| SpamapS | corvus: so promote has moved forward a bit, but what's interesting to me is that you can't have both change, and final sha, in the same pipeline, still. | 00:32 |
|---|---|---|
| SpamapS | corvus: the reason I want the sha is that I don't want to move a static tag like 'latest' or 'release' forward, but I want to always have the tag change, so I can force Kubernetes to redeploy. | 00:33 |
| SpamapS | But, I think I can do it by actually just reading the tag, and turning it into the layer hash. | 00:33 |
| clarkb | SpamapS: you update the image:sha in k8s to force it to redeploy? | 00:34 |
| SpamapS | clarkb: yes, otherwise k8s has no idea the container image change. | 00:35 |
| SpamapS | d | 00:35 |
| clarkb | ya. Our "plan" for that is to bump a var value | 00:35 |
| SpamapS | but if you poke a new image tag into the spec, k8s does a nice automatic rolling update | 00:35 |
| clarkb | the end user interaction is roughly the same "go change this value to something else" | 00:35 |
| SpamapS | And more importantly, if you are like "oh noes this is bad" you can roll back, and it will work. | 00:35 |
| SpamapS | Yeah no, yours is going to break. | 00:35 |
| clarkb | well it won't rollback | 00:35 |
| SpamapS | right | 00:36 |
| clarkb | unless you rollback the image | 00:36 |
| clarkb | which is doable | 00:36 |
| SpamapS | yeah, I think I'd rather poke the new tag in :) | 00:36 |
| SpamapS | Actually I'm looking at making the promote pipeline the one that does all the dirty work, and ditching my post jobs. | 00:36 |
| SpamapS | Since then I'll have the change #, I can just shove the change_# tag in. | 00:36 |
| clarkb | ya we don't have post jobs in our prmote setup | 00:37 |
| clarkb | gate builds an image and uploads it, promote promotes it | 00:37 |
| SpamapS | Yeah that's where I'm headed | 00:37 |
| clarkb | and ya you'll have the change tag on that which you could use instead of latest | 00:37 |
| clarkb | so that should work for your case | 00:37 |
| SpamapS | Yeah, that also is actually better for debuggers. | 00:38 |
| SpamapS | If somebody rolls back, they can also go "what changed?" and just find PR#X | 00:38 |
| SpamapS | And then the way I've written things, if there's no zuul.change, I can just fall back to pipeline_promote as "The last one promoted" | 00:39 |
| SpamapS | which is good for bootstrapping | 00:39 |
| SpamapS | I feel like this set of ECR roles might want to live in zuul-jobs... | 00:51 |
| SpamapS | https://github.com/GoodMoney/goodmoney-zuul-jobs/pull/19/files | 00:51 |
| *** strigazi has quit IRC | 00:55 | |
| *** strigazi has joined #zuul | 00:56 | |
| *** strigazi has quit IRC | 01:05 | |
| *** strigazi has joined #zuul | 01:06 | |
| *** jamesmcarthur has joined #zuul | 01:32 | |
| *** jamesmcarthur has quit IRC | 01:35 | |
| *** jamesmcarthur has joined #zuul | 01:36 | |
| *** jamesmcarthur has quit IRC | 02:01 | |
| *** jamesmcarthur has joined #zuul | 02:04 | |
| *** bhavikdbavishi has joined #zuul | 02:06 | |
| *** jamesmcarthur has quit IRC | 02:11 | |
| *** jamesmcarthur has joined #zuul | 02:12 | |
| *** jamesmcarthur has quit IRC | 02:20 | |
| *** jamesmcarthur has joined #zuul | 02:20 | |
| *** jamesmcarthur has quit IRC | 02:39 | |
| *** rlandy has quit IRC | 02:49 | |
| *** sdake has joined #zuul | 02:54 | |
| *** bhavikdbavishi has quit IRC | 03:18 | |
| *** bhavikdbavishi has joined #zuul | 04:06 | |
| *** sdake has quit IRC | 04:08 | |
| *** raukadah is now known as chandankumar | 05:09 | |
| *** saneax has joined #zuul | 06:42 | |
| *** quiquell|off is now known as quiquell | 07:01 | |
| *** saneax has quit IRC | 07:23 | |
