| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2017-11-03 | |||
| 21:40:59 | cdent | yeah, bellows is a good metaphor. Probably worthwhile _not_ reporting everything as it gets overwhelming and there’s no way to care about everything. | |
| 21:42:18 | fried_rice | cdent Okay, so IIUC on the code side one would need to modify GET /allocation_candidates to produce its allocation_requests in the new format; update the renos, api-ref, etc. accordingly; and... anything else? | |
| 21:43:16 | cdent | that’s pretty much it. the messy bits usually come with making sure the microversions do the right thing | |
| 21:43:22 | fried_rice | cdent May actually be best holding off on that until the dust settles on The Big Refactor. | |
| 21:43:28 | fried_rice | But I'll keep an eye on it. | |
| 21:43:52 | cdent | Is it naive of me to hope that the big refactor won’t have that big of an impact on the http layer? | |
| 21:44:04 | fried_rice | It should have a zero impact on the HTTP layer | |
| 21:44:07 | cdent | jmlowe: are you/will you be in oz? | |
| 21:44:09 | fried_rice | That's the goal, anyway. | |
| 21:44:15 | cdent | “should” is a dangerous word | |
| 21:44:26 | fried_rice | Well, except for possibly fixing bugs. | |
| 21:44:54 | jmlowe | cdent: nope, couldn't make the trip, especially not with SC the next week | |
| 21:45:34 | cdent | jmlowe: ah yeah, makes sense. Give jlf a big wet sloppy kiss from me. | |
| 21:45:35 | fried_rice | cdent But that's why I'm big on getting more functional test coverage in there. So that we can *prove* the refactor doesn't affect behavior. | |
| 21:45:45 | cdent | fried_rice++ | |
| 21:45:57 | jmlowe | cdent: I'm definitely going to Vancouver and I've got pretty good odds of being able to make Berlin | |
| 21:46:07 | fried_rice | But moreover, it's uncovering and/or highlighting existing bugs. | |
| 21:46:16 | cdent | if we can agree on behavior, that will help with the bug finding… | |
| 21:46:27 | cdent | jmlowe: awesome would be good to hang out again | |
| 21:46:37 | fried_rice | That's actually easier (in *most* cases) than trying to make the code behave accordingly. | |
| 21:46:44 | jmlowe | cdent: absolutely | |
| 21:47:22 | cdent | fried_rice: I’m unsure. I mean I hope that’s the case, but … ? | |
| 21:48:29 | fried_rice | The cases that are tough to reason about involve e.g. same-rc-local-and-shared, as we've discussed previously. But leakypipes and I found some of the places that deal with that; he put a stake in the ground that we ignore the shared provider in that scenario; and I forced him to put code comments describing that behavior and referencing the bug. | |
| 21:48:39 | fried_rice | cdent So I would say that's progress. | |
| 21:48:53 | cdent | progress is nice | |
| 21:49:10 | cdent | I’m going to progress around outside a bit to try and fake myself into thinking it is not nighttime | |
| 21:49:17 | fried_rice | cdent Enjoy! | |
| 21:58:38 | mriedem | dansmith: i filled in the placement update etherpad agenda a bit, feel free to modify https://etherpad.openstack.org/p/SYD-forum-nova-placement-update | |
| 23:05:44 | openstackgerrit | Jay Pipes proposed openstack/nova master: placement: adds REST API for nested providers https://review.openstack.org/384807 | |
| 23:05:45 | openstackgerrit | Jay Pipes proposed openstack/nova master: placement: update client to set parent provider https://review.openstack.org/385693 | |
| #openstack-nova - 2017-11-04 | |||
| 00:28:26 | gunix | hey guys | |
| 00:28:42 | gunix | how does nova configure key pairs on the VM? i mean, how does it upload the key to the instance? | |
| 00:33:26 | efried | gunix I'm no expert, but possibly through config drive? | |
| 00:33:38 | gunix | what is a config drive? | |
| 00:34:13 | efried | It's a thing that gets attached to the instance when it's spawned. It can contain instance-specific data like its hostname and IP address. | |
| 00:34:23 | efried | But it's entirely possible that key pairs get handled through some other mechanism. | |
| 00:34:47 | gunix | yea, ip confuses me too | |
| 00:35:11 | gunix | i mean well, dhcp agent handles that | |
| 00:35:14 | gunix | i guess based on mac | |
| 00:35:25 | efried | For DHCP setups | |
| 00:35:33 | gunix | however, ssh key gets somehow injected. you find it on the vm, within .ssh/authorized_keys | |
| 00:36:14 | efried | I'm afraid you may not have much luck in IRC at the moment - the summit starts on Monday and it's in Sydney, so most people are either on planes or already there, where it's tomorrow. | |
| 00:36:21 | gunix | i wonder if qemu for example opens the qcow2 file, finds the file located at /root/.ssh/authorized_keys and adds a line to it | |
| 00:36:43 | gunix | yea and the flight is long | |
| 00:37:13 | efried | So the config drive contains data, and a thing called cloud-init runs and pulls data from it somehow to do things that are instance-specific. | |
| 00:37:27 | gunix | strange part is i sent an email to someone from OSA regarding a bug fix which i am proud of since it's my first bugfix but i won't get an answer anytime soon | |
