_prince is now known as lazy_prince | 11:07 | |
| *** lazy_prince2 has joined #openstack-ironic | 11:11 | |
| *** lazy_prince2 is now known as killer_prince | 11:11 | |
| *** killer_prince has quit IRC | 11:20 | |
| dtantsur | Core folks, I now, I'm annoying (you missed it, while I was absent, right? ;), but please your opinion on discovery https://review.openstack.org/#/c/102565/ | 11:20 |
| dtantsur | (please read carefully intro on what was deliberately excluded) | 11:20 |
| openstackgerrit | Mikhail Durnosvistov proposed a change to openstack/ironic: Check whether specified FS is supported https://review.openstack.org/98102 | 11:22 |
| *** lucasagomes is now known as lucas-hungry | 11:31 | |
| dtantsur | lucas-hungry, addressed your comment on https://review.openstack.org/#/c/102247/ tl;dr UUID is always an internal attribute in REST API, I'm addressing only DB API itself | 11:35 |
| openstackgerrit | Matthew Gilliard proposed a change to openstack/ironic: Make ComputeCapabilitiesFilter work with Ironic https://review.openstack.org/105802 | 11:36 |
| openstackgerrit | Matthew Gilliard proposed a change to openstack/ironic: Make ComputeCapabilitiesFilter work with Ironic https://review.openstack.org/105802 | 11:37 |
| *** rakesh_hs has quit IRC | 11:52 | |
| *** jdob has joined #openstack-ironic | 12:03 | |
| *** vinbs_ has joined #openstack-ironic | 12:04 | |
| *** lazy_prince has quit IRC | 12:05 | |
| *** vinbs has quit IRC | 12:05 | |
| *** vinbs_ is now known as vinbs | 12:06 | |
| openstackgerrit | A change was merged to openstack/ironic-specs: DRAC power driver https://review.openstack.org/99352 | 12:08 |
| Shrews | morning all | 12:09 |
| Shrews | lucas-hungry: thx for the comments on the node lock retry spec. will address those today :) | 12:09 |
| *** vinbs has quit IRC | 12:10 | |
| Shrews | dtantsur: welcome back! | 12:10 |
| dtantsur | Shrews, morning, thanks :) | 12:10 |
| *** yuriyz has quit IRC | 12:16 | |
| *** geekyogi has quit IRC | 12:16 | |
| *** yuriyz has joined #openstack-ironic | 12:19 | |
| *** rloo has joined #openstack-ironic | 12:20 | |
| *** amitpp has joined #openstack-ironic | 12:20 | |
| *** killer_prince has joined #openstack-ironic | 12:21 | |
| *** killer_prince is now known as lazy_prince | 12:21 | |
| *** lucas-hungry is now known as lucasagomes | 12:36 | |
| lucasagomes | dtantsur, oh my bad, I see the defaults now | 12:37 |
| lucasagomes | will change my vote | 12:37 |
| dtantsur | thnx :) | 12:37 |
| lucasagomes | Shrews, np, and congratz for the nomination :) | 12:38 |
| lucasagomes | jroll, u too congratz :) | 12:39 |
| Shrews | thx | 12:39 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic-specs: New driver ManagementInterface https://review.openstack.org/100218 | 12:48 |
| jroll | lucasagomes: thanks :) | 12:49 |
| jroll | welcome back dtantsur \o/ | 12:49 |
| *** shausy has quit IRC | 12:53 | |
| lucasagomes | dtantsur, on the autodiscovery spec doing a POST {... hardware characteristics ...} /nodes/autodiscovery, isnt the same as POST {... hardware characteristics + maintenance=True + newly_discovered=True} /nodes ? | 12:53 |
| lucasagomes | dtantsur, just wondering if we really new a new endpoint for it | 12:54 |
| lucasagomes | I see the update part now | 12:57 |
| dtantsur | jroll \o/ | 12:59 |
| dtantsur | lucasagomes, we at least should give a driver an ability to jump in + we may not know Node UUID | 12:59 |
| dtantsur | jroll, Shrews, congrats on nominating, well deserved! | 12:59 |
| dtantsur | and jroll I'm gonna annoy you again with discovery spec, please have a look :) https://review.openstack.org/102565 | 13:04 |
| rloo | hello ironickers on this bright and wonderful Monday morning. (zzzzz) | 13:04 |
| rloo | welcome back dtantsur! | 13:04 |
| dtantsur | rloo, hi \o/ | 13:05 |
| rloo | dtantsur, if you have a chance, would you please look at the ipxe spec: https://review.openstack.org/#/c/98822/ | 13:05 |
| dtantsur | sure! | 13:05 |
| rloo | dtantsur: oh wait, well, it'd be good to look at but I think that's a like-to-have for j2. | 13:06 |
| rloo | lucasagomes: is the mgt interface spec a high priority? it seems like lots of things are gating on it. | 13:06 |
| lucasagomes | rloo, yeah | 13:06 |
| lucasagomes | it's needed for a couple of things | 13:06 |
| lucasagomes | discovery | 13:06 |
| lucasagomes | ceilometer data | 13:06 |
| rloo | lucasagomes: i'll take a look at it again. it wasn't mentioned in last week's meeting but I think we should get it approved soon... | 13:07 |
| lucasagomes | +1 | 13:07 |
| rloo | lucasagomes: btw, deva wanted this for J2 too and I said I'd get the code in after the spec was approved, so if you have a chance: https://review.openstack.org/#/c/100218/ | 13:08 |
| dtantsur | rloo, did lucasagomes addressed your concerns on iPXE spec? I'm ok with it and thinking about approving | 13:08 |
| lucasagomes | rloo, oh will do | 13:08 |
| *** faizan has joined #openstack-ironic | 13:09 | |
| rloo | dtantsur: yeah I think. Otherwise I'd have -1'd him. Sec, let me review, my head is a bit foggy still. | 13:09 |
| dtantsur | ack | 13:09 |
| rloo | dtantsur: he sort of answered my questions. i think we'll find out in the future if things need to change, but it is easy to deprecate config options so i'm not worried ;) | 13:10 |
| dtantsur | ok, approved | 13:11 |
| rloo | thx dtantsur. | 13:11 |
| * romcheg wonders how stable jobs are working in OS CI... | 13:12 | |
| lucasagomes | rloo, approved | 13:12 |
| lucasagomes | thanks for addressing the optional/required field thing | 13:12 |
| rloo | thx lucasagomes. finally, maybe I should wait still to see if there are any objections ;) | 13:13 |
| rloo | lucasagomes: thx for simplifying it for me ;) | 13:13 |
| lucasagomes | rloo, wait for jenkins :P | 13:13 |
| *** matty_dubs|gone is now known as matty_dubs | 13:13 | |
| openstackgerrit | A change was merged to openstack/ironic-specs: iPXE boot https://review.openstack.org/98822 | 13:13 |
| lucasagomes | zo/ | 13:13 |
| rloo | lucasagomes: ha ha. All hail the glorious jenkins... | 13:13 |
| openstackgerrit | A change was merged to openstack/ironic-specs: API to get driver_info properties https://review.openstack.org/102914 | 13:14 |
| lucasagomes | rloo, :) ^ | 13:14 |
| rloo | lucasagomes: whoo hoo. Past the first hurdle. Going for 1st base now... | 13:15 |
| lucasagomes | ;) | 13:15 |
| * Shrews giggles at rloo's monday morning email humor | 13:18 | |
| romcheg | Congrats Shrews and jroll! :) | 13:18 |
| Shrews | rloo: gmail seems to think your emails are junk. i tend to disagree. something with emails from yahoo-inc.com. happens with josh too | 13:19 |
| rloo | Shrews: yeah, I was giddy with happiness at more reviewers :D | 13:19 |
| rloo | Shrews: yeah. Yahoo did something and I suspected it'd think it was junk but I am too lazy to change to another non-yahoo-inc email address. | 13:19 |
| rloo | Shrews: I just try never to send email to the list ;) | 13:20 |
| jroll | oh god the puns | 13:20 |
| jroll | JayF will love it | 13:20 |
| Shrews | hehe | 13:20 |
| rloo | jroll, Shrews: congrats. A lot is riding on you two now. ha ha. | 13:20 |
| jroll | dtantsur: question for you... | 13:21 |
| dtantsur | sure | 13:21 |
| jroll | if the inventory field is an opaque json blob | 13:21 |
| jroll | how would ironic tell that the node is already registered? | 13:21 |
| jroll | and is that pluggable? | 13:21 |
| jroll | e.g. if I don't trust mac addresses to be unique (I do at the moment) | 13:22 |
| jroll | also, idk if I agree with that being a 200 | 13:22 |
| jroll | should almost be a 409 (conflict), but we use that everywhere right now :| | 13:23 |
| jroll | (I guess I see the answer for the first question) | 13:23 |
| matty_dubs | rloo: I've noticed that with normal yahoo.com webmail users, too. I think if they don't send through webmail, the SPF headers are bad or something. | 13:24 |