| *** saneax has joined #zuul | 07:24 | |
| *** bjackman has joined #zuul | 07:43 | |
| *** sdake has joined #zuul | 07:44 | |
| *** quiquell is now known as quiquell|brb | 07:54 | |
| *** sdake has quit IRC | 08:23 | |
| *** quiquell|brb is now known as quiquell | 08:26 | |
| openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Implied branch matching must use fullmatch https://review.openstack.org/640272 | 08:33 |
| *** jpena|off is now known as jpena | 08:48 | |
| *** sdake has joined #zuul | 08:55 | |
| *** panda|ruck|off is now known as panda|ruck | 09:05 | |
| *** saneax has quit IRC | 09:08 | |
| *** saneax has joined #zuul | 09:09 | |
| *** saneax has quit IRC | 09:10 | |
| *** saneax has joined #zuul | 09:10 | |
| *** bhavikdbavishi has quit IRC | 09:32 | |
| *** bhavikdbavishi has joined #zuul | 09:32 | |
| *** sshnaidm is now known as sshnaidm|off | 10:00 | |
| *** electrofelix has joined #zuul | 10:02 | |
| *** ttx has quit IRC | 10:07 | |
| *** ttx has joined #zuul | 10:08 | |
| *** sdake has quit IRC | 10:20 | |
| *** saneax has quit IRC | 10:25 | |
| *** saneax has joined #zuul | 10:26 | |
| *** saneax has quit IRC | 10:30 | |
| *** gtema has joined #zuul | 10:39 | |
| *** bhavikdbavishi has quit IRC | 10:46 | |
| *** bhavikdbavishi has joined #zuul | 11:29 | |
| *** bjackman has quit IRC | 11:45 | |
| *** hashar has joined #zuul | 11:46 | |
| *** gtema has quit IRC | 11:58 | |
| *** EmilienM is now known as EvilienM | 11:58 | |
| *** dkehn has quit IRC | 12:01 | |
| *** panda|ruck is now known as panda|ruck|lunch | 12:16 | |
| *** gtema has joined #zuul | 12:18 | |
| *** threestrands has quit IRC | 12:24 | |
| *** jpena is now known as jpena|lunch | 12:51 | |
| *** panda|ruck|lunch is now known as panda|ruck | 13:04 | |
| *** rlandy has joined #zuul | 13:34 | |
| *** jamesmcarthur has joined #zuul | 13:51 | |
| *** jpena|lunch is now known as jpena | 13:54 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: [WIP] Add AuthorizationRule Configuration Object, claim validation https://review.openstack.org/639855 | 13:54 |
| *** bhavikdbavishi has quit IRC | 13:55 | |
| *** sdake has joined #zuul | 14:01 | |
| openstackgerrit | Matthieu Huin proposed openstack-infra/zuul master: [WIP] Add AuthorizationRule Configuration Object, claim validation https://review.openstack.org/639855 | 14:08 |
| *** TheJulia is now known as needssleep | 14:14 | |
| mordred | tobiash: we landed the patch to keystoneauth to use the FairSempahore for rate limiting. this will let us remove the TaskManager class from openstacksdk, which should in turn reduce the thread overhead | 14:17 |
| mordred | tobiash: it'll still take a little bit to filter this all the way through to nodepool of course, but just wanted to let you know we've still been slogging through things since your original "there are too many threads here" issue | 14:17 |
| mordred | I'm really happy about how it's all come together | 14:19 |
| *** sdake has quit IRC | 14:19 | |
| *** dkehn has joined #zuul | 14:20 | |
| *** sdake has joined #zuul | 14:21 | |
| *** sdake has quit IRC | 14:24 | |
| *** chandankumar is now known as raukadah | 14:26 | |
| *** sdake has joined #zuul | 14:39 | |
| *** hashar has quit IRC | 14:41 | |
| *** hashar_ has joined #zuul | 14:41 | |
| *** quiquell is now known as quiquell|off | 14:51 | |
| *** sdake has quit IRC | 15:10 | |
| *** bhavikdbavishi has joined #zuul | 15:20 | |
| *** hashar_ has quit IRC | 15:24 | |
| *** hashar has joined #zuul | 15:49 | |
| corvus | mordred, clarkb, AJaeger: i'd like to land https://review.openstack.org/639779 today and test it out | 15:49 |
| *** sdake has joined #zuul | 15:57 | |
