| dmsimard | ansible 2.2 is out and it's pretty darn fast, amongst other things | 00:13 |
|---|---|---|
| *** herlo has joined #zuul | 02:08 | |
| *** herlo has joined #zuul | 02:08 | |
| *** bcoca has quit IRC | 02:42 | |
| *** saneax-_-|AFK is now known as saneax | 03:15 | |
| *** saneax is now known as saneax-_-|AFK | 04:27 | |
| *** bhavik1 has joined #zuul | 04:45 | |
| *** bhavik1 has quit IRC | 05:12 | |
| *** openstackgerrit has quit IRC | 05:48 | |
| *** openstackgerrit has joined #zuul | 05:49 | |
| *** bhavik1 has joined #zuul | 05:59 | |
| *** saneax-_-|AFK is now known as saneax | 06:43 | |
| *** yolanda has joined #zuul | 06:51 | |
| *** yolanda_ has joined #zuul | 06:51 | |
| *** yolanda_ has quit IRC | 06:52 | |
| *** bhavik1 has quit IRC | 07:02 | |
| *** openstackgerrit has quit IRC | 07:48 | |
| *** openstackgerrit has joined #zuul | 07:49 | |
| *** hashar has joined #zuul | 07:57 | |
| *** abregman has joined #zuul | 08:22 | |
| rcarrillocruz | jeblair , fungi : seems pycrypto doesn't have support for pkcs7 , it does for pkcs1 (rsa based) | 10:05 |
| rcarrillocruz | https://www.dlitz.net/software/pycrypto/api/2.6/Crypto.Cipher-module.html | 10:06 |
| rcarrillocruz | i'll work on a poc patch using it and push so you have a look | 10:06 |
| *** mmedvede has quit IRC | 11:09 | |
| *** mmedvede has joined #zuul | 11:22 | |
| pabelanger | dmsimard: ya, I haven't tested it yet. Wonder how much work it will be to upgrade zuul to start using it | 12:01 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_needed_changes_enqueue testing https://review.openstack.org/393001 | 12:16 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_source_cache unit test https://review.openstack.org/393278 | 12:19 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_one_job_project unit test https://review.openstack.org/393304 | 13:04 |
| pabelanger | jeblair: are you okay with create a new class when we need to use a different tenant_config_file ^? Figured it would be a good time to slim down the single-tenant configuration | 13:08 |
| fungi | rcarrillocruz: pkcs#7 is closely tied to s/mime, so it's possible we just need to look at a different library. unfortunately the most popular one for doing s/mime and pkcs#7 seems to still be m2crypto, which i think we had abandoned some years ago in openstack for reasons i no longer recall | 13:17 |
| *** abregman is now known as abregman|afk | 13:18 | |
| pabelanger | oh, ya, m2crypto had embedded things IIRC | 13:19 |
| *** hashar is now known as hasharAway | 13:20 | |
| *** jasondotstar has joined #zuul | 13:20 | |
| openstackgerrit | Monty Taylor proposed openstack-infra/zuul: Revert "Work with ansible 2.1" https://review.openstack.org/379769 | 13:27 |
| mordred | that should be good now that 2.2 is out | 13:27 |
| *** saneax is now known as saneax-_-|AFK | 13:49 | |
| *** yolanda has quit IRC | 14:22 | |
| *** yolanda has joined #zuul | 14:23 | |
| *** abregman|afk is now known as abregman | 14:31 | |
| *** yolanda has quit IRC | 14:33 | |
| *** yolanda has joined #zuul | 14:38 | |
| *** abregman has quit IRC | 14:44 | |
| *** hasharAway is now known as hashar | 15:44 | |
| timrc | "Channel created on Sun, 23 Aug 2009 11:50:29" -- really? | 16:06 |
| Shrews | most likely. it was originally for a now defunct project. jeblair grabbed the channel through some magic | 16:07 |
| jeblair | Shrews: i used the most secret and dark spell i know of on irc -- so mysterious and terrible it is almost never spoken of. it's name: "ask nicely". | 16:12 |
| Shrews | *shudder* | 16:13 |
| SotK | sounds terrifying | 16:13 |
| mordred | jeblair: I heard that was deprecated and you're supposed to use docker somehow now | 16:20 |
| jeblair | mordred: docker make me a sandwich | 16:21 |
| timrc | mordred: Yeah, you're suppose to run "ask nicely" in a sidekick container. | 16:21 |
| mgagne | do you know if it's possible to have nested templates in Zuul? Like a template including multiple other templates. | 16:22 |
| jeblair | mgagne: i don't believe so | 16:23 |
| mgagne | alright, thanks! | 16:25 |
| pabelanger | jeblair: speaking of templates, looking at the zuulv3 spec, we say templates are still support, but don't cite an example. Currently working towards enabling test_job_from_templates_launched and could use some syntax help if you have something handly | 16:37 |
| jeblair | pabelanger: the code exists already, and there's a template test that exercises it in the test_v3 file | 16:40 |
| jeblair | pabelanger: so hopefully you should be able to update the syntax based on the config that test uses | 16:40 |
