| *** rfolco has joined #zuul | 00:01 | |
| *** jamesmcarthur has joined #zuul | 00:27 | |
| *** rfolco has quit IRC | 00:44 | |
| *** jamesmcarthur has quit IRC | 00:45 | |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Don't search github api for projects not in Zuul https://review.opendev.org/694390 | 00:45 |
|---|---|---|
| clarkb | pabelanger: ^ I think that should cover tobiash and my concern | 00:45 |
| *** jamesmcarthur has joined #zuul | 00:53 | |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Look for depends-on lines in github dependency search https://review.opendev.org/697595 | 01:00 |
| clarkb | pabelanger: tobiash ^ I think that implements the other thing I mentioned | 01:00 |
| *** rlandy|afk has quit IRC | 01:03 | |
| *** jamesmcarthur has quit IRC | 01:04 | |
| clarkb | someone that groks github better than me will need to clarify if the old behavior was intentional | 01:04 |
| *** jamesmcarthur has joined #zuul | 01:05 | |
| *** jamesmcarthur has quit IRC | 01:10 | |
| *** rfolco has joined #zuul | 01:13 | |
| *** jamesmcarthur has joined #zuul | 01:14 | |
| *** jamesmcarthur has quit IRC | 01:15 | |
| *** jamesmcarthur has joined #zuul | 01:15 | |
| *** michael-beaver has quit IRC | 01:19 | |
| *** rfolco has quit IRC | 01:33 | |
| *** rfolco has joined #zuul | 01:33 | |
| *** jamesmcarthur has quit IRC | 01:35 | |
| *** rfolco has quit IRC | 01:38 | |
| *** jamesmcarthur has joined #zuul | 01:39 | |
| ianw | ohhh, we've (i've) forgotten to pass in the ZUUL_SIBLIGNS build arg via the dockerfile | 01:43 |
| *** jamesmcarthur has quit IRC | 01:44 | |
| *** jamesmcarthur has joined #zuul | 02:02 | |
| *** jamesmcarthur has quit IRC | 02:13 | |
| *** jamesmcarthur has joined #zuul | 02:13 | |
| openstackgerrit | Ian Wienand proposed zuul/nodepool master: Also build sibling container images https://review.opendev.org/697393 | 02:25 |
| openstackgerrit | Ian Wienand proposed zuul/nodepool master: [wip] move openstack testing to use containerised daemon https://review.opendev.org/693464 | 02:25 |
| *** bhavikdbavishi has joined #zuul | 03:11 | |
| *** bhavikdbavishi1 has joined #zuul | 03:14 | |
| *** bhavikdbavishi has quit IRC | 03:16 | |
| *** bhavikdbavishi1 is now known as bhavikdbavishi | 03:16 | |
| *** jamesmcarthur has quit IRC | 03:18 | |
| openstackgerrit | Ian Wienand proposed zuul/zuul-jobs master: build-docker-image: fix up siblings copy https://review.opendev.org/697614 | 03:34 |
| *** openstackstatus has joined #zuul | 03:42 | |
| *** ChanServ sets mode: +v openstackstatus | 03:42 | |
| openstackgerrit | Ian Wienand proposed zuul/zuul-jobs master: build-docker-image: fix up siblings copy https://review.opendev.org/697614 | 04:49 |
| *** jamesmcarthur has joined #zuul | 05:20 | |
| *** raukadah is now known as chkumar|ruck | 05:20 | |
| *** jamesmcarthur has quit IRC | 05:25 | |
| openstackgerrit | Merged zuul/zuul master: Log item warning messages at info level https://review.opendev.org/697571 | 05:54 |
| openstackgerrit | Merged zuul/zuul master: Fix potential wedge with provides/requires/dependencies https://review.opendev.org/697579 | 06:01 |
| *** swest has joined #zuul | 06:10 | |
| openstackgerrit | Merged zuul/zuul master: Pagure - add the enqueue_ref unit test https://review.opendev.org/687351 | 06:17 |
| *** hashar has joined #zuul | 07:54 | |
| *** jangutter has joined #zuul | 08:00 | |
| *** pcaruana has joined #zuul | 08:11 | |
| *** jcapitao has joined #zuul | 08:11 | |
| *** saneax has joined #zuul | 08:31 | |
| *** avass has joined #zuul | 08:35 | |
| *** hashar has quit IRC | 09:05 | |
| openstackgerrit | Albin Vass proposed zuul/nodepool master: Documentation fixes https://review.opendev.org/697635 | 09:06 |
| *** hashar has joined #zuul | 09:33 | |
| *** bhavikdbavishi has quit IRC | 10:21 | |
| *** reiterative has joined #zuul | 10:21 | |
| *** rfolco has joined #zuul | 10:28 | |
| *** sshnaidm|afk is now known as sshnaidm|off | 11:28 | |