| matty_dubs | rloo: Just in case you wanted to single-handedly fix Yahoo! webmail. | 13:24 |
| jroll | lolll | 13:25 |
| rloo | thx matty_dubs. | 13:27 |
| openstackgerrit | A change was merged to openstack/ironic: Add more INFO logging to ironic/common/service.py https://review.openstack.org/99495 | 13:28 |
| *** jbjohnso has joined #openstack-ironic | 13:30 | |
| rloo | lucasagomes: wrt the mgmt interface and set-boot-device. if it isn't being stored in the db, what happens if the conductor goes away? | 13:31 |
| lucasagomes | jroll, we don't use 409 for when registering nodes | 13:31 |
| lucasagomes | I mean, creating resources in general | 13:32 |
| lucasagomes | rloo, api will try to create a topic to talk to the conductor responsable for that driver | 13:32 |
| lucasagomes | rloo, it will fail saying no conductor in the cluster supports that driver | 13:33 |
| lucasagomes | so the node can't be managed | 13:33 |
| *** Nisha has joined #openstack-ironic | 13:33 | |
| jroll | lucasagomes: what do we use? I don't think we currently (can/do) validate that a resource is unique when creating it | 13:33 |
| lucasagomes | jroll, yeah we don't, just saying about ur argument that we use it everywhere | 13:34 |
| lucasagomes | jroll, just saying that none of the resources creations would return 409 (conflict) right now, because that doesn't required a node lock | 13:34 |
| lucasagomes | it's creating the node | 13:34 |
| lucasagomes | so I think ur suggestion for using 409 if node is already registered is valid | 13:35 |
| rloo | lucasagomes: sorry. I should have been clearer. What happens if someone sets-boot-device for a node/driver, and then the conductor goes away. we're able to get another conductor to handle that node, but the set-boot-device setting will be gone. | 13:36 |
| jroll | lucasagomes: right, we just have a lot of things returning a 409 | 13:36 |
| jroll | in general | 13:36 |
| lucasagomes | yeah | 13:36 |
| dtantsur | jroll, sorry, was out for a moment | 13:37 |
| *** ramineni has joined #openstack-ironic | 13:38 | |
| lucasagomes | rloo, why? it won't... like you had a conductor managing that node and the user set the boot device, request completely. Now this conductor is down and a new one took over that node | 13:38 |
| dtantsur | jroll, 409 is an error message, will cause an error to be logged and so on. In our case there's no error | 13:38 |
| lucasagomes | rloo, the configuration still on the BMC, so the device you set through the previous conductor still valid | 13:38 |
| lucasagomes | (if I got the question) | 13:38 |
| dtantsur | jroll, detecting existing node is not pluggable as of now. I wanted to include it, but dropped to make the spec as simple as possible. | 13:39 |
| dtantsur | I guess it can be introduced later in a backward-compatible manner | 13:39 |
| *** amitpp has quit IRC | 13:40 | |
| rloo | lucasagomes: so the new conductor will know the setting that the user specified via the bmc? including the persistent flag? | 13:40 |
| lucasagomes | rloo, yeah because the conductor is just an abstraction layer, the configuration is set on the BMC | 13:41 |
| jroll | dtantsur: 409 is only an error if you treat it as an error :/ | 13:41 |
| rloo | lucasagomes: ok thx. | 13:41 |
| lucasagomes | rloo, or in the case of the SSH driver, who holds the configuration of which is the default boot device is virsh | 13:41 |
| dtantsur | jroll, you should, 4xx is a family of error codes | 13:41 |
| lucasagomes | rloo, we just go there and ask it | 13:41 |
| *** amitpp has joined #openstack-ironic | 13:42 | |
| jroll | dtantsur: sure... but idk. trying to re-enroll a node that is already enrolled sounds like an error case. unless you treat every re-enroll as a Node.update(), then maybe it's fine? | 13:42 |
| dtantsur | jroll, returning an error code assumes,that we tried not to take any action, when in fact we succeeded in updating hardware parameters. Yes, I treat this endpoint as an update_or_create() rather than just create | 13:43 |
| jroll | ok | 13:44 |
| jroll | fair enough :) | 13:44 |
| jroll | I wasn't thinking about update() when I asked about 409 | 13:44 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic: Raise appropriate errors on duplicate Node and Chassis creation https://review.openstack.org/102506 | 13:45 |
| *** Mikhail_D_ltp has quit IRC | 13:49 | |
| jroll | dtantsur: reviewed | 13:50 |
| jroll | a few more questions, ping me if you want to chat real-time about them | 13:50 |
| GheRivero | nls | 13:51 |
| lucasagomes | jroll, I remember we started talking about #100734 | 13:52 |
| jroll | lucasagomes: hey, were you planning to remove your -1 here? (we talked friday about it) https://review.openstack.org/#/c/100734/ | 13:52 |
| jroll | lol | 13:52 |
| jroll | good timing | 13:52 |
| lucasagomes | lol | 13:52 |
| lucasagomes | yeah | 13:52 |
| lucasagomes | yeah so what's the problem of having tftp_master_path on the pxe group? | 13:53 |
| jroll | lucasagomes: still have questions or are we good to go here | 13:53 |
| jroll | ah | 13:53 |
| jroll | I guess it depends how you think of the pxe group | 13:53 |
| jroll | I think of it as "pxe driver configs" | 13:53 |
| lucasagomes | jroll, there's a nit on that patch anyway that would keep my -1, importing image_cache at tftp.py is not required cause we are not using it | 13:53 |
| jroll | in which case it's weird... because I want to use that config in the agent | 13:54 |
| dtantsur | jroll, thanks. I suggest to postpone auth issue. I understand it's pressing, but I also understand that we can't grow this particular spec infinitely. Also, I'm not too strong in this topic. | 13:54 |
| jroll | lucasagomes: ah, right, pep8 caught that too | 13:54 |
| dtantsur | jroll, I suggest to start discussing it immediately, but in a new spec | 13:54 |
| lucasagomes | jroll, right... yeah... what I think when I see a tftp config in ironic | 13:54 |
| lucasagomes | is that ironic is actually managing a tftp server or something | 13:54 |
| jroll | dtantsur: I just... would -1 the heck out of any implementation that uses the regular auth mechanism | 13:54 |
| lucasagomes | that's why on my patch I was moving back | 13:54 |
| jroll | lucasagomes: right. names are hard. | 13:54 |
| lucasagomes | having it on the pxe group, IMO it represents better that pxe depends on the tftp | 13:55 |
| lucasagomes | jroll, yeah | 13:55 |
| jroll | on the other hand... it configures what directory the tftp server uses, so | 13:55 |
| jroll | or like, what dir to put the files in because the deployer knows tftpd is pointed there | 13:55 |
| lucasagomes | yeah the 2) is more accurate | 13:56 |
| dtantsur | jroll, that's fair, but we _do_ have this auth right now and this spec changes exactly nothing | 13:56 |
| jroll | dtantsur: right... but we don't put admin creds in an image stored in glance today :) | 13:56 |
| jroll | dtantsur: or, at least, I don't. idk about others. | 13:56 |
| dtantsur | jroll, yeah, we put it into TFTP, that's what will be used here as well (as of now) | 13:56 |
| lucasagomes | jroll, and another thing I don't know what to do is, after renaming tftp.py to pxe_utils.py I should keep a 'tftp' config group as part of pxe_utils? | 13:57 |
| jroll | uhhhhhhh | 13:57 |
| dtantsur | jroll, again, I'm all for inventing something better, I only state that this spec does not make it worse | 13:57 |
| jroll | dtantsur: what do you mean by "we put it into TFTP" | 13:57 |
| jroll | dtantsur: as in, we put it in an image, or the pxe options? | 13:58 |
| *** jistr has quit IRC | 13:58 | |
| dtantsur | jroll, we put an admin token on a TFTP for the ramdisk to ping back Ironic after it booted, no? | 13:58 |
| lucasagomes | yes | 13:58 |
| jroll | lucasagomes: hrm, true. maybe we should leave it in pxe :/ | 13:58 |
| lucasagomes | jroll, :/ yeah and it get more confused after I add http_root and http_server | 13:58 |