| 00:37:59 | efried | Mm, new to the community? Get ready for a surprising amount of cultural indoctrination. | |
| 00:38:28 | gunix | what do you mean with cultura indoctrination? | |
| 00:38:52 | efried | It just takes a while to get used to the pace and cadence of how work gets done. | |
| 00:39:09 | gunix | yea, i sent an email instead of doing just a git push | |
| 00:39:21 | gunix | i don't want to damage any process | |
| 00:40:17 | gunix | whatever, i guess i am going to sleep with the sound knowledge that nova-compute key injection is dark magic, powered by robotic penguins | |
| 00:42:24 | efried | mdrabe Any ideas here? | |
| 00:43:23 | efried | gunix You can do git review to put your change up in gerrit. Sending an email is likely to be an even slower process, if it gets you anywhere at all. You won't hurt anything by putting a review up in gerrit. | |
| 00:43:53 | gunix | what is gerrit? i am not puttin anything in people i don't know | |
| 00:44:12 | efried | Okay, let's back up. What's the bug number? | |
| 00:44:18 | gunix | no bug number | |
| 00:44:27 | gunix | i can FW you the mail and you will get a good feeling what it's about | |
| 00:44:30 | efried | Okay. | |
| 00:48:31 | efried | gunix Okay, so you've never sent code to openstack before? | |
| 00:48:43 | gunix | nope | |
| 00:49:31 | efried | gunix I can't stick around, but you'll want to start here: https://wiki.openstack.org/wiki/How_To_Contribute | |
| 00:50:22 | efried | gunix Pay special attention to the "If you're a developer" section to figure out how to contribute code. | |
| 00:50:54 | gunix | efried: any section for "If you're a wannabe developer? | |
| 00:51:21 | gunix | s/r\?/r\"\?/g | |
| 00:51:45 | efried | gunix From your email, it looks like you've got a start on what you think the fix needs to be. That's good enough for a beginning. | |
| 00:52:55 | gunix | efried: please excuse the bad jokes, but if (brain.uptime >= 18h) jokemode = yes | |
| 00:53:15 | gunix | i find it hard to talk seriously at 3 am | |
| 00:53:27 | efried | gunix You'll want to clone the project openstack/openstack-ansible-os_tempest - see https://review.openstack.org/#/admin/projects/openstack/openstack-ansible-os_tempest | |
| 00:53:53 | gunix | efried: all 3 branches ... | |
| 00:53:57 | efried | gunix I'm unfortunately in that jokeless no-man's land between EOD and end of week. | |
| 00:53:59 | gunix | the error is on all branches :D | |
| 00:54:15 | efried | Start with master, always. Once your fix is approved, you can worry about backporting. | |
| 00:54:20 | efried | if appropriate. | |
| 00:54:34 | gunix | i mean, it's not an error ... it's just bad user experience for begginers | |
| 00:54:45 | gunix | people trying out that osa aio will have no idea how to fix that | |
| 00:55:10 | gunix | efried: can i just git clone the branch, modify the file and push ? | |
| 00:55:23 | gunix | with appropriate comment | |
| 00:55:25 | efried | Not push. Otherwise, yeah, pretty much. | |
| 00:55:45 | gunix | oh wait | |
| 00:55:50 | efried | Spider your way through that contributor guide, accept the licenses, get your launchpad ID, bootstrap your gerrit account, figure out how to work your sandbox, make your fix, and submit it via `git review`. That'll create a patch in the gerrit system (which we use for code reviews). | |
| 00:55:53 | gunix | Learn how to work with our Gerrit review system. | |
| 00:55:57 | efried | Just so. | |
| 00:56:27 | efried | You may want to join the #openstack-ansible channel, which is where I imagine people who know/care about this thing might hang out. | |
| 00:56:33 | gunix | i have a launchpad id. i used it for some article on ubuntu wiki | |
| 00:56:37 | efried | But like I said, it may be pretty sparsely populated for the next couple of weeks. | |
| 00:57:13 | gunix | efried: i stay a lot on that channel. that's why i mailed cloudnull. he is like ALWAYS helping people out. helped me a lot too. i hope he'll be proud to see me bugfixing, since i learned OSA from him | |
| 00:57:35 | gunix | how long's the summit? a week? | |
| 00:57:54 | efried | Oh, perfect. You should buzz him or email him once you have a patch up in gerrit. Copy me too - I'll be interested to see how you've done. | |
| 00:58:27 | efried | The actual summit is three days. But you know, travel to Sydney etc. And some folks take vacation before/after. So it's going to be a couple weeks before life grinds back into action. | |
| 00:59:10 | gunix | you do brasilian jiu jitsu? | |
| 00:59:17 | gunix | that's like the best fighting technique in the world | |
| 00:59:19 | efried | Yes. I own a school. | |
| 00:59:25 | gunix | nice man | |
| 00:59:29 | efried | :) | |
| 00:59:54 | gunix | that means you are part of the 0.0001% of people that can beat mcgregor? | |
| 01:00:06 | efried | I'd have to get him to the ground first. | |
| 01:00:25 | gunix | he avoids that recently | |
| 01:00:31 | gunix | since the nate event | |
| 01:00:33 | gunix | :D | |