| pabelanger | jeblair: Ah, thanks. grep is my friend | 16:41 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_job_from_templates_launched job https://review.openstack.org/393412 | 16:50 |
| pabelanger | ^ is a little different in zuulv3, seems using a template variable is no longer valid. | 16:52 |
| pabelanger | but I might have had my syntax wrong | 16:52 |
| pabelanger | still working on it | 16:52 |
| *** bcoca has joined #zuul | 17:03 | |
| jeblair | pabelanger: yes, no more variables -- the only one was 'name' and we're retiring the practice of including project names in jobs. | 17:14 |
| pabelanger | jeblair: great | 17:16 |
| *** harlowja has quit IRC | 17:43 | |
| *** harlowja has joined #zuul | 17:46 | |
| *** abregman has joined #zuul | 17:56 | |
| *** abregman has quit IRC | 18:09 | |
| *** hashar is now known as hasharAway | 18:33 | |
| SpamapS | FYI, I'm adding stories to storyboard at the moment.. pretty difficult fighting through bugs in the UI.. but we all should start seeing stories appear in this board... | 18:38 |
| SpamapS | https://storyboard.openstack.org/#!/board/41 | 18:38 |
| cinerama | cool | 18:39 |
| auggy | thanks for doing that SpamapS ! | 18:44 |
| auggy | if you have notes or something somewhere or would like some help, i'm sure folks would be happy to be delegated to, myself included :) | 18:44 |
| auggy | even if it's just throw some stubs up there and then have some folks in the project here fill them out better | 18:45 |
| SpamapS | auggy: I'm pulling from Jim's notes .. shouldn't be too hard, I've got a system now :) | 18:46 |
| SpamapS | once they're in, we can treat storyboard as the source of truth for these itesm | 18:46 |
| auggy | ok, i have no idea what kind of volume we're talking about here, so if there's any opportunity for other folks to participate, it seemed like a potentially good opportunity for that | 18:49 |
| SpamapS | auggy: I'm about 30% of the way through | 18:50 |
| SpamapS | so.. not so crazy high | 18:50 |
| auggy | SpamapS: where are these notes? some of the tickets don't have a lot of information in them and i'd be happy to review the notes and maybe add some context | 18:57 |
| SpamapS | Yeah the notes are super sparse | 18:57 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_layered_templates test https://review.openstack.org/393445 | 18:57 |
| SpamapS | jeblair: to auggy's point, a lot of these stories could use more information to help noobs get started on them. Not sure how we're going to flesh those out. | 18:58 |
| * SpamapS includes himself in the noobs list :) | 18:58 | |
| auggy | jeblair: SpamapS: maybe it's something we could divide and conquer on amongst the interested contributors | 18:58 |
| auggy | ie, once we have the list then people can work on getting more info to flesh them out | 18:59 |
| SpamapS | yeah definitely filling in the details will help get us going | 18:59 |
| * SpamapS has to switch to arch-wg for an hour | 18:59 | |
| auggy | oh right i was gonna attend that | 18:59 |
| auggy | :D | 18:59 |
| SpamapS | auggy: I have about 10 left to add | 18:59 |
| auggy | whoohoo! | 18:59 |
| auggy | dang you're a machine | 18:59 |
| SpamapS | then I will definitely revisit them and start filling in descriptions where I can | 18:59 |
| SpamapS | -> #openstack-meeting-alt if you want to attend. :) | 19:00 |
| auggy | thx! | 19:00 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_check_queue_success test https://review.openstack.org/393452 | 19:15 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_check_queue_success / test_check_queue_failure https://review.openstack.org/393452 | 19:17 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable 3 disabled tests https://review.openstack.org/393452 | 19:22 |
| Shrews | I'd add some context to some of those SB things, but I do not seem to be able to edit | 19:22 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Enable test_new_patchset_check test https://review.openstack.org/393458 | 19:24 |
| mordred | Shrews: I seem to be able to edit them - but I don't know if that's just because I'm a superadmin | 19:25 |
| pabelanger | starting to get into a rhythm now | 19:25 |
| auggy | i added a comment to one | 19:25 |
| mordred | \o/ | 19:25 |
| mordred | it's about the minor successes | 19:25 |
| auggy | https://storyboard.openstack.org/#!/story/2000773 | 19:25 |
| * Shrews shrugs and keeps his context to himself | 19:25 | |
| mordred | Shrews: did you click through to the actual story? | 19:26 |
| mordred | Shrews: you can't edit the pop-up window thing | 19:26 |