| jroll | wait, that's a keystone token? I thought that was a per-node token of some sort | 13:58 |
| jroll | lucasagomes: lol | 13:59 |
| dtantsur | jroll, it's a general admin token IIRC | 13:59 |
| lucasagomes | jroll, it gets more inconsistency, because then would I need to add a new 'http' config group? | 13:59 |
| lucasagomes | :( | 13:59 |
| dtantsur | lucasagomes, am I right ^^^? | 13:59 |
| *** jistr has joined #openstack-ironic | 13:59 | |
| lucasagomes | jroll, that's a keystone token yes | 13:59 |
| jroll | lucasagomes: right, hm | 13:59 |
| jroll | gah | 13:59 |
| lucasagomes | yeah the admin token is on the tftp | 13:59 |
| jroll | that's... not awesome | 13:59 |
| lucasagomes | and the ramdisk will suck it and use it to talk to the ironic api | 14:00 |
| lucasagomes | yeah | 14:00 |
| lucasagomes | so, couple of ways to solve that | 14:00 |
| lucasagomes | keystone trusts is the most generic one IMO | 14:00 |
| jroll | so anyone that can dhcp has admin tokens for your cloud | 14:00 |
| lucasagomes | or using the driver's virtual media to pass the token in a secure way | 14:00 |
| dtantsur | jroll, I believe it's sort of a problem, but I don't see how the spec in question make it worse or better | 14:00 |
| jroll | hmm | 14:00 |
| lucasagomes | jroll, yeah that's far from being a good case | 14:00 |
| jroll | dtantsur: right... I didn't realize that was a keystone token. that spec is fine, then | 14:01 |
| lucasagomes | ofc the token is deleted after deploy | 14:02 |
| lucasagomes | FYI https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/pxe.py#L397 | 14:02 |
| jroll | lucasagomes: that doesn't revoke it from keystone, though, does it? | 14:02 |
| dtantsur | jroll, cool. Re bad value for default_driver - I don't know :( Do you have some ideas? | 14:03 |
| jroll | dtantsur: idk, crash the conductor? :) | 14:03 |
| jroll | (actually not the worst idea) | 14:03 |
| lucasagomes | jroll, afair no, it just use the token used for the request of dpeloying a machine and saves it to a file | 14:03 |
| lucasagomes | but that's a admin token because our api is admin only | 14:03 |
| dtantsur | jroll, that seems reasonable to me: if discovery is enabled and no sane value is provided, it's much better, than failing during the discovery | 14:03 |
| jroll | lucasagomes: right... so I boot a bare metal node. I poke your tftp server as if I'm booting a ramdisk, and get a token. then I use that token to do whatever I want. | 14:04 |
| jroll | dtantsur: +1 | 14:04 |
| *** jrist has joined #openstack-ironic | 14:04 | |
| lucasagomes | jroll, yeah, that's what the default ramdisk now | 14:04 |
| lucasagomes | jroll, the token is saved on the tftp with a deterministic name | 14:05 |
| lucasagomes | and the ramdisk gets it after boot | 14:05 |
| lucasagomes | and use it to talk to the ironic api (passing the iscsi iqn etc...) | 14:05 |
| lucasagomes | jroll, oh got ur question now... yeah that's a risk | 14:06 |
| lucasagomes | yeah it's quite bad actually... | 14:06 |
| jroll | right... | 14:06 |
| lucasagomes | jroll, so there's a spec for the deploy ilo driver | 14:06 |
| jroll | because we also don't have network isolation automated | 14:07 |
| jroll | so hopefully someone turned off access to the control plane after deploy :) | 14:07 |
| lucasagomes | yeah :/ the way to work around is have a network for the deploy | 14:07 |
| lucasagomes | and another for production | 14:07 |
| lucasagomes | jroll, the token gets deleted after deploy | 14:07 |
| lucasagomes | tho token for other machines being deployed may be there | 14:07 |
| jroll | ahhh, right | 14:08 |
| lucasagomes | jroll, so on the ilo deploy, afai remember, they want to create a floppy disk image with the token and use the virtual media device | 14:08 |
| lucasagomes | to pass the token to the node | 14:08 |
| jroll | right | 14:09 |
| jroll | I know there's good ways to solve this :) | 14:09 |
| *** overlayer has joined #openstack-ironic | 14:09 | |
| lucasagomes | yeah | 14:09 |
| jroll | but at first I thought for discovery, we would pass more creds than usual to the ramdisk | 14:09 |
| openstackgerrit | Dmitry Tantsur proposed a change to openstack/ironic-specs: Generic Hardware Discovery Bits https://review.openstack.org/102565 | 14:09 |
| lucasagomes | right, yeah not really :/ I mean... we are a admin-only api right now so | 14:10 |
| dtantsur | jroll, lucasagomes, ifarkas updated please have a look ^^^ | 14:10 |
| lucasagomes | jroll, and back to the #100734 talk, so ur ok in keeping it on the pxe group? | 14:11 |
| lucasagomes | I understand its not perfect, but I think that separating may be worse | 14:11 |
| lucasagomes | :/ | 14:11 |
| jroll | lucasagomes: sure | 14:11 |
| lucasagomes | ack thanks | 14:11 |
| jroll | lucasagomes: and then I'll file a bug to rename the 'pxe' driver to 'iscsi' driver | 14:11 |
| jroll | :) | 14:11 |
| lucasagomes | lol | 14:11 |
| * jroll bbiab | 14:12 | |
| *** lazy_prince is now known as killer_prince | 14:17 | |
| *** overlayer has quit IRC | 14:17 | |
| *** jgrimm has joined #openstack-ironic | 14:26 | |
| lucasagomes | if someone is willing to review something here's a good one https://review.openstack.org/#/c/91585/ | 14:27 |
| NobodyCam | good morining Ironic | 14:27 |
| dtantsur | NobodyCam, morning! | 14:27 |
| lucasagomes | we need it to be supported fro the client asap, we are waiting for it to do a new release | 14:27 |
| lucasagomes | NobodyCam, morning | 14:27 |
| NobodyCam | morning dtantsur , lucasagomes | 14:28 |
| rloo | morning NobodyCam. | 14:28 |
| NobodyCam | jroll: rloo too :) | 14:28 |
| rloo | lucasagomes: yes, I will take a look at that pagination. Seems like I had comments about it before ;) | 14:28 |
| lucasagomes | rloo, yup, there's a new patch set up for it | 14:29 |
| lucasagomes | thanks | 14:29 |
| jroll | morning NobodyCam | 14:30 |
| NobodyCam | :) | 14:30 |
| dtantsur | NobodyCam, and of course I'm advertising discovery spec: https://review.openstack.org/#/c/102565 :) Could you have a look? I guess I addressed your comment | 14:32 |
| NobodyCam | :) | 14:32 |
| NobodyCam | will do | 14:32 |
| *** jistr has quit IRC | 14:35 | |
| *** jistr has joined #openstack-ironic | 14:36 | |
| *** martyntaylor has joined #openstack-ironic | 14:40 | |
| dtantsur | back in ~hour | 14:46 |
| *** jcoufal has quit IRC | 14:54 | |
| *** faizan has quit IRC | 14:58 | |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic-specs: Add support for retry on NodeLocked exceptions https://review.openstack.org/103996 | 15:18 |
| Shrews | lucasagomes: That patch set ^^^^ should be a bit more explicit now. Thanks for your suggestions. | 15:21 |
| Shrews | more welcomed | 15:22 |
| lucasagomes | Shrews, ta much! will take a look (finishing something up here quickly before) | 15:22 |
| Shrews | no worries | 15:22 |
| Shrews | i have many other things i'm trying to knock off my list today, too :) | 15:23 |
| Shrews | most involving tempest... *grumble* *grumble* | 15:23 |
| openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Add IloDriver and its IloPower module https://review.openstack.org/89500 | 15:28 |
| *** mdorman has joined #openstack-ironic | 15:30 | |
| *** amitpp has quit IRC | 15:32 | |
| *** mdorman has quit IRC | 15:35 | |
| *** mdorman has joined #openstack-ironic | 15:36 | |
| *** hemna has quit IRC | 15:42 | |
| *** athomas has quit IRC | 15:50 | |
| *** derekh__ is now known as derekh_ | 15:53 | |
| *** mdorman_ has joined #openstack-ironic | 15:54 | |
| *** mdorman has quit IRC | 15:56 | |
| *** mdorman_ is now known as mdorman | 15:56 | |
| dtantsur | Shrews, what's the point of having 0 retries by default? | 15:56 |
| *** athomas has joined #openstack-ironic | 15:56 | |
| *** jistr has quit IRC | 15:56 | |
| dtantsur | value 1 makes much more sense to me | 15:56 |