| *** sdake has quit IRC | 16:03 | |
| AJaeger | corvus: Seeing your reply to evrardjp, go ahead ;) | 16:04 |
| *** sdake has joined #zuul | 16:05 | |
| AJaeger | zuul-jobs cores, could you review https://review.openstack.org/627534 for puppet building, please? | 16:05 |
| corvus | AJaeger: +2 with comment | 16:18 |
| *** kmalloc is now known as needscoffee | 16:22 | |
| *** sdake has quit IRC | 16:23 | |
| *** smyers_ has joined #zuul | 16:24 | |
| *** smyers has quit IRC | 16:25 | |
| *** smyers_ is now known as smyers | 16:25 | |
| tobiash | mordred: will that remove the taskmanager stopped exception that is problematic for single provider config reload? | 16:25 |
| mordred | tobiash: probably? | 16:26 |
| tobiash | mordred: I was about to look into that issue today but reproducing this in a test case is hard because openstacksdk is mocked there | 16:26 |
| tobiash | mordred: if I can help you with plugging this into nodepool tell me | 16:29 |
| *** hashar has quit IRC | 16:29 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Handle multiple docker images with the same repository https://review.openstack.org/639779 | 16:36 |
| corvus | tobiash: i'm going to use https://review.openstack.org/638596 as a test case for ^ to make sure image builds/publications work. hope you don't mind. :) | 16:39 |
| tobiash | sure :) | 16:40 |
| *** bhavikdbavishi has quit IRC | 17:11 | |
| *** gtema has quit IRC | 17:12 | |
| *** bhavikdbavishi has joined #zuul | 17:17 | |
| *** panda|ruck is now known as panda|ruck|off | 17:21 | |
| *** sdake has joined #zuul | 17:46 | |
| *** jpena is now known as jpena|off | 18:03 | |
| *** irclogbot_3 has joined #zuul | 18:12 | |
| *** jamesmcarthur has quit IRC | 18:20 | |
| *** jamesmcarthur has joined #zuul | 18:20 | |
| *** jamesmcarthur has quit IRC | 18:24 | |
| *** mrhillsman is now known as mrhillsman_lunch | 18:28 | |
| *** bhavikdbavishi has quit IRC | 18:46 | |
| *** electrofelix has quit IRC | 18:52 | |
| corvus | SpamapS, mordred, tobiash: zuul-preview in action: http://site.926bb0aaddad4bc3853269451e115dcb.openstack.zuul-preview.opendev.org/ | 18:54 |
| SpamapS | *nice* | 18:54 |
| tobiash | Cool | 18:54 |
| SpamapS | I lost track of the thread.. did we end up with the C++ version + whitelabel fixes? | 18:54 |
| mordred | SpamapS: we have the c++ version, I think it still needs whitelabel fixes- we also have a replace-with-rust stack up so that we can discuss amongst ourselves | 19:00 |
| *** sdake has quit IRC | 19:00 | |
| SpamapS | ruh roh, I used {build} in my failure-url and this happend: | 19:00 |
| SpamapS | https://screenshots.firefox.com/HQeZtoLROLnIV0Lq/github.com | 19:00 |
| SpamapS | I theenk I found a buuug | 19:00 |
| SpamapS | mordred: cool, wasn't sure if that went anywhere | 19:01 |
| clarkb | mordred: I was thinking it might be good to also capture why/what performance we think is important there. Like python startup time is bad but once its started we are only doing IO. (At least that might help my understand why C++ or Rust might be preferable | 19:02 |
| SpamapS | clarkb: no, python failed pretty hard even after startup time. | 19:03 |
| clarkb | SpamapS: oh? | 19:03 |
| clarkb | was there testing of this somewhere? | 19:03 |
| clarkb | (there is a big lack of info is my point) | 19:03 |
| SpamapS | http://lists.zuul-ci.org/pipermail/zuul-discuss/2019-February/000750.html | 19:03 |
| clarkb | SpamapS: re your bug, ya zuul doesn't type check those attrs, you have to be careful to set the correct ones | 19:03 |
| SpamapS | http://lists.zuul-ci.org/pipermail/zuul-discuss/2019-February/000747.html has the initial tests | 19:04 |