| sugaar | corvus going back to what we where discussing yesterday about the node name, I suppose that nodepool gets feed from some independent "element" that provides the nodes that nodepool gives to zuul. In the case of kubernetes, a node would be the server given by a could provider, so when in the nodepool/zuul config we are referencing nodes are we | 11:34 |
| sugaar | actually referencing the name of the actual kubernetes node, the kubernetes cluster or the pod? because I think that my problem is in there | 11:34 |
| *** zbr has quit IRC | 11:47 | |
| *** zbr has joined #zuul | 11:47 | |
| *** tosky has joined #zuul | 11:50 | |
| openstackgerrit | Albin Vass proposed zuul/nodepool master: Documentation fixes https://review.opendev.org/697635 | 11:56 |
| openstackgerrit | Ian Wienand proposed zuul/zuul-jobs master: build-docker-image: fix up siblings copy https://review.opendev.org/697614 | 12:02 |
| *** rfolco is now known as rfolco|bbl | 12:03 | |
| *** jcapitao is now known as jcapitao|lunch | 12:05 | |
| *** saneax has quit IRC | 12:43 | |
| *** rlandy has joined #zuul | 13:03 | |
| *** jamesmcarthur has joined #zuul | 13:06 | |
| *** jcapitao|lunch is now known as jcapitao | 13:07 | |
| *** hashar has quit IRC | 13:07 | |
| *** jamesmcarthur has quit IRC | 13:28 | |
| *** jamesmcarthur has joined #zuul | 13:29 | |
| *** jamesmcarthur has quit IRC | 13:34 | |
| *** mgoddard has quit IRC | 13:44 | |
| *** mgoddard has joined #zuul | 13:45 | |
| *** armstrongs has joined #zuul | 13:46 | |
| *** jamesmcarthur has joined #zuul | 13:46 | |
| *** mgoddard has quit IRC | 13:58 | |
| *** mgoddard has joined #zuul | 14:05 | |
| *** hashar has joined #zuul | 14:22 | |
| *** jamesmcarthur has quit IRC | 14:29 | |
| *** jamesmcarthur has joined #zuul | 14:30 | |
| *** jamesmcarthur has quit IRC | 14:35 | |
| *** Goneri has joined #zuul | 14:38 | |
| fungi | sugaar: with other nodepool drivers the nodepool launcher contacts an api for a service which provides compute resources (openstack, aws, azure, whatever) and then hands those off to zuul to fill a node request as soon as they're available | 14:42 |
| fungi | aside from the static node driver of course, which just expects you to have already built the nodes and listed them in nodepool yourself | 14:42 |
| fungi | i don't know enough about the kubernetes and openshift drivers to be sure, but i expect they get containers the same way the other cloudy node drivers get servers | 14:43 |
| fungi | i'll read the docs | 14:43 |
| fungi | ahh, the nodepool launcher talks to a cluster manager and gets kubernetes pods to hand over to zuul for corresponding node requests, if i'm reading it correctly | 14:45 |
| sugaar | can you point me to that doc? | 14:47 |
| fungi | sugaar: https://zuul-ci.org/docs/nodepool/configuration.html#kubernetes-driver | 14:48 |
| fungi | so looks like you would precreate a kubernetes cluster, provide admin credentials for that to the nodepool launcher service, and then in your jobs you would specify nodes or nodesets which referred to labels for pods you want nodepool to provide | 14:50 |
| mordred | fungi, sugaar: there are 2 different options - you can either get a pod from k8s or you can get a namespace from k8s | 14:50 |
| mordred | but yes - what fungi said | 14:50 |
| fungi | yeah, i see the namespace option in there too. documentation seems to suggest that namespace is a subset of the resources provided by pod? i.e. when you request a pod you get a namespace with a pod in it | 14:51 |
| fungi | but when you request a namespace you just get a namespace with no pod in it | 14:51 |
| fungi | i suppose the latter would be for when you have a job which plans to create its own pods in the namespace? | 14:51 |
| * fungi is still very fuzzy on container ecosystem terminology | 14:52 | |
| *** hashar has quit IRC | 14:53 | |
| *** jamesmcarthur has joined #zuul | 15:00 | |
| *** chkumar|ruck is now known as raukadah | 15:01 | |
| mordred | fungi: a k8s namespace is actually more a subset of a full k8s in which you create pods | 15:02 |
| fungi | that's roughly how i understood it | 15:03 |
| mordred | fungi: so if what you want for your jobs isn't some containers inside of which you want to run content, but instead a k8s api against which you can run kubectl to do things - you would request a namespace and then ... ah - yes, I read your words better :) | 15:03 |