| auggy | that's where we talked about grouping the tests by feature/functionality/whatever | 19:26 |
| mordred | Shrews: but you should be able to edit here: https://storyboard.openstack.org/#!/story/2000767 | 19:26 |
| Shrews | mordred: ugh, ANOTHER click? *sigh* | 19:26 |
| mordred | Shrews: you saw that jeblair has been working on boartty yeah? | 19:26 |
| Shrews | i wanted to be super lazy | 19:26 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Enable test_new_patchset_check / test_abandoned_check https://review.openstack.org/393458 | 19:26 |
| Shrews | mordred: yup | 19:26 |
| auggy | i'm able to edit | 19:27 |
| auggy | i has the power! | 19:27 |
| Shrews | auggy: ID-10-T error | 19:27 |
| mordred | it's worth pointing out pabelanger's patch above to everyone -- it mentions the concept of an "abide" which is a vocabulary word in v3 - but also always an opportunity to remind people what The Dude does | 19:28 |
| auggy | hahahaha | 19:29 |
| pabelanger | the dude | 19:29 |
| *** mmedvede has quit IRC | 19:29 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable 4 disable tests https://review.openstack.org/393461 | 19:30 |
| pabelanger | I'm sure a lot of this can be squashed if needed, I'm basically going test, by test now | 19:30 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable 3 disabled tests https://review.openstack.org/393452 | 19:33 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Enable test_new_patchset_check / test_abandoned_check https://review.openstack.org/393458 | 19:33 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_layered_templates test https://review.openstack.org/393445 | 19:33 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable 4 disable tests https://review.openstack.org/393461 | 19:33 |
| mordred | pabelanger: you are a machine | 19:34 |
| pabelanger | mordred: now to fill my gas tank with some food | 19:34 |
| timrc | pabelanger: I didn't read up completely so maybe this was discussed but re:templated projects... | 19:37 |
| timrc | Take a look at https://review.openstack.org/gitweb?p=openstack-infra/zuul.git;a=tree;f=tests/fixtures/config;hb=ff6582d75c95859d456e588e0e74c602b9664fdb -- are project-template and templated-project fundamentally different? | 19:38 |
| *** hasharAway is now known as hashar | 19:46 | |
| pabelanger | timrc: Ya, they appear to be the same thing. I suspect we can remove templated-project patch | 19:46 |
| timrc | Also, this is just me thinking outloud here, but do we really consider these templates? It feels like a deviation from what people have come to expect as a template. Can projects just be composed of other projects? | 20:02 |
| jeblair | timrc: a project is a git repo; i don't think project composition makes sense. a template is a pattern of pipeline-job configuration which can be applied to multiple projects. having a project follow the 'publish-to-pypi' template makes sense to me. it is also exactly what we call templates in zuulv2. | 20:07 |
| timrc | jeblair: I'm just thinking about what the term template means across projects not necessarily what it meant in zuulv2->v3. But yeah, I see how project/sub-project would not work given the definition of a project primitive. | 20:11 |
| timrc | erm | 20:11 |
| timrc | overloaded the term project there | 20:11 |
| timrc | about what the term template menas across other software projects (puppet, ansible, django, whatever) | 20:11 |
| timrc | even in jjb is closer to what I'd consider a template, generating something new with input + boilerplate. | 20:13 |
| jeblair | timrc: i think that's an expansion of the plain english form of the word which allows for use without change -- in fact, a template is used to make the same thing over and over. i don't expect people to have a problem with it either way. | 20:15 |
| *** abregman has joined #zuul | 20:17 | |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Enable test_statsd test https://review.openstack.org/393479 | 20:28 |
| SpamapS | jeblair: FYI, on the board, 'todo' is "next steps", backlog is 'things we think we need to do' | 20:42 |
| SpamapS | that should probably go in the description | 20:43 |
| jeblair | oh yeah, that'd be a good thing for the board description | 20:43 |
| SpamapS | also I've been making stories when I'm not sure how isolated things are from one another. The line between story and task I think is sometimes blurry. | 20:44 |
| jeblair | SpamapS: there's something that i'm not quite sure how to express yet, which is that we have a couple of arcs -- in our 3 or 4 phase rollout plan, we have 1) roll out nodepool zk builders; 2) roll out nodepool zk launchers; 3) roll out zuulv3 | 20:44 |