| dtantsur | (or even 2, can be discussed) | 15:56 |
| Shrews | dtantsur: so we don't affect operators who upgrade and don't want the change in behavior | 15:57 |
| Shrews | because there is a penalty for enabling it | 15:57 |
| dtantsur | Shrews, I guess a user will prefer to opt-in for sudden errors, not vice verse | 15:57 |
| Shrews | dtantsur: yes, it's an opt-in feature | 15:58 |
| dtantsur | Shrews, there's also penalty in _not_ enabling it, i.e. users will get errors from time-to-time. Yes, and I suggest that users want to opt-out this feature, not opt-in | 15:58 |
| Shrews | dtantsur: well, that's what they get today :) | 15:59 |
| dtantsur | and that's what we're trying to fix | 15:59 |
| dtantsur | Shrews, seriously, could you explain why would anyone set this to 0? In what case getting vague error is better than delay? | 15:59 |
| Shrews | right, but the "real" fix is the async api stuff. but that's not until K | 15:59 |
| *** eghobo has joined #openstack-ironic | 16:00 | |
| dtantsur | Shrews, yeah, I know. But this looks like we're fixing things only for devstack (where we'll have proper non-0 default) and leaving it broken for others | 16:01 |
| Shrews | dtantsur: so, what we're trying to do is fix this in the gate b/c it's annoying. since the real fix is the async api, we don't want having this in affect anyone with current deploys. | 16:01 |
| Shrews | dtantsur: nothing is "broken", really. | 16:01 |
| dtantsur | Shrews, it is. People are seeing errors that are completely related to our internal business | 16:02 |
| dtantsur | Shrews, if we're fixing tempest, we should fix it in tempest, not in our code | 16:02 |
| Shrews | it's not fixing a bug, but changing behavior. if someone wants the new behavior, they should explicitly enable it | 16:02 |
| Shrews | dtantsur: the tempest fix is really hard b/c they use their own client, so it would affect a lot of tests | 16:03 |
| dtantsur | Shrews, any bug can be called "old behaviour" :D We have several bug reports for this, I guess, and a special workaround in a quick start guide telling people to disable periodic task | 16:03 |
| *** eghobo has quit IRC | 16:03 | |
| dtantsur | Shrews, it's more than enough to me to call for changing the default behaviour | 16:04 |
| Shrews | dtantsur: feel free to bring this up in today's meeting if you wish | 16:04 |
| dtantsur | (especially provided how many things we're already changing in Juno) | 16:04 |
| Shrews | dtantsur: or just post a -1 on the spec with your comments :) people can weigh in there, too | 16:07 |
| dtantsur | yeah, I'm trying to summarize what I think on it | 16:08 |
| *** matty_dubs is now known as matty_dubs|lunch | 16:09 | |
| *** eghobo has joined #openstack-ironic | 16:10 | |
| *** bandicot has joined #openstack-ironic | 16:10 | |
| dtantsur | ok, left my looooong letter :) thanks for bringing it anyway | 16:11 |
| *** bandicot has quit IRC | 16:11 | |
| Shrews | dtantsur: cool :) | 16:11 |
| *** pcrews has joined #openstack-ironic | 16:11 | |
| NobodyCam | Nisha: dtantsur: are you both in channel? | 16:14 |
| dtantsur | I am, though wanted some tea | 16:14 |
| NobodyCam | heheh you can get tea | 16:14 |
| dtantsur | ack, you can raise whatever you want, I'll answer a bit later | 16:16 |
| NobodyCam | has every one looked over the agenda? lucasagomes still want "Config option deprecation strategy" on the agenda? | 16:17 |
| *** martyntaylor1 has joined #openstack-ironic | 16:17 | |
| lucasagomes | NobodyCam, oh, no it can be removed | 16:17 |
| lucasagomes | last meeting someone pointed to a doc about how to do it on openstack | 16:17 |
| lucasagomes | so it's grand | 16:17 |
| NobodyCam | dtantsur: I was going to see if I could get you and Nisha toghether to chat about the inventory interface | 16:17 |
| *** martyntaylor has quit IRC | 16:18 | |
| NobodyCam | lucasagomes: removed :) | 16:18 |
| lucasagomes | cheers | 16:18 |
| *** derekh_ has quit IRC | 16:20 | |
| *** eghobo has quit IRC | 16:21 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Rename/update common/tftp.py to common/pxe_utils.py https://review.openstack.org/103595 | 16:23 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add iPXE support for Ironic https://review.openstack.org/99318 | 16:23 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add iPXE Installation Guide documentation https://review.openstack.org/106809 | 16:23 |
| *** eghobo has joined #openstack-ironic | 16:24 | |
| *** eghobo has quit IRC | 16:25 | |
| *** eghobo has joined #openstack-ironic | 16:25 | |
| dtantsur | NobodyCam, yeah... we have some preliminary talk, but I'm not sure about the outcome. At least Nisha saw the spec. | 16:31 |
| *** romcheg1 has joined #openstack-ironic | 16:34 | |
| *** martyntaylor1 has quit IRC | 16:37 | |
| *** ramineni has quit IRC | 16:37 | |
| NobodyCam | :) | 16:38 |
| devananda | morning, all | 16:42 |
| dtantsur | devananda, morning! | 16:42 |
| romcheg1 | Morning devananda! | 16:42 |
| devananda | dtantsur: welcome back :) | 16:42 |
| dtantsur | thanks :) | 16:43 |
| *** romcheg1 is now known as romcheg_ltp | 16:43 | |
| Nisha | dtantsur: :) | 16:43 |
| NobodyCam | good morning devananda | 16:43 |
| dtantsur | Nisha, hi! :) | 16:44 |
| Nisha | dtantsur: Hi :) Welcome back | 16:44 |
| Nisha | dtantsur: we were discussing about inventory specs (yours and mine ) when you were on vacation | 16:45 |
| dtantsur | aaaah I see. Could you summarize for me? | 16:46 |
| Nisha | dtantsur: last conclusion which i had from NobodyCam is that yours and mine are duplicate specs :) | 16:46 |
| Nisha | almost duplicate | 16:46 |
| dtantsur | really? I don't think so, unless things have changed a lot | 16:46 |
| dtantsur | we may have some duplication in what we define, but mine does not cover user-driven discovery (deliberately, I left it for you) | 16:47 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding support for decommissioning https://review.openstack.org/104379 | 16:47 |
| *** martyntaylor has joined #openstack-ironic | 16:47 | |
| Nisha | dtantsur: nothing has changed except that i populate the same data field inventory as you do | 16:48 |
| Nisha | :) | 16:48 |
| dtantsur | let me quickly look through your spec now, ok? | 16:48 |
| Nisha | dtantsur: yes | 16:48 |
| Nisha | That should be great | 16:48 |
| Nisha | dtantsur: https://review.openstack.org/#/c/100951/ | 16:49 |
| dtantsur | yeah I know | 16:49 |
| Nisha | dtantsur: Just for reference (to speed up) :) | 16:49 |
| NobodyCam | dtantsur: Nisha: both of your spec are adding a inventory field what I what to make sure of is that they dont both add it in different ways. | 16:49 |
| dtantsur | Nisha, one thing to warn you: devananda is against storing the whole hardware inventory in Node, that's why I dropped this point from my spec | 16:49 |
| dtantsur | NobodyCam, mine no longer add inventory field. that was banned by devananda. | 16:50 |
| devananda | IMNSHO, Ironic is not a CMDB | 16:50 |
| * devananda goes back to catching up on emails | 16:50 | |
| NobodyCam | Nisha: ^^^ did you see that | 16:50 |
| Nisha | dtantsur: I added it as per your comment | 16:50 |
| dtantsur | Nisha, yeah and I proved to be wrong. Sorry for that. | 16:51 |
| dtantsur | Nisha, now what I see for your spec is to cover the whole procedure starting with new CLI command and ending with calling to store_inventory() defined in my spec | 16:51 |
| dtantsur | what do you think? | 16:52 |
| *** Mikhail_D_ltp has joined #openstack-ironic | 16:53 | |
| *** Nisha has quit IRC | 16:55 | |
| lucasagomes | devananda, morning | 16:55 |
| * NobodyCam needs to run afk for a few minutes... bbs | 16:59 | |
| devananda | lucasagomes: mornin! | 17:00 |
| *** matty_dubs|lunch is now known as matty_dubs | 17:01 | |
| *** harlowja_away is now known as harlowja | 17:03 | |
| *** Alexei_987 has quit IRC | 17:03 | |