| clarkb | SpamapS: right you measured on host cpu bound lookups | 19:05 |
| SpamapS | basically, python is 10x slower. We did not do much to try and debug it. | 19:05 |
| clarkb | SpamapS: real world deployments will be IO bound | 19:05 |
| SpamapS | actually I don't think it was CPU bound. | 19:05 |
| SpamapS | 90% of one core utilized by python | 19:05 |
| SpamapS | 76% by Rust | 19:05 |
| SpamapS | I think GC played a role. | 19:06 |
| SpamapS | but I'd have to go deep to confirm that, it's just a theory | 19:06 |
| clarkb | basically once you starting proxying you are talking many milliseconds between webserver and brwoser and the overhead of cpu lookup python vs c++ should be minimized and its that I think we would want to measure (however the startup time issue is also totally at hing) | 19:07 |
| SpamapS | either way, when you're talking about something that takes about 2 hours to port to basically any language and will run for every worker process on your server, using the strongly typed high performance language seems like a net positive. | 19:07 |
| SpamapS | I mean, in 72 hours I believe we had 3 ports from the original C++. | 19:08 |
| clarkb | sure | 19:09 |
| clarkb | but as a reviewer I don't want to review the Golang rewrite and the ruby rewrite all becquse its trivial and we don't have criteria for what is reasonable | 19:09 |
| SpamapS | I tend to think that the better result was achieved with minimal effort and a low expectation of ongoing feature creep, so, C++ or Rust seems to make a lot more sense than "spend time debugging python" | 19:10 |
| clarkb | I agree | 19:11 |
| clarkb | I'm just saying we should write down what "better" means here | 19:11 |
| SpamapS | And honestly, maybe some like Golang.. it's a really broken, trash language IMO. (sorry, had to get a shot in there). | 19:11 |
| clarkb | sounds like for you it is fast with strong typing | 19:11 |
| clarkb | (which is a fine better) | 19:11 |
| SpamapS | Fast with memory safety actualy. | 19:11 |
| SpamapS | The strong typing is just a bonus. :) | 19:12 |
| SpamapS | ok, I see, {build.uuid} is what I needed | 19:17 |
| SpamapS | I wonder if we should make Attributes return a default attribute for __str__ | 19:17 |
| SpamapS | clearly if I say {build}, I mean {build.uuid} | 19:18 |
| SpamapS | or maybe have a schema for the URL components and explode so the noop job doesn't pass. | 19:18 |
| clarkb | SpamapS: in the past I've through we should only expose a limited number of attributes to that url builder. Will make it easier to document and for users to get it right | 19:19 |
| clarkb | add a schema to that and we can provide nice errors too | 19:19 |
| clarkb | rather than expose the entire build object for example have build_uuid and job_name and whatever. I think the problem there is when you realize you ahven't exposed something someone needs | 19:20 |
| *** rfolco|rover has quit IRC | 19:21 | |
| *** mrhillsman_lunch is now known as mrhillsman | 19:29 | |
| *** dmellado has quit IRC | 19:44 | |
| *** dmellado has joined #zuul | 19:45 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Return npm site location as artifact https://review.openstack.org/640498 | 19:45 |
| SpamapS | clarkb: the entire build object is not exposed to be clear, it's only the safe attributes. But w/o attribute, it's the result of str(build) | 19:45 |
| SpamapS | Also I just tried setting success-url and failure-url to a build URL, and it overrode the log_url on the build page! | 19:46 |
| * SpamapS gives up trying to make the failure comments more useful | 19:48 | |
| *** irclogbot_3 has quit IRC | 19:48 | |