| jeblair | SpamapS: (indeed... ^ that goes to that point about story/task lines as well) | 20:45 |
| SpamapS | Right, well I think we could manage each of those arcs as a work board.. or just put all the things for arc #1 in todo first, and then leave the rest in backlog | 20:45 |
| SpamapS | I kind of prefer the latter.. but it means the workboard won't be usable as an overall work status view.. it's just for an idea of what's going on _now_. | 20:46 |
| jeblair | SpamapS: yeah. also, we're totally in "pick something and try it for a bit" mode... so, easy come easy go | 20:46 |
| jeblair | SpamapS: at any rate, in the "roll out nodepool zk builders" plan, we're actually fairly close to that. we need a few more tasks to flesh it out. but i think the next one on the list is actually the "switch nodepool zk module to use classes instead of dicts" | 20:47 |
| SpamapS | Yep.. the only reason I want the board is so people pick different things from eachother. | 20:47 |
| jeblair | SpamapS: that's in service of "incorporate zk builder back into main nodepool" and "enable more zk builder tests" -- both tasks that i should probably go create right now :) | 20:47 |
| *** mmedvede has joined #zuul | 20:48 | |
| jeblair | but those are the things that, roughly speaking, i think need to get done before we roll it out into production (Shrews may also know of some things i'm missing there) | 20:49 |
| SpamapS | jeblair: so I'll get all of these stories/tasks input here in the next few.. and then maybe you can sift through backlog and make sure everything that needs to be is in todo | 20:51 |
| jeblair | SpamapS: k. i added those two things i just said as tasks on https://storyboard.openstack.org/#!/story/2000767 | 20:55 |
| SpamapS | hooray | 20:58 |
| SpamapS | jeblair: "use ssl with gearman (openstack puppet deployment) | 20:58 |
| SpamapS | that doesn't seem related? | 20:58 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_live_reconfiguration test https://review.openstack.org/393488 | 21:00 |
| pabelanger | ^ actually exposed a bug I think | 21:01 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Re-enable test_live_reconfiguration test https://review.openstack.org/393488 | 21:02 |
| SpamapS | jeblair: ok, the board now has everything from your email, except ssl with gearman and python3, since I think neither of those are needed for base operational functionality. Though they are needed, I presume, for infra rollout? | 21:04 |
| SpamapS | I have updated the description for https://storyboard.openstack.org/#!/board/41 | 21:07 |
| SpamapS | auggy: around? Can you try moving a task around on that? I want to see if non-owner non-users can edit the board | 21:07 |
| auggy | SpamapS: sure, i was able to edit the ticket about the tests that you created before | 21:09 |
| auggy | lol 400: GET /api/v1/users/preferences: Invalid input for field/attribute user_id. Value: 'preferences'. unable to convert to int | 21:09 |
| SpamapS | auggy: when do you get that? | 21:10 |
| SpamapS | auggy: just trying to move things on the board? | 21:10 |
| auggy | hrm trying but sorryboard doesn't seem to be loading correctly | 21:10 |
| auggy | still working on it | 21:10 |
| jeblair | SpamapS: we need to switch to using gear ssl for openstack infra since we will be shipping credentials across it. the python3 stuff will be needed for websockets based console log streaming. | 21:10 |
| auggy | how do you move things across, is dragon drops a thing ? | 21:12 |
| jeblair | SpamapS: so yeah, the gear thing is an infra-specific rollout. python3 can be kicked down the road if we don't want to include that in the initial release, but i think we need it early on. it is a feature that is... widely expected. :) | 21:12 |
| jeblair | auggy: yes | 21:15 |
| Shrews | jeblair: SpamapS: fwiw, i believe i was the one to add ssl support to gear waaaaaay back when. i hope it works :) | 21:15 |
| jeblair | Shrews: it's tested :) | 21:15 |
| Shrews | w00t | 21:15 |
| jeblair | (all the gear tests run with both ssl and non-ssl. that sounds excessive now that i say it.) | 21:16 |
| Shrews | wow, i added that more than 3 years ago. and for the life of me, i cannot remember what for | 21:17 |
| Shrews | oh, maybe libra | 21:17 |
| SpamapS | jeblair: oh I think socket streaming viewing is in our mvp. | 21:17 |
| SpamapS | "hang on I'm running your job" is a really awful experience | 21:17 |
| auggy | so yeah i have no dragon drops | 21:18 |