| jroll | devananda: proposal to add "Ironic is not a CMDB" to the topic :P | 17:06 |
| *** Nisha has joined #openstack-ironic | 17:07 | |
| *** krtaylor has quit IRC | 17:07 | |
| dtantsur | jroll, +1 | 17:08 |
| *** ndipanov is now known as ndipanov_gone | 17:08 | |
| dtantsur | or we can reiterator every meeting | 17:08 |
| dtantsur | * reiterate | 17:08 |
| devananda | jroll: lol | 17:09 |
| Nisha | dtantsur: Why we want to define a new CLI? | 17:10 |
| *** kylestev_ has joined #openstack-ironic | 17:10 | |
| devananda | just gonna toss this out there | 17:10 |
| *** kylestev_ has left #openstack-ironic | 17:10 | |
| dtantsur | Nisha, I mean arguments, you introduce. Sorry. it's evening here and it's my first day after PTO :) | 17:11 |
| devananda | do you think the mission statement over-reaches into CMDB territory? http://git.openstack.org/cgit/openstack/governance/tree/reference/programs.yaml#n142 | 17:11 |
| kylestev | wat… how did my second nick join | 17:11 |
| devananda | if it's perceived that way, I'd like to know, since that wasn't my intent | 17:11 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: fix https://review.openstack.org/106816 | 17:11 |
| dtantsur | devananda, it may be treated like this. The problem is in word "manage" which is too vague | 17:11 |
| jroll | ^ | 17:11 |
| devananda | i see | 17:11 |
| rloo | devananda: who reads the mission statement? | 17:11 |
| jroll | JoshNang: nice commit message :) | 17:12 |
| devananda | rloo: people? I dunno | 17:12 |
| rloo | devananda: this is the first time I'm looking at it ;) | 17:12 |
| kylestev | “fix” | 17:12 |
| kylestev | lol | 17:12 |
| JoshNang | jroll: lol. whoops. | 17:12 |
| devananda | would s/manage/prepare/ clear it up? | 17:12 |
| Nisha | dtantsur: store_inventory() even discovers the properties? | 17:12 |
| jroll | just s/managing and // (imo) | 17:12 |
| dtantsur | devananda, s/of managing and provisioning physical machines,/of provisioning physical machines,/ | 17:12 |
| dtantsur | jroll, heh :) | 17:13 |
| rloo | +1 for jroll's + dtantsur's suggestion. | 17:13 |
| devananda | ack | 17:13 |
| dtantsur | Nisha, it's the final point, which stores everything in the Node (merely database update) | 17:13 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Adding support for decommissioning https://review.openstack.org/104379 | 17:13 |
| dtantsur | Nisha, try to make it clearer: keep everything, except for data model impact (you can just use store_inventory() defined in my spec) | 17:14 |
| Nisha | dtantsur: you said that the discovered properties should not be stored in Node object | 17:15 |
| dtantsur | Nisha, that was the outcome of the discussion before my PTO: update only Node.properties, do not store the full inventory | 17:15 |
| dtantsur | i.e. store only what is stored now, nothing new | 17:15 |
| jroll | should not be stored in *Node.inventory* | 17:15 |
| * devananda adds that to the agenda | 17:16 | |
| jroll | they are still stored in Node.properties | 17:16 |
| dtantsur | Nisha, see my spec for details on what we agreed should be done | 17:16 |
| Nisha | dtantsur: Ok. i got that and that was my first proposal too | 17:16 |
| Nisha | I just had some common properties more which may be required by other drivers also in future | 17:17 |
| dtantsur | Nisha, you can reuse the endpoint (with it's REST API) and store_inventory(). As you see in my spec, the full inventory is passed to a driver and it can do whatever it wants with it | 17:17 |
| Nisha | like Boot type : UEFI or BIOS etc | 17:17 |
| * NobodyCam is back | 17:18 | |
| Nisha | dtantsur: which line shall i see | 17:20 |
| dtantsur | Nisha, #53 explains storing parameters, #111 - REST API for storing collected parameters, #213 new method in management interface for pluging in | 17:21 |
| *** hemna_ is now known as hemna | 17:33 | |
| *** krtaylor has joined #openstack-ironic | 17:40 | |
| NobodyCam | nice chatch dtantsur with 102247 | 17:43 |
| NobodyCam | catch even | 17:43 |
| jroll | hmmm | 17:43 |
| jroll | I almost think that should check if uuid changed | 17:44 |
| jroll | not just if it was in values | 17:44 |
| NobodyCam | jroll: oh sure now that it's approved | 17:46 |
| dtantsur | how can it be changed if it's not in values? oo | 17:46 |
| jroll | :| | 17:46 |
| jroll | dtantsur: I mean | 17:47 |
| jroll | dtantsur: if I do: node.update({'uuid': node.uuid, 'driver': 'agent_ipmitool'}), that should work | 17:47 |
| jroll | because it doesn't actually change the uuid | 17:47 |
| *** pelix has quit IRC | 17:47 | |
| jroll | because I think that now, if a PATCH request is issued with uuid in it... it will fail | 17:48 |
| jroll | which isn't *too* bad | 17:48 |
| jroll | but gives kind of a crappy user experience | 17:48 |
| dtantsur | jroll, PATCH with uuid will fail for different reason | 17:48 |
| jroll | will it? | 17:48 |
| dtantsur | changing /uuid is banned on API level | 17:48 |
| * jroll looks at code | 17:48 | |
| jroll | changing, yes, but including the current uuid in the PATCH body? | 17:49 |
| dtantsur | let me see... | 17:49 |
| jroll | I don't think it will fail today | 17:50 |
| dtantsur | jroll, https://github.com/openstack/ironic/blob/master/ironic/api/controllers/v1/types.py#L134 | 17:50 |
| dtantsur | judging by this line, it will just fail, even if value is duplicated, no? | 17:50 |
| jroll | oh huh | 17:51 |
| jroll | yeah, I think so | 17:51 |
| jroll | I'm not super passionate about this, feel free to tell me to stop talking about it :) | 17:51 |
| dtantsur | I'm always glad to double-check my understanding | 17:52 |
| jroll | right | 17:52 |
| * jroll tests | 17:52 | |
| *** Penick has joined #openstack-ironic | 17:53 | |
| jroll | dtantsur: you're right: {"error_message": "{\"debuginfo\": null, \"faultcode\": \"Client\", \"faultstring\": \"'/uuid' is an internal attribute and can not be updated\"}"} | 17:56 |
| jroll | ignore me :) | 17:57 |
| dtantsur | ;) | 17:57 |
| * NobodyCam add jroll to the ignore list | 18:00 | |
| dtantsur | LOL | 18:01 |
| jroll | :P | 18:01 |
| * jroll talks trash about NobodyCam | 18:01 | |
| NobodyCam | lol | 18:01 |
| * Shrews learns new tech word: memristors | 18:03 | |
| Shrews | i only learned the word. i don't know what it actually *is* yet | 18:03 |
| * Shrews too easily distracted by internal corporate emails | 18:04 | |
| dtantsur | should be quite an interesting thing. in my institute a lot of folks were excited by it | 18:04 |
| devananda | Shrews: those look fun. | 18:04 |
| matty_dubs | I think the important thing with new words is to begin using them immediately, before one is clear on the meaning. | 18:04 |
| matty_dubs | That party was totally memristors! | 18:05 |
| Shrews | matty_dubs: you sure it's not a band? | 18:05 |
| Shrews | matty_dubs and the memristors. one night only! | 18:05 |
| JayF | I used to be forgetful, then I started taking my memristors. One in the morning and one before bed, and now I never forget anything! | 18:05 |
| jroll | lol | 18:06 |
| dtantsur | LOL | 18:06 |
| matty_dubs | Haha | 18:06 |
| matty_dubs | The technology is pretty neat, though! | 18:07 |
| NobodyCam | just fyi for all: I will need to leave about 1/2 through the meeting :( | 18:16 |
| dtantsur | that's pity :( I wanted to discuss discovery specs in the end (we actually made a good progress towards sorting overlapping) | 18:19 |
| NobodyCam | dtantsur: please do that. You don't need me for that conversation. | 18:20 |
| *** shakamunyi has joined #openstack-ironic | 18:21 | |
| Shrews | devananda: adam_g: fyi, greatly modified ironic rebuild tempest test -- https://review.openstack.org/94439 | 18:21 |
| devananda | ooooh | 18:22 |
| * devananda looks | 18:22 | |
| Shrews | devananda: adam_g: i plan to add the 'rebuild' compute feature flag as a separate patch. including here and introducing its use properly would add much complexity | 18:22 |