| fungi | well, i'm still only just starting to wrap my head around it anyway | 15:04 |
| mordred | fungi: of course, what we've found so far with our k8s workloads is that requesting some vms and installing k8s on them, then creating pods in that k8s is actually nicer than requesting a namspace :) | 15:04 |
| mordred | which is why I think growing nodepool drivers similar to the namespace driver but that use magnum/gke/eks to create a full k8s for you would be a positive thing | 15:05 |
| fungi | seems like it, yeah | 15:05 |
| zbr | can i get some hints regarding starting zuul using compose file? somehow that compose file does not work for me because it needs a network=host in order to build the containres | 15:07 |
| zbr | the default network mode on docker does not allow network access and I prefer to avoid altering default setup. | 15:07 |
| zbr | are we really not able to use dockerfile file formats newer than v2 ? | 15:07 |
| *** jamesmcarthur has quit IRC | 15:08 | |
| mordred | zbr: "the default network mode on docker does not allow network access" - I have not found that to be true - I can run that docker compose on any unaltered docker-compose I've found - across ubuntu, centos and mac osx - I'm confused about the problem you are encountering | 15:11 |
| mordred | what sort of environment are you trying to run that in? | 15:11 |
| mordred | (also - some of those containers are just for the quick-start example - so you could just use the upstream published images for a docker-compose file that didn't also build the node and logs container images) | 15:12 |
| zbr | mordred: centos 8 + docker-ce (upstream version) and running compose command from a remove machine (and using DOCKER_HOST=ssh://. for that) | 15:12 |
| mordred | ah - I have never attempted to run this using docker on a remote machine | 15:13 |
| zbr | i can try to narrow it down but I this happened to me multiple times, unless I run docker *build with | 15:13 |
| zbr | with --network-host it would fail to start. | 15:13 |
| zbr | i doubt this issue is related to remoting, but I can easily check directly there. | 15:14 |
| zbr | on the other hand, I do not exclude a local docker config issue. but is not the first time when I see this issue realted to --network=host | 15:15 |
| mordred | zbr: the two things you're doing that are new are centos8 and remoting | 15:15 |
| mordred | if you can verify that the issue isn't related to remoting, then I think we can start figuring out what's broken in centos8 | 15:15 |
| zbr | in fact, let me disable remoting and use local (macos) | 15:15 |
| mordred | zbr: I am 100% certain it works on docker for mac | 15:16 |
| zbr | i also happen to have a full dual stack at home, and apparently this was permanent source of problems | 15:16 |
| *** jangutter has quit IRC | 15:19 | |
| Shrews | i still find it frightening that mordred admits to any mac usage | 15:21 |
| Shrews | :) | 15:21 |
| *** jamesmcarthur has joined #zuul | 15:22 | |
| zbr | Shrews: believe me that, every time i have to confess that I expect some kind of retaliation. | 15:22 |
| fungi | i still find it sad that having working ipv6 means docker falls apart | 15:22 |
| zbr | mordred: and you were right, using mac-docker worked fine. now time to narrow down the issue, testing a remote on ubuntu. | 15:23 |
| zbr | there is still a question here: why we do not publish pre-build containers for zuul, so user can run them without needing any post pull operations. | 15:24 |
| zbr | because pulling works fine, the problem is related to building phase. | 15:25 |
| zbr | i know for sure that on molecule I had to add a special config items to force docker to use host network during build to avoid errors installing packages (unrelated to zuul) | 15:25 |
| zbr | yep, remoting to ubuntu+docker worked fine too, so now i time to compare the setup. | 15:31 |
| *** jamesmcarthur has quit IRC | 15:31 | |
| *** jamesmcarthur has joined #zuul | 15:32 | |
| mordred | zbr: we do publish pre-built containers for zuul | 15:32 |