| SpamapS | right now I believe gearmand's SSL support is borken because of modern OpenSSL versions.. so gear is ahead of the curve. ;) | 21:18 |
| SpamapS | auggy: so it looks like a static list for you? | 21:19 |
| auggy | yeah when i click on things it just highlights | 21:19 |
| auggy | i can't pick anything up | 21:19 |
| jeblair | SpamapS: yeah. "telnet" is, of course superior, but requires "user re-education" which is not in the mvp | 21:19 |
| jeblair | SpamapS, auggy: we probably need to add auggy as either a user or owner of the board, not sure what the diff is. | 21:20 |
| auggy | https://usercontent.irccloud-cdn.com/file/0In2GwOg/Screenshot%20from%202016-11-03%2014%3A19%3A39.png | 21:20 |
| auggy | so when i click on any one of those cards, nothing happens, when i try to drag, it just highlights the text | 21:21 |
| jeblair | SpamapS: the websockets / python3 thing is going to be tricky. it's not a simple task -- we need to think carefully about the gear api in python3. so we should budget appropriately for that. | 21:21 |
| SpamapS | yeah bytes and strings, fun | 21:21 |
| jeblair | zactl | 21:21 |
| jeblair | y | 21:22 |
| SpamapS | though I think for the most part the thing we found in pysaml2 (which has similar challenges) was "all public api calls ingest and emit strings unless an encoding is already included in the bytes content" | 21:23 |
| SpamapS | then again, pysaml2 has the benefit of xml | 21:23 |
| jeblair | SpamapS: yeah, the problem arises from you might be using this to send json. or you might be using it to send jpeg. | 21:23 |
| SpamapS | auggy: ok I'm going to add you as a user.. | 21:23 |
| SpamapS | auggy: can you refresh now? | 21:24 |
| auggy | SpamapS: ooh it's like tron | 21:24 |
| jeblair | (my *guess*: users can move boxes around, owners can additionally modify the acls) | 21:24 |
| jeblair | auggy: can tell us if that's the case in a minute :) | 21:24 |
| SpamapS | jeblair: yeah with gear, I think bytes may be the thing. | 21:24 |
| auggy | YESH IT WORKS | 21:25 |
| auggy | i has the dragon drops! | 21:25 |
| SpamapS | though we could definitely duck type and say if you give us a string it's going to get encoded as utf-8. | 21:25 |
| jeblair | SpamapS: anyway, there's a whole ml list chain about it from 5 months ago we can dust off when we're ready. mostly wanted to convey that it's not just "go sprinkle some six around". | 21:25 |
| SpamapS | but that doesn't mean it's coming out as a string on the other end, just as bytes. | 21:25 |
| SpamapS | Yeah, all things with I/O are challenging with python3.. but if you think long and hard about it, that's what they point was of being explicit. | 21:26 |
| SpamapS | less magic | 21:26 |
| auggy | jeblair: which ml list? infra? maybe we could paste a link to that discussion in the bug for now? i'm happy to do it :) | 21:26 |
| SpamapS | jeblair: forgive me the deep dive.. I've been project managing all day and my engineer muscle is really twitchy ;) | 21:26 |
| jeblair | auggy: yes infra. | 21:26 |
| auggy | when my dog takes bytes it often comes out as a string of sorts on the other end | 21:26 |
| SpamapS | auggy: ok so we basically have to add everyone who wants to work on zuul to it | 21:26 |
| SpamapS | I feel a feature request coming on | 21:27 |
| auggy | hahaha yeahhh, how do we make storyboard requests? | 21:27 |
| auggy | just pop into story board and ask? | 21:28 |
| jeblair | auggy: or make a story in storyboard | 21:28 |
| jeblair | or both | 21:28 |
| auggy | although i think the idea might be a permissions thing... like do you want non-zuul people to be able to move things around? | 21:28 |
| auggy | or yeah i dk.. i just realized i can modify zuul bugs | 21:28 |
| auggy | so that shouldn't matter | 21:28 |
| auggy | the perms should be consistent either way | 21:28 |
| * auggy is just thinking about how to file the storyboard bug in storyboard | 21:29 | |
| auggy | so many layers to this one whoa | 21:29 |
| jeblair | auggy: i think the idea is that boards are used to set priority, which is the purview of an individual or group, whereas stories are the unfiltered input into that | 21:29 |
| auggy | ohhhh right | 21:30 |
| auggy | so it's priority *to you* | 21:30 |
| auggy | so you have to be part of the entity that has whatever that priority perspective is? | 21:30 |
| jeblair | auggy: yep. so this board will be "our team's priority" | 21:30 |
| auggy | ok | 21:31 |
| auggy | hrm so if we wanted to file a ticket for storyboard to change that... what exactly would we want? for anyone to be able to see that perspective? | 21:31 |
| auggy | not see but modify | 21:31 |