| Shrews | and i'm trying to get up with mtreinish in -qa to talk about this patch | 18:23 |
| adam_g | Shrews, cool | 18:23 |
| adam_g | Shrews, my tempest work has been blocked by some other stuff this past week. hope to focus more on it this week | 18:23 |
| Shrews | devananda: as we discussed, i just usurped the one basic test | 18:24 |
| devananda | adam_g: I saw the patch to lower instance RAM to 512 -- good stuff :) | 18:27 |
| Shrews | ooh. yay | 18:27 |
| adam_g | devananda, yeah. spinning that up now on rax to see how it fares with multiple spawns | 18:27 |
| devananda | how close do you think we are to being able to experiment with parallel tempest? | 18:27 |
| adam_g | devananda, we still need to figure out the isolated credentials / neutron networking stuff | 18:28 |
| devananda | ack | 18:28 |
| devananda | Shrews: your patch is adding a spurious blank line in schenario/manager.py | 18:29 |
| devananda | Shrews: silly thing to nit on, but it's there :p | 18:30 |
| Shrews | devananda: intentional. i thought a blank line there looked better | 18:30 |
| devananda | ah | 18:30 |
| Shrews | but i can remove it if you wish | 18:30 |
| devananda | nope, it's fine | 18:31 |
| devananda | i try to keep formatting/cleanup changes limited to the area of a file i'm changing, that's all | 18:31 |
| Shrews | ack | 18:31 |
| Shrews | devananda: to be fair, that are *was* changed in the previous patch sets :) | 18:32 |
| devananda | Shrews: i figured as much | 18:32 |
| Shrews | s/are/area/ | 18:32 |
| comstud | devananda: InstanceDeployFailure == returning 500 seems kinda crappy | 18:32 |
| comstud | devananda: even crappier that tear down raises that same exception.. kinda weirdly named for that | 18:33 |
| comstud | My fix for build + delete races.. where nova lways calls unprovision.. is not quite right, I guess. | 18:33 |
| comstud | if you call tear down for a node in NOSTATE, InstanceDeployFailure is raised, and a 500 is returned | 18:34 |
| comstud | and the client_wrapper in nova retries | 18:34 |
| devananda | ugh | 18:34 |
| comstud | .. because it's a 500 | 18:34 |
| comstud | or something. | 18:34 |
| devananda | so 500 shouldn't be retried | 18:34 |
| comstud | somehow... | 18:34 |
| devananda | InstanceDeployFailure *is* an internal server error | 18:34 |
| comstud | the client is detecting it as a 503, actually | 18:34 |
| comstud | because it prins 'Service Unavailable' | 18:34 |
| devananda | ahh | 18:34 |
| comstud | prints | 18:34 |
| comstud | But | 18:34 |
| comstud | from what I can tell in ironic... it should be a 500 | 18:34 |
| comstud | so I dunno what's going on there. | 18:35 |
| devananda | so NodeLocked is 409 CONFLICT | 18:35 |
| comstud | yeah, in this case, the node is not locked. | 18:35 |
| devananda | and requesting to unprovision something with no provision state should also be 409 | 18:35 |
| devananda | as that's a client side errror | 18:35 |
| comstud | this is a slightly different case. | 18:35 |
| devananda | requesting a state transition that's impossible | 18:35 |
| devananda | oh? | 18:35 |
| comstud | Say nova is somehow just... out of sync. | 18:35 |
| comstud | I dunno, this is a really bad case no matter what... heh but | 18:36 |
| comstud | Say you have a node that is not provisioned | 18:36 |
| comstud | and you try to unprovision it | 18:36 |
| comstud | You get InstanceDeployFailure | 18:36 |
| Shrews | quick walk before the meeting. brb | 18:37 |
| comstud | it'd be kinda nice if it raised something more like 'YeahItsAlreadyToreDownDufus' | 18:37 |
| comstud | I dunno, I have multiple problems with this stuff. | 18:38 |
| comstud | 1) Tear down raising an exception labeled "InstanceDeployFailure" | 18:38 |
| comstud | when it's not a deploy failure. | 18:38 |
| comstud | 2) InstanceDeployFailure returning a 5xx code. It should probably be a 400 or something. | 18:38 |
| comstud | 3) Trying to tear down something already torn down... Some way to know that it's already done would be nice.. so that callers can just 'pass' on that. | 18:40 |
| *** Nisha has quit IRC | 18:40 | |
| comstud | .. | 18:40 |
| devananda | right, so differentiating between "this deploy failed because of user error" and "this deploy failed because the hardware malfunctioned" | 18:41 |
| devananda | *this deploy request ... | 18:41 |
| comstud | handling tear down has nothing to do with the initial deploy | 18:41 |
| comstud | i'm trying to un-deploy. | 18:41 |
| comstud | :) | 18:41 |
| devananda | *the request to the deploy driver failed because ... | 18:42 |
| devananda | :) | 18:42 |
| comstud | haha oaky. | 18:42 |
| comstud | that's fair. | 18:42 |
| comstud | although in this case.. | 18:42 |
| comstud | InstanceDeployFailure is raised right from conductor manager | 18:42 |
| devananda | these errors are either generated within the deply driver or by the ConductorManager wrapping it | 18:42 |
| comstud | checking states. | 18:42 |
| devananda | right | 18:42 |
| comstud | even before calling driver | 18:42 |
| devananda | so that's silly | 18:42 |
| devananda | probably my fault | 18:42 |
| comstud | (let me verify i'm not on crack) | 18:43 |
| comstud | but this is what i recall from last week | 18:43 |
| * jroll verifies comstud is likely on crack | 18:43 | |
| comstud | https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L447 | 18:43 |
| devananda | yah. that's an easy fix | 18:44 |
| comstud | i'm trying to fix up https://review.openstack.org/#/c/104649/ | 18:44 |
| comstud | but I'm wondering how far I go there. | 18:44 |
| devananda | comstud: https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L387 | 18:45 |
| devananda | https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L370 | 18:45 |
| devananda | https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L208 | 18:46 |
| comstud | 370 seems reasonable | 18:46 |
| comstud | err | 18:46 |
| comstud | 387 seems reasonable | 18:46 |
| devananda | no .it's the same thing | 18:46 |
| *** Nisha has joined #openstack-ironic | 18:46 | |
| comstud | i know, i can't read numbers. | 18:46 |
| comstud | hehe | 18:46 |
| devananda | also, 387 is still a user error | 18:47 |
| devananda | it's 409 CONFLICT | 18:47 |
| devananda | resoruce in wrong state, can't perform requested action | 18:47 |
| devananda | *it should be | 18:47 |
| comstud | well, i suppose... I guess i was looking that the deploy driver raised | 18:47 |
| comstud | i guess an InvalidParameterValue would be a user error | 18:47 |
| devananda | which is allowed to propagate back to the user in other areas | 18:47 |
| devananda | i dunno why it's converted to a server-side error on 387 :( | 18:48 |
| devananda | ditto on 458 | 18:48 |
| comstud | but also: i don't think 500 is any sort of reason http code though for any exceptions we are purposely generating and returning. | 18:48 |
| comstud | "Internal Server Error" | 18:48 |
| comstud | That generally means you like... your code is really fucked up. | 18:49 |
| comstud | or you like ran out of memory or some stupidness. | 18:49 |
| comstud | :) | 18:49 |
| devananda | comstud: except none of the other 5XX codes make sense either | 18:49 |
| comstud | 400 would make sense for these | 18:49 |
| devananda | no | 18:49 |
| devananda | 400 means the syntax of the request itself was bad (malformed) | 18:50 |
| devananda | dont repeat it ever | 18:50 |
| * devananda steps afk for a few min before meeting | 18:50 | |
| comstud | hrm, yeah. | 18:50 |
| comstud | ok. | 18:50 |
| *** rwsu has joined #openstack-ironic | 18:51 | |
| *** Nisha has quit IRC | 18:55 | |
| NobodyCam | one minutes | 18:59 |
| *** mrda-away is now known as mrda | 18:59 | |
| mrda | Morning Ironic | 18:59 |
| NobodyCam | morning mrda | 18:59 |
| jroll | heya mrda | 18:59 |
| mrda | \o | 18:59 |
| dtantsur | mrda, morning :) | 18:59 |
| * NobodyCam heads over to meeting room | 19:00 | |