| clarkb | SpamapS: log_url is set by the job | 19:49 |
| clarkb | (which is confusing yes) | 19:49 |
| clarkb | but that gives the job the ability to say my logs are over here and you to set the success-url to say the build page | 19:49 |
| *** hashar has joined #zuul | 19:56 | |
| *** rlandy is now known as rlandy|brb | 19:57 | |
| clarkb | SpamapS: in the rust implementation why is hostname converted to string from str? split seems valid on str as well as string and we don't appear to modify hostname. | 20:01 |
| clarkb | (this is me being a rust noob after not looking at rust since it had like 4 different pointer types) | 20:01 |
| *** irclogbot_3 has joined #zuul | 20:03 | |
| *** hashar has quit IRC | 20:06 | |
| *** hashar has joined #zuul | 20:07 | |
| mordred | clarkb: (you should go check out the guide explaining 2018 edition - there's some super cool stuff in there) | 20:12 |
| mordred | clarkb: https://doc.rust-lang.org/nightly/edition-guide/rust-2018/index.html | 20:12 |
| mordred | (just in your off time) | 20:12 |
| clarkb | ya I'm noticing a lot of simplifications by way of syntactic sugar that I'm not sure were around when I first looked at it | 20:13 |
| clarkb | things like ? suffix operator | 20:13 |
| mordred | yeah. that's one of them | 20:13 |
| mordred | clarkb: https://doc.rust-lang.org/nightly/edition-guide/rust-2018/error-handling-and-panics/the-question-mark-operator-for-easier-error-handling.html | 20:14 |
| mordred | clarkb: they did a bunch of really good updates | 20:14 |
| *** EvilienM is now known as EmilienM | 20:14 | |
| mordred | clarkb: I wrote https://review.openstack.org/#/c/636796/ based on reading the slice pattern chapter in that doc in fact :) | 20:15 |
| *** sdake has joined #zuul | 20:18 | |
| *** sdake has quit IRC | 20:18 | |
| clarkb | mordred: The pointer simplification was a much needed change. The old docs for pointers literally said try this one first its most common, then try the other is it fails to compile TODO actually write down what these different pointers are for | 20:19 |
| mordred | ++ | 20:20 |
| *** sdake has joined #zuul | 20:22 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Return npm site location as artifact https://review.openstack.org/640498 | 20:33 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-preview master: Use artifact name rather than log_url https://review.openstack.org/640509 | 20:35 |
| *** rfolco has joined #zuul | 20:37 | |
| corvus | clarkb, mordred, SpamapS: i think the thread is at: several of us are keen on poking at rust. Shrews was interested in c++ but then went on PTO for a bit. i wanted to get his thoughts on rust before merging the switch | 20:37 |
| corvus | clarkb, mordred, SpamapS: but also, before we switch, can we land this change to remove the TODO: https://review.openstack.org/640509 | 20:38 |
| corvus | that'll let me finish the initial rollout | 20:38 |
| Shrews | so, rust seems like the new cool kid in school. but much like actual school, i don't know the cool kid. but i have no objection to hanging out with him | 20:39 |
| Shrews | corvus: i agree there is enough interest in rust (possibly more than c++?) to use that as a reason for using that version | 20:40 |
| clarkb | corvus: there is a bug in that change, one sec | 20:41 |
| clarkb | posted | 20:41 |
| *** sdake has quit IRC | 20:42 | |
| corvus | clarkb: thanks :) | 20:42 |
| corvus | i did not catch that in my testing since they are currently the same :| | 20:42 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-preview master: Use artifact name rather than log_url https://review.openstack.org/640509 | 20:43 |