| jeblair | auggy: well, the boards are public, so anyone can see "what we think needs to be worked on to make zuulv3 operational", they just can't change what we think needs to be worked on. | 21:32 |
| auggy | right, so when we talk about wanting to file a ticket, is it that we want to change that ability? | 21:32 |
| auggy | how about if there was some message like "if you want to modify this you need to be part of x group" and a link to add yourself or request to be added? | 21:33 |
| jeblair | auggy: well, i don't actually have anything better to offer the storyboard team right now in terms of ideas other than "just used fixed priority on the story". so i'm comfortable trying this out the way they thought it might should be used and seeing how it goes. | 21:33 |
| jeblair | auggy: yeah, an explanation might be nice. even something like "This board is managed by ..." | 21:34 |
| auggy | yeah so one thing i am thinking about now, i didn't even know that modifying that was even an option | 21:34 |
| jeblair | that would be nice for storyboard to have, though we could probably put that into our description right now... | 21:34 |
| auggy | so some indication of group membership would be useful so people know.. | 21:34 |
| jeblair | auggy: can you edit the board description? | 21:34 |
| auggy | yeah as an interim | 21:34 |
| auggy | jeblair: yeah i can | 21:34 |
| auggy | what's the group that people should belong to? | 21:35 |
| auggy | and how do they request to be in it? | 21:35 |
| jeblair | hrm, let's go to #storyboard | 21:35 |
| auggy | well actually i can't | 21:35 |
| auggy | i thought i could edit it | 21:35 |
| auggy | oh wait i think i get it | 21:41 |
| auggy | so SpamapS did you have to specifically add me to have access to that board? | 21:41 |
| auggy | whereas if we had some generic "zuul team" that had that access, we could just add people to that and then as we run into things, just add that team to have access | 21:41 |
| jeblair | auggy: yes, the board has an acl | 21:42 |
| jeblair | auggy, SpamapS: so i guess for the moment, our process will be "complain if you need to move boxes around and someone will give you access to move boxes around" | 21:44 |
| auggy | haha | 21:44 |
| auggy | how can i edit the description to update it with that info? | 21:44 |
| jeblair | auggy: since, yeah, storyboard has "teams" but i can't see that they are able to be used for this | 21:45 |
| jeblair | auggy: do you have a pencil icon next to the board title on https://storyboard.openstack.org/#!/board/41 ? | 21:45 |
| auggy | no pencils for me | 21:47 |
| auggy | only dragon drops | 21:47 |
| SpamapS | Right, auggy has user access | 21:47 |
| SpamapS | auggy: you can add things right? | 21:48 |
| auggy | well i'm not sure | 21:48 |
| auggy | i'm trying to | 21:48 |
| auggy | ah i see | 21:49 |
| auggy | so if you create a new story it doesn't automatically select it for you in the add stories, you still have to search for it first | 21:50 |
| auggy | *sigh* | 21:50 |
| auggy | this is taking dog fooding to a whole new level i see ;D haha | 21:50 |
| auggy | SpamapS: yes i can add a story | 21:50 |
| * auggy needs to go afk for a bit | 21:53 | |
| auggy | so if i don't respond immediately... it's cuz i'm not here! | 21:53 |
| SpamapS | auggy: thanks for helping us test | 21:54 |
| mordred | jeblair, SpamapS: I believe we can do the websockets thing without python3 at this point - although I'd like to discuss the various options with people at some point when we're a little futher along | 21:55 |
| jeblair | mordred: okay. i mean, python3 is a thing we need to do eventually anyway... the last time we discussed it, the options not involving python3 seemed terrible... but if they are better, we can certainly reconsider :) | 21:56 |
| mordred | the python2 options include some things I personally and infra more broadly consider 'ick' - but that aren't necessarily bad if they're properly contained. considering the care gear3 will need, it may be more expeditious to do one of them instead of lose ourselves down the gear3 hole | 21:56 |
| *** harlowja has quit IRC | 21:56 | |
| jeblair | mordred: i actually do not think that the gear3 hole will be that bad. i think it will be far better than 'ick'. | 21:57 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Enable 4 CRD unit tests https://review.openstack.org/393494 | 21:58 |
| mordred | jeblair: awesome. I mostly wanted to add into the conversation that, depending on how gear3 winds up being scoped, there are phyiscal possibilities that do not necessitate solving gear3. that said - solving gear3 and moving forward cleanly is my preference | 21:59 |