| mordred | zbr: for this quickstart - there are two utility images we make. they are quick-start specific | 15:33 |
| zbr | errata: did not really worked, got another issue | 15:33 |
| mordred | we could make a more pared down compose file that only ran zuul - it would be harder to use as a self-contained quick-start example tutorial-style (becaus you'd need to set up nodepool sources and stuff) | 15:33 |
| mordred | but I'm curious to learn what you find out is the issue here | 15:34 |
| mordred | Shrews: yeah. I find it frightening too | 15:36 |
| sugaar | fungi mordred thanks for the info, I will digest all that and let's see if I can clear my mind | 15:43 |
| *** jamesmcarthur has quit IRC | 15:46 | |
| openstackgerrit | David Moreau Simard proposed zuul/zuul master: DNM: Test zuul-stream-functional with ara 1.3 https://review.opendev.org/694622 | 15:47 |
| zbr | yep, remoting failed me, while same command worked fine when run locally http://paste.openstack.org/show/787249/ | 15:47 |
| mordred | zbr: image building seemed to work ok - it seems like it's the connections between containers that is unhappy, yeah? | 15:49 |
| zbr | already found the bug, almost sure is https://github.com/docker/compose/issues/6463 -- reading it now. clearly the problem was different than the one with centos8 | 15:50 |
| zbr | i am sure that by the time I sort this I will forget why I wanted to start local zuul. | 15:50 |
| openstackgerrit | Fabien Boucher proposed zuul/zuul-jobs master: Make ara-report role to zuul_return an artifact https://review.opendev.org/697681 | 15:50 |
| zbr | learned a new trick today: set ssh MaxSessions to 500 on any docker server deployment if you plan to do ssh remoting to it. | 16:00 |
| *** jcapitao is now known as jcapitao|afk | 16:02 | |
| mordred | zbr: ah fun! | 16:04 |
| *** rfolco|bbl is now known as rfolco | 16:12 | |
| *** sgw has quit IRC | 16:36 | |
| *** rlandy is now known as rlandy|ruck | 16:44 | |
| SpamapS | Has anyone suggested changing the log viewer in the zuul UI to use xterm.js just like the streamer? | 16:49 |
| SpamapS | I have a lot of color codes in my log files, they look great in streaming.. they pollute the txt tho | 16:49 |
| zbr | regarding ipv6 inside docker containers, does anyone know a good guide (other than adding the --ipv6 option)? ... or i am better off not attempting to make it work? | 16:49 |
| mordred | corvus: your stack at https://review.opendev.org/#/c/695051/ has 3x+2 on it - I didn't +A in case you wanted to be around for that | 16:51 |
| *** sgw has joined #zuul | 16:51 | |
| mordred | SpamapS: I'm not aware of any such suggestion - by log viewer, do you mean the console log: https://zuul.opendev.org/t/zuul/build/fe0fd72c74e346db98dedb03cd2f23e9/console ? or the 'raw' log viewer: https://zuul.opendev.org/t/zuul/build/fe0fd72c74e346db98dedb03cd2f23e9/log/job-output.txt ? | 16:52 |
| mordred | zbr: I have no useful suggestions. :) In our production things we run in containers we run them all with network=host and it all just works - I don't know much useful about using docker with network namespacing (not a thing that's super valuable in our world) | 16:54 |
| zbr | mordred: that is good info anyway, mainly i do the same in my homelab, all containers are running with network=host, which is a PITA as I need to assure ports do not overlap. | 16:58 |
| *** jamesmcarthur has joined #zuul | 16:59 | |
| corvus | mordred: thanks, approved | 17:00 |
| corvus | clarkb: small note on 697548 | 17:03 |
| clarkb | corvus: can do, thanks | 17:03 |
| *** hashar has joined #zuul | 17:09 | |
| corvus | oh wow -- the last openshift patch i threw up before switching to podman actually worked | 17:10 |
| corvus | https://zuul.opendev.org/t/zuul/build/6966f52859f14477a77ece8f1d769951/console#1/0/47/controller | 17:11 |
| clarkb | corvus: maybe review https://review.opendev.org/#/c/675403/ too? (another ease of use change but this one for operations) | 17:12 |
| *** jamesmcarthur has quit IRC | 17:14 | |
| openstackgerrit | Merged zuul/zuul-jobs master: use-buildset-registry: Vendor pytoml and remarshal https://review.opendev.org/695050 | 17:15 |