| devananda | morning, mrda ! | 19:00 |
| gilliard | The creation of flavours would need to be based on the different node types specified in that file, and flavours would be matched to nodes using nova scheduler's ComputeCapbilitiesFilter. | 19:01 |
| gilliard | umm, wrong room, sorry. | 19:01 |
| *** mitz_ has quit IRC | 19:06 | |
| jroll | mrda: fwiw, we have a patch coming soon to fix virt driver performance in a different way | 19:14 |
| mrda | hopefully not requiring too great a rebase if you get there first :P | 19:16 |
| jroll | nah | 19:16 |
| jroll | we put in a hack to hit /nodes/details, so you don't have to hit the api for each node in a loop | 19:16 |
| jroll | going to make that better and submit it | 19:17 |
| openstackgerrit | David Shrewsbury proposed a change to openstack/ironic-specs: Add support for retry on NodeLocked exceptions https://review.openstack.org/103996 | 19:21 |
| *** dwalleck has joined #openstack-ironic | 19:32 | |
| dtantsur | jroll, remind please, what we decided re coordination of your spec with discovery spec? | 19:40 |
| jroll | um | 19:40 |
| jroll | did we decide anything? | 19:40 |
| jroll | and do you mean, as far as the lookup stuff? | 19:40 |
| dtantsur | I don't remember :) | 19:41 |
| jroll | ha, ko | 19:41 |
| jroll | well | 19:41 |
| dtantsur | that's why I'm asking | 19:41 |
| jroll | (s/ko/ok/) | 19:41 |
| jroll | goal is to land this during the mid cycle | 19:41 |
| jroll | so we need to kick the spec through | 19:41 |
| jroll | idk if there should be any dependencies between the two... | 19:41 |
| dtantsur | jroll, does it work for you, if we land your spec first, than do any fixes in code, if my spec will require it? | 19:42 |
| jroll | +1 | 19:42 |
| dtantsur | good! | 19:42 |
| jroll | I think that's all it will be, is instead of vendor_passthru/lookup, we will hit your new endpoint | 19:42 |
| jroll | probably | 19:42 |
| jroll | which makes me appreciate your extra key in the response | 19:43 |
| jroll | although now that I think about it | 19:43 |
| jroll | is that what the driver returns? | 19:43 |
| jroll | dtantsur: ^ | 19:43 |
| dtantsur | jroll, I made it specially for you :) driver both accept extra and return it | 19:43 |
| jroll | right, so | 19:44 |
| jroll | that makes it async, I think | 19:44 |
| jroll | and your spec explicitly says that it should not be | 19:44 |
| jroll | IMBW | 19:44 |
| dtantsur | I don't really care, I'm just avoiding putting there special provisioning for async case | 19:44 |
| jroll | I guess you can call to drivers synchronously... we do it in driver_vendor_passthru | 19:45 |
| jroll | which is... odd | 19:45 |
| jroll | but ok | 19:45 |
| jroll | sounds good :) | 19:45 |
| dtantsur | anyway everything will be async somewhere in K :) | 19:45 |
| jroll | right | 19:45 |
| *** tatyana has joined #openstack-ironic | 19:49 | |
| *** Penick has quit IRC | 19:59 | |
| mrda | :) | 20:03 |
| dtantsur | folks, please have a look at discovery spec in question: https://review.openstack.org/#/c/102565/ :) Do not forget to read intro about what this spec covers and what deliberately not. Thanks :) | 20:06 |
| * dtantsur can be annoying ^_^ | 20:06 | |
| dtantsur | and now I'm slowly moving to bed, good night everyone | 20:06 |
| lucasagomes | dtantsur, night | 20:07 |
| lucasagomes | I'm leaving as well, g'night everyone! | 20:08 |
| *** lucasagomes is now known as lucas-dinner | 20:08 | |
| Shrews | night lucas-dinner, dtantsur | 20:08 |
| *** dtantsur is now known as dtantsur|afk | 20:08 | |
| *** athomas has quit IRC | 20:18 | |
| *** Penick has joined #openstack-ironic | 20:19 | |
| *** tatyana has quit IRC | 20:26 | |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Make ComputeCapabilitiesFilter work with Ironic https://review.openstack.org/105802 | 20:30 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add versioning to Agent decommission https://review.openstack.org/106859 | 20:32 |
| *** Hefeweizen has quit IRC | 20:35 | |
| devananda | jroll: there are two of you in gerrit | 20:35 |
| jroll | uh oh | 20:36 |
| rloo | so that's how jroll gets twice the work done. | 20:36 |
| jroll | devananda: username I always use is jimrollenhagen | 20:36 |
| devananda | heh | 20:36 |
| devananda | jroll: they both appear with the same name and email address, which is all the web UI lets me see | 20:36 |
| jroll | fun | 20:37 |
| devananda | jroll: right. so selecting either one gives me an error now, when trying to add you | 20:38 |
| devananda | jroll: which is great, since it clearly shows you in the ipa-core group already | 20:38 |
| jroll | sweet | 20:38 |
| devananda | jroll: can you poke folks in infra to try to clean it up? I need to get lunch... if not, i'll try when i get back | 20:39 |
| jroll | devananda: sure, where are you seeing this? | 20:40 |
| devananda | jroll: attempting to add you on https://review.openstack.org/#/admin/groups/165,members | 20:40 |
| JayF | jroll: maybe you should get your clone destroyed by -infra? | 20:40 |
| jroll | probably | 20:41 |
| jroll | I tried to fix my launchpad ID once | 20:41 |
| * devananda lunches, bbiab | 20:41 | |
| jroll | which is probably where this comes from | 20:41 |
| *** Mikhail_D_ltp has quit IRC | 20:48 | |
| *** jgrimm has quit IRC | 20:52 | |
| *** martyntaylor has quit IRC | 20:55 | |
| *** dwalleck has quit IRC | 20:56 | |
| * NobodyCam is back but need to tend to some things.. so will around but think of /me as afk... | 20:57 | |
| jroll | NobodyCam: that's when you pretend you're afk rather than telling people you're not really afk | 20:57 |
| jroll | :P | 20:57 |
| *** matty_dubs is now known as matty_dubs|gone | 21:00 | |
| *** dwalleck_ has joined #openstack-ironic | 21:01 | |
| *** harlowja is now known as harlowja_away | 21:02 | |
| *** dwalleck has joined #openstack-ironic | 21:02 | |
| *** harlowja_away is now known as harlowja | 21:03 | |
| NobodyCam | jroll: LOL :-p I am here just heavly distracted atm. | 21:05 |
| jroll | :P | 21:05 |
| *** dwalleck_ has quit IRC | 21:05 | |
| *** jdob has quit IRC | 21:06 | |
| *** jbjohnso has quit IRC | 21:07 | |
| devananda | jroll: any luck fixing that? | 21:09 |
| jroll | devananda: no response | 21:09 |
| jroll | I wonder if I can kill one account | 21:09 |
| jroll | or something | 21:09 |
| devananda | jroll: it's probably coming from LP, yea | 21:09 |
| jroll | wheeeee, don't have password for the other account... forgot password doesn't land in my inbox immediately | 21:11 |
| jroll | yay. | 21:11 |
| openstackgerrit | Josh Gachnang proposed a change to openstack/ironic-python-agent: Add versioning to Agent decommission https://review.openstack.org/106859 | 21:22 |
| NobodyCam | devananda: style type question. | 21:26 |
| NobodyCam | line 215 of https://review.openstack.org/#/c/105802/12/ironic/nova/virt/ironic/driver.py is a log.warn for malformed key:value pair. what are thoughts on adding the word "skipping"? or is that implyed in a warning message? | 21:28 |
| jroll | "Explicit is better than implicit." | 21:31 |
| *** openstackgerrit has quit IRC | 21:31 | |
| NobodyCam | also we may want a %r vs the current %s | 21:32 |
| *** openstackgerrit has joined #openstack-ironic | 21:32 | |
| jroll | NobodyCam: why? it's just a string... | 21:33 |
| devananda | NobodyCam: why? it's a string | 21:33 |
| devananda | heh | 21:33 |
| jroll | lol | 21:33 |
| NobodyCam | as it already failed one check it may be something totaly unexpected | 21:34 |
| devananda | no | 21:34 |
| NobodyCam | like an int | 21:34 |
| devananda | it is the result of splitting a string | 21:34 |
| jroll | I guess deva thinks more highly of strings than I do | 21:34 |
| jroll | NobodyCam: capabilities = str(node.properties.get('capabilities')) | 21:34 |
| jroll | it will be a string :) | 21:34 |
| NobodyCam | doh | 21:34 |
| *** killer_prince has quit IRC | 21:34 | |