| jeblair | mordred: yeah, i hope we can do that and reserve 'contained ick' as a fallback plan :) | 22:01 |
| clarkb | right I think its a good idea to not have a hard requirement there if it isn't actually a hard requirement. Which is what I think I was trying to argue for in the past | 22:01 |
| clarkb | makes solving the problems easier that way | 22:01 |
| openstackgerrit | Paul Belanger proposed openstack-infra/zuul: Enable 4 CRD unit tests https://review.openstack.org/393494 | 22:03 |
| pabelanger | okay, that's it for me today. If people are interested in reviewing zuulv3 patches: https://review.openstack.org/#/q/status:open+project:openstack-infra/zuul+branch:feature/zuulv3+topic:enable-tests | 22:03 |
| pabelanger | pretty straighforward stuff | 22:03 |
| SpamapS | The gear3 hole should be pretty shallow. I'm pretty comfortable jumping into that hole actually. :) | 22:06 |
| SpamapS | Just needs careful thought. | 22:06 |
| mordred | SpamapS: yup | 22:06 |
| mordred | clarkb: and yup | 22:06 |
| mordred | I agree with everyone all at the same time!!! | 22:06 |
| SpamapS | Also I'll have a hard time not also tumbling into the "make geard do the cool things gearmand does" hole at the same time. :) | 22:06 |
| mordred | pabelanger: you're winning at all the things | 22:07 |
| mordred | https://review.openstack.org/#/c/393479 is the best patch, btw | 22:08 |
| jeblair | pabelanger: you probably found the one test like that :) | 22:13 |
| openstackgerrit | Caleb Boylan proposed openstack-infra/nodepool: Fix subnode deletion https://review.openstack.org/370455 | 22:30 |
| *** toabctl has quit IRC | 22:42 | |
| *** toabctl has joined #zuul | 22:42 | |
| *** hashar has quit IRC | 23:03 | |
| SpamapS | jeblair: btw, gear's functional tests already pass in python3. Is the coverage really bad? | 23:12 |
| SpamapS | well the whole tox suite passes I should say | 23:12 |
| jeblair | SpamapS: yeah, i think the actual issue is that *using* gear as written in py3 is the worst thing ever | 23:16 |
| SpamapS | yeah playing around, requiring bytes for function names makes very little sense. | 23:16 |
| jeblair | SpamapS: .encode()/.decode() is required on like every call into or out of gear. | 23:17 |
| SpamapS | also it does not appear to work | 23:17 |
| SpamapS | INFO:gear.Server:Received packet from <gear.ServerConnection 0x7fb65e0ac470 name: None host: 127.0.0.1 port: 60750>: <gear.Packet 0x7fb65e0ac3c8 type: SUBMIT_JOB_BG function: b'foo' unique: b''> | 23:17 |
| jeblair | details | 23:17 |
| SpamapS | ERROR:gear.Server:Received unhandled packet: <gear.Packet 0x7fb65e0ac3c8 type: SUBMIT_JOB_BG function: b'foo' unique: b''> | 23:17 |
| jeblair | SpamapS: yeah, i don't think the server implements that | 23:18 |
| SpamapS | oh BG's? ok | 23:18 |
| SpamapS | I just use them to get non block, doh | 23:18 |
| jeblair | SpamapS: (the server is very limited in functionality; the client should be fully featured) | 23:18 |
| mordred | SpamapS: do you own the server now too? | 23:19 |
| mordred | (the C one) | 23:19 |
| *** saneax-_-|AFK is now known as saneax | 23:19 | |
| mordred | (of course you do - duh- it's drink o'clock and I don't have a drink yet) | 23:20 |
| mordred | jeblair: what's the thing gear does that we use that never made it in to the C server which prevents us using the C server with zuul/nodepool? | 23:20 |
| SpamapS | mordred: I do | 23:20 |
| jeblair | the only reason left for us not to use the c server is that we need strict fifo for each of the 3 priorities. i know that's not the default for the c server, but i don't know if it's possible to coax it into that behavior. | 23:20 |
| jeblair | mordred: ^ (was already typing :) | 23:20 |
| mordred | jeblair: jinx | 23:21 |
| mordred | jeblair: the cancel admin function made it in already, right? | 23:21 |
| SpamapS | strict fifo? | 23:21 |
| jeblair | mordred: aiui yes | 23:21 |
| SpamapS | did you ever type that up in a launchpad bug or GH issue? | 23:21 |
| SpamapS | Also I believe the recent addition of registering many functions at once isn't in gearmand. | 23:21 |
| SpamapS | Though I'd happily merge it. | 23:22 |
| jeblair | SpamapS: no idea. was years ago. | 23:22 |
| jeblair | SpamapS: yes, though happily we won't need 'mass_do' in zuulv3. (we will only have like 10 functions) | 23:22 |
| SpamapS | Right | 23:23 |
| SpamapS | (and that's generally been the position of the gearman authors in the past... "Just stuff your sub-job into your payload") | 23:23 |