| corvus | Shrews: cool, then i think let's switch next week | 20:44 |
| clarkb | we'll need to update the rust version to do the artifact_url lookup as in ^ | 20:44 |
| corvus | yep. it'll be fun! | 20:45 |
| corvus | Shrews or i should do that :) | 20:45 |
| Shrews | i can take a look at it | 20:46 |
| corvus | i just received my copy of https://www.worldcat.org/title/effective-modern-c-42-specific-ways-to-improve-your-use-of-c11-and-c14/oclc/1026465821 via inter-library loan | 20:46 |
| corvus | just in time :) | 20:46 |
| Shrews | unless rust decides to give me a wedgie | 20:46 |
| clarkb | Shrews: corvus you'll want to base it on https://review.openstack.org/#/c/636796/2/src/main.rs which changes how artifact is retrieved from parts | 20:49 |
| *** rlandy|brb is now known as rlandy | 20:51 | |
| clarkb | re my earlier question about to_string the reason for that is parts[x] goes out of scope at the end of the loop but hostname must live in the cache beyond that | 21:06 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Fix typo in docker push role https://review.openstack.org/640515 | 21:06 |
| *** jamesmcarthur has joined #zuul | 21:07 | |
| *** jamesmcarthur has quit IRC | 21:09 | |
| *** jamesmcarthur has joined #zuul | 21:10 | |
| *** jamesmcarthur has quit IRC | 21:14 | |
| corvus | 2 docker things are broken right now -- the docker build/publish/promote jobs are broken due to https://review.openstack.org/640515 which is preventing zuul changes from landing | 21:15 |
| corvus | and something has gone wrong with the new buildset registry jobs which is preventing zuul-preview changes from landing. i'm hoping https://review.openstack.org/638794 will tell more | 21:16 |
| clarkb | and to_string heap allocates a new String object which can live beyond the scope of that loop. And now I've learned myself a rust for the day | 21:16 |
| *** jamesmcarthur has joined #zuul | 21:29 | |
| *** irclogbot_3 has quit IRC | 21:37 | |
| openstackgerrit | Merged openstack-infra/zuul-jobs master: Fix typo in docker push role https://review.openstack.org/640515 | 21:40 |
| *** tobiash has quit IRC | 21:49 | |
| *** tobiash has joined #zuul | 21:51 | |
| *** josefwells_ has joined #zuul | 21:54 | |
| *** jamesmcarthur has quit IRC | 22:07 | |
| *** jamesmcarthur has joined #zuul | 22:08 | |
| *** rlandy has quit IRC | 22:10 | |
| openstackgerrit | Josef Wells proposed openstack-infra/zuul master: quickstart: web and others wait on mysql to start https://review.openstack.org/640548 | 22:22 |
| josefwells_ | yay, I did a thing | 22:25 |
| *** jamesmcarthur has quit IRC | 22:40 | |
| *** jamesmcarthur has joined #zuul | 22:44 | |
| *** odyssey4me_ has joined #zuul | 22:45 | |
| *** jbryce_ has joined #zuul | 22:45 | |
| *** jamesmcarthur has quit IRC | 22:48 | |
| *** odyssey4me has quit IRC | 22:52 | |
| *** jbryce has quit IRC | 22:52 | |
| *** odyssey4me_ is now known as odyssey4me | 22:52 | |
| *** jbryce_ is now known as jbryce | 22:52 | |
| *** hashar has quit IRC | 23:03 | |
| *** jesusaur has quit IRC | 23:03 | |
| *** frickler has quit IRC | 23:06 | |
| *** frickler has joined #zuul | 23:06 | |
| *** jesusaur has joined #zuul | 23:08 | |
| *** jesusaur has quit IRC | 23:14 | |
| *** jesusaur has joined #zuul | 23:23 | |
| *** josefwells_ has quit IRC | 23:31 | |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Fix buildset registry https://review.openstack.org/640557 | 23:56 |
| openstackgerrit | James E. Blair proposed openstack-infra/zuul-jobs master: Fix docker image retag https://review.openstack.org/640558 | 23:56 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!