| zbr | btw, what was the magic string to be added to populate the status filter. I remember that adding #123 to the url used to work but I checked now and apparently it does not. | 17:15 |
| openstackgerrit | Merged zuul/zuul-jobs master: Add build-container-image role https://review.opendev.org/695251 | 17:15 |
| SpamapS | mordred: I mean https://zuul.opendev.org/t/zuul/build/fe0fd72c74e346db98dedb03cd2f23e9/log/job-output.txt | 17:15 |
| zbr | asking because I want to post a status url that auto-filters, intead of seeing the entire status dashboard which may be of not much use to someone just wanted to see what happens with his own change. | 17:16 |
| corvus | SpamapS: i'm totally open to supporting ansi codes there; the reason we're not just using xterm.js is the severity filtering/colorization and line number anchoring | 17:17 |
| corvus | zbr: click on the icon next to the item on the status change | 17:18 |
| corvus | zbr: rather, 'on the status screen' | 17:18 |
| *** jamesmcarthur has joined #zuul | 17:18 | |
| corvus | that should get you an example | 17:18 |
| corvus | SpamapS: my guess is that it'll be easier to just add some ansi parser to what we have now, but any solution that has filtering+anchoring+color wfm :) | 17:21 |
| zbr | corvus: so what should I append to https://github.com/ansible/project-config/blob/master/zuul.d/pipelines.yaml#L22 in order to enable the filtering? | 17:21 |
| *** hashar has quit IRC | 17:22 | |
| *** hashar_ has joined #zuul | 17:23 | |
| zbr | SpamapS: are you really working on bringing ANSI to our console logs? that would be the best UX we could add to zuul right now. I am more than happy to help here because I wanted to do it myself but never found the time. | 17:23 |
| *** pcaruana has quit IRC | 17:23 | |
| zbr | for me color is *very* important, my eyes are not good at spotting errors in endless B/W files, in fact that is why I am using https://opendev.org/x/coats/src/branch/master/coats/FoxReplace.json to bring back some of the lost colors. | 17:24 |
| *** jamesmcarthur has quit IRC | 17:25 | |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Record job build attempts in inventory https://review.opendev.org/697548 | 17:26 |
| *** jamesmcarthur has joined #zuul | 17:26 | |
| corvus | zbr: please feel free to add those to https://opendev.org/zuul/zuul/src/branch/master/web/src/actions/logfile.js so that you don't have to use that anymore and others can befenit from it | 17:27 |
| openstackgerrit | Merged zuul/zuul-jobs master: use-buildset-registry: Add podman support https://review.opendev.org/695051 | 17:28 |
| *** jamesmcarthur has quit IRC | 17:28 | |
| *** rlandy|ruck is now known as rlandy|ruck|brb | 17:29 | |
| *** jamesmcarthur has joined #zuul | 17:29 | |
| corvus | zbr: re status, try something like "/change/${change.number},${change.patchset}" | 17:30 |
| zbr | i hope it will work with github integration too | 17:30 |
| clarkb | what I find helps me most is ability to grep. I think xterm makes that weird so if we do switch to xterm.js we should keep the use case of being able to ^F in mind | 17:30 |
| clarkb | the current renderer works fine with ^F though | 17:31 |
| zbr | in fact that is why i was trying to setup my private zuul, so I can fully experiment with these before proposing changes. | 17:31 |
| corvus | clarkb: agreed. so color+anchor+severity+grep are the reqs :) | 17:31 |
| corvus | zbr: just fyi, you can run the js locally against a public zuul. that's helpful for some kinds of changes. for others, it's helpful to have a private zuul. | 17:32 |
| corvus | (but if i were just looking at things like colorizing logs, i'd run js locally against opendev's zuul -- it has lots of logs already) | 17:32 |
| corvus | https://zuul-ci.org/docs/zuul/developer/javascript.html has the instructions for that | 17:33 |
| *** hashar_ has quit IRC | 17:33 | |
| *** hashar has joined #zuul | 17:34 | |
| SpamapS | corvus: thanks that makes sense. I think we could get that with xterm.js, but there's a nice React ANSI thing that might be better for this more static rendering. | 17:35 |