| jeblair | mordred, SpamapS: i would be happy to switch to the C server if we can (performance yay!), though i'll still use the geard server for zuul unit tests. and the py3 issue is related to the client api, so it doesn't solve that. | 23:24 |
| SpamapS | yeah | 23:24 |
| mordred | ++ | 23:24 |
| SpamapS | no I was just poking at _BG... that's not needed | 23:24 |
| SpamapS | IMO most of the interesting uses of gearman do not involve _BG jobs | 23:25 |
| jeblair | SpamapS: here's the last place my brain was on this issue: http://lists.openstack.org/pipermail/openstack-infra/2016-June/004464.html | 23:26 |
| SpamapS | jeblair: thanks, I had been trying to dig that up and gotten distracted | 23:27 |
| jeblair | yeah, i think i was halfway through digging it and then squirrel! | 23:28 |
| Shrews | jeblair: re: "enable more zk builder tests"... there are surprisingly few tests... 2 really. only 1 is disabled | 23:34 |
| Shrews | perhaps we should change that to "write new builder tests" | 23:34 |
| Shrews | s/new/additional/ | 23:34 |
| mordred | Shrews: or maybe the task is really easy and only has one step ;) | 23:38 |
| clarkb | SpamapS: jeblair re making gears api nicer for py3k one of the ideas I liked (and I am not sure who had it or if it was the one settled on) was to keep the core of gear binary, then have wrapper classes for things that would be configured with an encoding at instantiation for people dealing with strings in python | 23:38 |
| clarkb | then its StringClient('utf8) instead of Client() or whatever | 23:38 |
| clarkb | and the raeson for that is gearman is a binary protocol and as you say payload could be jpg | 23:39 |
| jeblair | Shrews: well, if we need to. i'd rather not have more tests than needed. it does get exercised a lot, so may have better coverage than is apparent. :) but at any rate, new tests probably aren't a blocker for initial rollout. | 23:39 |
| jeblair | clarkb: yeah, that idea had slipped my mind. i think that might be a good solution. | 23:39 |
| SpamapS | clarkb: +1 | 23:40 |
| clarkb | there should be quite a few tests that exercise the builder in nodepool | 23:40 |
| clarkb | there are the cli tests, the normal functioanl tests, and then the builder tests themselves | 23:40 |
| jeblair | clarkb, Shrews: the entire TestNodepool test class is skipped currently because every nodepool test uses the builder | 23:41 |
| jeblair | (can't boot a node without an image) | 23:41 |
| jeblair | so the good news is that, unlike zuul, re-enabling them should be easy... | 23:42 |
| jeblair | we do the dict/object conversion (so that the zk api is closer to the existing db api), incorporate the zk api into nodepool itself, then the tests should work without significant alteration (should just be the configuration of the test fixtures) | 23:43 |
| jeblair | then we turn on the devstack test, then production | 23:43 |
| jeblair | clarkb, SpamapS: yeah, i think a gear where all of the function names and unique ids are 'utf8' (i know it's not in the spec, but i'd be comfortable with gear asserting that -- see the first part of my email), and then data is binary by default, but have a StringWorker that gives you decoded data (and possibly a corresponding StringClient) would fit the bill. | 23:48 |
| SpamapS | jeblair: I was just poking at that, but calling it TextWorker and TextClient because String is too overloaded in python land. | 23:56 |
| SpamapS | and it aligns with six.text_type | 23:56 |
| jeblair | SpamapS: it has fewer chrctrs too. ++ | 23:56 |
| SpamapS | I'm also just needing to code-nerd-out a bit.. | 23:57 |
| SpamapS | the challenge I have right now is how to test it without rewriting tests | 23:57 |
| SpamapS | just seeing if I can do it with some testscenarios magic | 23:57 |
| mordred | testscenarios | 23:58 |
| mordred | SpamapS: also - sorry - I haven't solved the testscenarios/load_tests thing with ttrun yet | 23:58 |
| jeblair | *magic* | 23:58 |
| mordred | SpamapS: just don't want you to think I forgot | 23:58 |
| SpamapS | mordred: I figure you'll get annoyed with that at some point too and do it. | 23:58 |
| mordred | yah | 23:58 |
| SpamapS | the only way I know to do it is to go ahead and grep subunit the way testr does | 23:58 |
| SpamapS | which may defeat the purpose you had | 23:59 |
| mordred | well - you can do a discover step | 23:59 |
| mordred | which, yeah, is slow - but all that really needs to happen is running load_tests in the module | 23:59 |
| mordred | that you already know | 23:59 |
| mordred | so I _think_ we should be able to do it without too much cost | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