| devananda | for capability in capabilities.split(","): | 21:34 |
| NobodyCam | yes it will | 21:34 |
| jroll | that too | 21:34 |
| devananda | so it is the result of a string manipulation op (split(",")) on a object that was cast to a string | 21:34 |
| devananda | what's the worst case here | 21:35 |
| devananda | capabilities = str(node.properties.get('capabilities')) could fail | 21:35 |
| jroll | empty string might show up as "Malformed capability ." blabla | 21:35 |
| devananda | but it would still cast as an empty string | 21:35 |
| devananda | jroll: right | 21:35 |
| jroll | also, str(None) casts to 'None' | 21:36 |
| jroll | if .get() fails | 21:36 |
| jroll | I'd kind of prefer quotes around the %s | 21:36 |
| devananda | which will yield Malformed capability 'None'. Sould be 'key:val' | 21:36 |
| devananda | yep | 21:36 |
| devananda | that's missing | 21:36 |
| devananda | as is a closing . | 21:36 |
| jroll | yep | 21:37 |
| NobodyCam | grrrr my phone keeps downloading the HP org chart..lol and I have no idea as to why :-p | 21:37 |
| *** shakamunyi has quit IRC | 21:40 | |
| *** killer_prince has joined #openstack-ironic | 21:41 | |
| *** killer_prince is now known as lazy_prince | 21:41 | |
| devananda | oh, heh | 21:41 |
| devananda | this logs a lot of warnings right now | 21:41 |
| devananda | http://logs.openstack.org/02/105802/12/check/check-tempest-dsvm-virtual-ironic/8bd5491/logs/screen-n-cpu.txt.gz?level=WARNING | 21:41 |
| devananda | because nothing is passing out a capability by default | 21:41 |
| devananda | Shrews: also, sme to you -- welcome to ironic core | 21:43 |
| devananda | it's all official and stuff now | 21:43 |
| devananda | for you and jroll | 21:43 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Make ComputeCapabilitiesFilter work with Ironic https://review.openstack.org/105802 | 21:49 |
| openstackgerrit | Devananda van der Veen proposed a change to openstack/ironic: Make ComputeCapabilitiesFilter work with Ironic https://review.openstack.org/105802 | 21:49 |
| NobodyCam | congtatz jroll Shrews | 21:50 |
| jroll | danke | 21:57 |
| *** Penick has quit IRC | 21:57 | |
| Shrews | devananda: NobodyCam: thx! | 22:00 |
| NobodyCam | :) | 22:00 |
| *** dwalleck has quit IRC | 22:01 | |
| Shrews | now jroll and I have something to do to fill all the free time we have | 22:01 |
| Shrews | :) | 22:01 |
| *** Penick has joined #openstack-ironic | 22:01 | |
| jroll | lol wat is free time | 22:01 |
| Shrews | just something i heard thrown around somewhere. not sure | 22:02 |
| jroll | :) | 22:03 |
| *** dwalleck has joined #openstack-ironic | 22:07 | |
| NobodyCam | devananda: Nice, 105802 will get a +2 from me as soon as Mr J is done with it. :) | 22:11 |
| *** romcheg_ltp has quit IRC | 22:17 | |
| NobodyCam | brb | 22:18 |
| *** lazy_prince has quit IRC | 22:19 | |
| *** romcheg1 has joined #openstack-ironic | 22:19 | |
| *** Shrews has quit IRC | 22:20 | |
| *** Shrews has joined #openstack-ironic | 22:21 | |
| *** killer_prince has joined #openstack-ironic | 22:23 | |
| *** killer_prince is now known as lazy_prince | 22:24 | |
| *** radsy has joined #openstack-ironic | 22:30 | |
| *** radsy has joined #openstack-ironic | 22:30 | |
| *** romcheg1 has quit IRC | 22:32 | |
| NobodyCam | devananda: +2'd (only the first one) is it ok to land, I ask based on its location, or should I advise others to just +/- it with out +a'ing it? | 22:33 |
| devananda | NobodyCam: my CCF patch? yea, OK to land | 22:37 |
| devananda | i have a big chunk of merge squashing to do anyway :( | 22:37 |
| NobodyCam | :( | 22:37 |
| NobodyCam | jroll: Shrews: need a first review as core member? -> https://review.openstack.org/#/c/105802 | 22:38 |
| NobodyCam | should test out that new +a power | 22:39 |
| NobodyCam | lol | 22:39 |
| devananda | jroll: pls see my comment on https://review.openstack.org/#/c/105590/ | 22:42 |
| jroll | devananda: it's more of a weird thing on ironic side, I've seen this before | 22:43 |
| jroll | ttp://127.0.0.1:6385/v1/v1/nodes/?associated=True | 22:43 |
| jroll | er, http://127.0.0.1:6385/v1/v1/nodes/?associated=True | 22:44 |
| devananda | right, that /v1/v1/ is bad | 22:44 |
| jroll | yeah | 22:44 |
| jroll | is something setting IRONIC_URL=http://127.0.0.1:6385/v1 ? | 22:44 |
| jroll | because that should be http://127.0.0.1:6385 | 22:44 |
| jroll | ah ha | 22:45 |
| jroll | DEBUG nova.openstack.common.service [-] ironic.api_endpoint = http://127.0.0.1:6385/v1 | 22:45 |
| devananda | ah | 22:45 |
| jroll | not sure where that comes from... but seems like a devstack (or something)bug | 22:45 |
| devananda | lib/nova_plugins/hypervisor-ironic: iniset $NOVA_CONF ironic api_endpoint http://$SERVICE_HOST:6385/v1 | 22:51 |
| NobodyCam | yep dib seems correctly set: https://github.com/openstack/tripleo-incubator/blob/master/scripts/setup-endpoints#L160 | 22:53 |
| NobodyCam | s/dib/devtest/ | 22:54 |
| devananda | adam_g: i'm cancelling one of your meetup tickets (you have two) | 22:54 |
| adam_g | devananda, ah sorry, too many email addresses :) | 22:54 |
| devananda | adam_g: oh wait. you have three, lol | 22:54 |
| adam_g | jeeez | 22:54 |
| adam_g | sorry :) | 22:54 |
| devananda | adam_g: which address do you want to keep? | 22:55 |
| adam_g | devananda, @gmail.com is fine | 22:55 |
| * NobodyCam thinks adam_g has sign up bot running | 22:55 | |
| devananda | k | 22:55 |
| devananda | comstud: ditto - you have two. | 22:56 |
| devananda | russell_h: ditto | 22:57 |
| adam_g | i call dibs on the extra cookies | 22:57 |
| NobodyCam | ahh /me was too slow | 22:58 |
| NobodyCam | brb | 22:58 |
| * NobodyCam thinks it would be cool for the nova to be able to boot nodes from a specific chassis, "nova boot --flavor blah --image blah_blah --from-chassis <ironic.chassis.uuid>" but also thinks devananda would murder him for speaking such things ATM. | 23:09 | |
| *** dwalleck has quit IRC | 23:14 | |
| JayF | nova does have scheduler hints | 23:15 |
| JayF | which we used liberally early in our deployment to test things | 23:16 |
| JayF | I think comstud even wrote one that could target a specific ironic chassis uuid? | 23:16 |
| jroll | node uuid | 23:17 |
| jroll | but yeah | 23:18 |
| jroll | same idea, if nova knew about chassis :) | 23:18 |
| NobodyCam | ahh, I was thinking chassis for grouping nodes | 23:18 |
| jroll | yeah | 23:18 |
| jroll | I like that idea too :) | 23:18 |
| NobodyCam | shhh | 23:18 |
| NobodyCam | :-p | 23:18 |
| NobodyCam | deva will hear us | 23:19 |
| jroll | heh | 23:19 |
| NobodyCam | rloo: was looking over 91585, I agree with bug 1339731, just have to watch out for people switching sort order and invalidating the "next" marker | 23:23 |
| NobodyCam | s/order/key/ | 23:23 |
| *** lucas-dinner has quit IRC | 23:30 | |
| NobodyCam | just a question to all, not saying we should, but: I can work down from a given chassis uuid to a port finding node along the way, what we seem unable to do is work that in reverse. ie find a the chassis for a given port or even node given only the port.. would be nice to have port.ListNode and node.ListChassis calls | 23:36 |
| jroll | we can find node by port... | 23:36 |
| jroll | although idk if there's a function for it | 23:37 |
| * jroll looks | 23:37 | |
| jroll | NobodyCam: you're right | 23:38 |
| jroll | :) | 23:38 |
| NobodyCam | :) | 23:38 |
| jroll | see line 543 or so https://review.openstack.org/#/c/101020/9/ironic/drivers/modules/agent.py | 23:38 |
| NobodyCam | ya your doing all manually | 23:39 |
| NobodyCam | :-p | 23:39 |
| jroll | right | 23:39 |
| NobodyCam | ports should know their nodes and nodes should know their chassis | 23:40 |
| mrda | if it's being done manually internally, there's an argument that it might make sense to expose that | 23:50 |
| *** nosnos has joined #openstack-ironic | 23:50 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!