| SpamapS | zbr: I'm mostly just dreaming. | 17:36 |
| corvus | dreaming is more or less how i start all dev efforts :) | 17:36 |
| mordred | my favorite way to code is to dream loudly enough that clarkb writes the code | 17:37 |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: WIP: openshift speculative containers https://review.opendev.org/696939 | 17:37 |
| zbr | is it true that the ability to click the result column items in builset column was added to zuul after 3.10.2 ? i did not see a note in release notes but I see it working on our zuul but on rdo onw is not clickable. | 17:39 |
| corvus | zbr: i don't know; we usually only add release notes for bigger changes | 17:40 |
| corvus | (in order to keep them short enough that people will want to read them for the things we think everyone should know about) | 17:41 |
| zbr | corvus: just sent them a msg asking about when they will update. | 17:41 |
| *** rlandy|ruck|brb is now known as rlandy|ruck | 17:54 | |
| *** jamesmcarthur has quit IRC | 18:02 | |
| *** jamesmcarthur has joined #zuul | 18:02 | |
| *** jamesmcarthur has quit IRC | 18:17 | |
| openstackgerrit | Merged zuul/zuul master: Improve functionality and docs around ansible installation https://review.opendev.org/675403 | 18:19 |
| *** armstrongs has quit IRC | 18:41 | |
| *** notnone has quit IRC | 18:55 | |
| *** notnone has joined #zuul | 18:55 | |
| *** hashar has quit IRC | 19:03 | |
| *** tosky has quit IRC | 19:18 | |
| clarkb | I think I understand why we do a greedy search in github now for depends on | 19:24 |
| clarkb | it is because we do a greedy search in gerrit then filter the results in zuul | 19:24 |
| clarkb | github is just mimicing what we do for gerrit, except that gerrit is much more forgiving about having many many results and queries | 19:26 |
| clarkb | what we seem to do pretty universally is use code review tool search to find all cases of a "uri". The uri actually lacks the protocol prefix. Then we do local filtering to determine if that represents a valid depends on or not | 19:37 |
| openstackgerrit | Clark Boylan proposed zuul/zuul master: Look for depends-on lines in dependency searches https://review.opendev.org/697595 | 20:07 |
| clarkb | corvus: tobiash ^ that change feels dangerous but I also think it is more correct behavior if done right? | 20:08 |
| clarkb | at this point I think it is mostly a "hey we noticed this thing and here are some steps to update it to be friendlier to our code review servers" and input would be appreciated | 20:08 |
| clarkb | My hunch is that I'm missing some corner case where the query languages of these tools isn't rich enough to describe valid depends-on cases that we want to catch | 20:09 |
| clarkb | and maybe we leave some breadcrumbs to that in the code so we don't try to make this update in the future? | 20:09 |
| *** jcapitao|afk has quit IRC | 20:16 | |
| corvus | clarkb: can you verify that works in gerrit? there was some weird behavior with gerrit's search around that | 20:42 |
| corvus | it's probably okay 'cause we're just adding terms -- i want to say the thing i'm remembering is that searching for "depends-on" wasn't enough on its own (it would return results with the word "on"). | 20:44 |
| corvus | but i'm not sure | 20:44 |
| mordred | corvus: we should really get gerrit to add a depends-on parser and indexer :) | 20:50 |
| clarkb | corvus: ah I did test it generally works and is case insensitive but didnt check for extra results | 20:54 |
| *** jamesmcarthur has joined #zuul | 20:59 | |
| *** jamesmcarthur has quit IRC | 21:04 | |
| mordred | SpamapS: aws related docs patch: https://review.opendev.org/#/c/697635 - pretty small, should be an easy review | 21:05 |
| *** rlandy|ruck has quit IRC | 21:11 | |
| clarkb | corvus: https://review.opendev.org/#/c/589769/ has an 'on' in the commit message but message:{depends-on} does not match it | 21:18 |
| *** jamesmcarthur has joined #zuul | 21:30 | |
| *** rfolco has quit IRC | 21:33 | |
| corvus | clarkb: oh, it may have been extra "depends" | 21:35 |
| corvus | maybe on is a stop word, so we just got lots of "depends" changes if we searched for "depends-on". | 21:36 |
| *** jamesmcarthur has quit IRC | 21:36 | |
| openstackgerrit | Clark Boylan proposed zuul/zuul-jobs master: DO NOT MERGE test cleanup phase playbook https://review.opendev.org/680178 | 21:37 |
| corvus | clarkb: i think the main question to answer is, if you search for "depends-on: url" is it a subset of searching for "url" or are we going to get extra results | 21:37 |
| corvus | as long as it's a subset, then mission accomplished :) | 21:37 |
| clarkb | corvus: I think the quotes are the important bit because that means the matched string includes the whitespace? but I haven't confirmed that yet | 21:38 |
| corvus | clarkb: well, we probably want it to work for "depends-on: uri" too | 21:38 |
| clarkb | ok its something in between :/ | 21:38 |
| clarkb | corvus: see https://review.opendev.org/#/q/message:%257Bdepends+because%257D | 21:38 |
| clarkb | that matches my edited commit message with has 'depends because' in it but also mordreds 'depends.\n\nBecause' | 21:39 |
| corvus | https://review.opendev.org/#/q/message:%257Bbecause+depends%257D confirms that it's not just doing unordered token search | 21:40 |
| corvus | clarkb: that seems to suggest that we're getting the behavior we want from gerrit | 21:41 |
| corvus | this should reduce the number of results we get back, with no loss of results, even if we still have a few false positives | 21:42 |
| clarkb | corvus: ya, we still need the local side filter though | 21:42 |
| clarkb | yup | 21:42 |
| corvus | and your change leaves that in place, right? it basically just improves the input? | 21:43 |
| clarkb | yes | 21:43 |
| clarkb | I was thinking the next step would be to remove the local filter, but pretty sure that would be broken now | 21:43 |
| corvus | yeah. no harm leaving it there. | 21:43 |
| corvus | clarkb: lgtm; let's see if tobiash or pabelanger can take a look. probably next week. :) | 21:45 |
| *** irclogbot_1 has quit IRC | 21:45 | |
| openstackgerrit | Merged zuul/zuul master: Fix job result metric name for builds to retry https://review.opendev.org/697248 | 21:46 |
| clarkb | corvus: ya no rush. Was something that popped out at me reviewing pabelanger's related change | 21:46 |
| clarkb | I think part of the spidering and then rate limiting is we do a very greedy search :) | 21:47 |
| openstackgerrit | James E. Blair proposed zuul/zuul-jobs master: openshift speculative containers https://review.opendev.org/696939 | 21:50 |
| clarkb | corvus: https://review.opendev.org/#/c/694390/ is the related change if you want to take a look at that one while context is paged in | 21:53 |
| *** hashar has joined #zuul | 21:55 | |
| corvus | lgtm | 21:56 |
| tobiash | clarkb: did you also do some test queries fir github or just gerrit? | 22:00 |
| clarkb | tobiash: I did some queries with github, let me find them in browser history | 22:00 |
| openstackgerrit | Merged zuul/zuul master: Pagure: Fix handling of threshold_reached: None https://review.opendev.org/691254 | 22:00 |
| clarkb | tobiash: https://github.com/ansible-network/ansible_collections.cisco.nxos/pulls?utf8=%E2%9C%93&q=%22Depends-On%3A+https%3A%2F%2Fgithub.com%2Fansible-network%2Fansible_collections.cisco.nxos%2Fpull%2F36%22+is%3Apr+is%3Aopen+ is one example | 22:02 |
| tobiash | k, then lgtm | 22:02 |
| *** avass has quit IRC | 22:16 | |
| *** jamesmcarthur has joined #zuul | 22:33 | |
| *** tosky has joined #zuul | 22:35 | |
| *** jamesmcarthur has quit IRC | 22:38 | |
| openstackgerrit | Merged zuul/zuul master: Don't search github api for projects not in Zuul https://review.opendev.org/694390 | 22:48 |
| openstackgerrit | Merged zuul/zuul master: Look for depends-on lines in dependency searches https://review.opendev.org/697595 | 23:03 |
| SpamapS | ohhhhh | 23:20 |
| SpamapS | 694390 is a good one! :) | 23:20 |
| SpamapS | mordred: thanks! +3'd | 23:21 |
| *** jamesmcarthur has joined #zuul | 23:34 | |
| *** jamesmcarthur has quit IRC | 23:38 | |
| openstackgerrit | Merged zuul/nodepool master: Documentation fixes https://review.opendev.org/697635 | 23:57 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!