Monday, 2025-11-10

-@gerrit:opendev.org- Zuul merged on behalf of Simon Westphahl: [zuul/zuul] 966285: Fix OIDC secret config errors in zuul-web https://review.opendev.org/c/zuul/zuul/+/96628510:39
-@gerrit:opendev.org- Simon Westphahl proposed: [zuul/nodepool] 966526: Enable IPv6 endpoint when IMDS support enabled https://review.opendev.org/c/zuul/nodepool/+/96652612:22
-@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 966528: Enable IPv6 endpoint when IMDS support is enabled https://review.opendev.org/c/zuul/zuul/+/96652812:23
-@gerrit:opendev.org- Simon Westphahl proposed: [zuul/zuul] 966528: Enable IPv6 endpoint when IMDS support is enabled https://review.opendev.org/c/zuul/zuul/+/96652812:24
-@gerrit:opendev.org- Simon Westphahl proposed: [zuul/nodepool] 966526: Enable IPv6 endpoint when IMDS support is enabled https://review.opendev.org/c/zuul/nodepool/+/96652612:25
@mnaser:matrix.orga little bit first world problem, but does anyone notice/find the console text extremely small on a 4k display in firefox/linux?   rendering seems to be sortof trimmed15:13
@mnaser:matrix.orgwell.. it looks perfect in that screenshot, lol15:15
@mnaser:matrix.orgactually you can kinda see it a bit, things like the j is trimmed from the bottom15:16
@sean-k-mooney:matrix.orglooks like there is some type hinitign issues but that might just be a browser rednerign issues15:17
@sean-k-mooney:matrix.orgdo you have your browser scaling at 100%15:18
@mnaser:matrix.orgyeah, if i scale up, the text is bigger, but still bottom of chaarcters like p's and y's are trimmed off, but yeah, not implying a zuul issue15:18
@mnaser:matrix.orgi also wonder if missing font on my side15:20
@sean-k-mooney:matrix.orgjsut lookign at a live log now https://zuul.opendev.org/t/openstack/stream/70f17f36fe0a4260bc1fd4915bb1096e?logfile=console.log im not seeing it happen but i also use monospace fonts as my defatult fonts systemwid15:20
@sean-k-mooney:matrix.orgso i wonder is is font specific15:20
@mnaser:matrix.orginteresting, dont see the issue in chromium15:21
@mnaser:matrix.orgso might be firefox thing i guess15:21
@fungicide:matrix.orgmnaser: what i observe in my primary browser (qutebrowser, webkit based) is that if i try to increase the font size the text box isn't rescaled to fit into the window and runs off the bottom of the screen without any way for me to scroll down to the bottom of the embedded text area15:23
@sean-k-mooney:matrix.orgim using firefox but i see each span with "font-family: courier-new, courier, monospace; font-size: 12px"15:23
@mnaser:matrix.orgthat was my initial investigation, i did end up installing courier new (but i wonder if i havent rebooted since and it didnt pick it up in firefox)15:24
@mnaser:matrix.orgbrowwsers too complicated :(15:24
@sean-k-mooney:matrix.orgah if i go to 150% zoom i see the same clipiing issue15:25
@fungicide:matrix.orgyeah, i have some small-screen devices with very high resolution panels, and zoom ends up necessary for me on those15:28
@sean-k-mooney:matrix.orgah i see the issue15:28
@sean-k-mooney:matrix.org```<textarea class="xterm-helper-textarea" aria-label="Terminal input" aria-multiline="false" autocorrect="off" autocapitalize="none" spellcheck="false" tabindex="0" style="left: 533px; top: 880px; width: 20px; height: 20px; line-height: 15.339px; z-index: 1000;"></textarea>```15:28
@sean-k-mooney:matrix.orgwhen you zuum in the lin hight and hight of the text area for each line is nto adjusted15:28
@fungicide:matrix.orgaha, good catch15:28
@sean-k-mooney:matrix.orgso the fot gets bigger btut the line height stays the same and just pushes the bottom off the end15:29
@sean-k-mooney:matrix.orgno idea how to fix that because CSS is magic but ya thats the problem15:29
@sean-k-mooney:matrix.orgmaybe do min heihg tinstead of height15:29
@sean-k-mooney:matrix.orgwell the question i was going to ask has ansered itself. yes if i just wait long enough after updating zuuls tenant file the project will show up after the scdhuelr finishes setting them all up15:31
@jangutter:matrix.org^--- if you do smart-reconfigure or full-reconfigure.15:31
@fungicide:matrix.orgright, if you follow the scheduler log you should see it indicate when reconfiguration starts/completes15:32
@sean-k-mooney:matrix.orgi did neither i just updated the configmap vai flux and zuul noticed and did the smart-reconfigure itself15:32
@jangutter:matrix.orgThat's.... not supposed to happen? Usually something triggers that.15:33
@sean-k-mooney:matrix.orghttps://zuul.teim.app/t/main/projects i added basiclly all the openstack pyton project in one goo so it took maybe 5 minutes to caluate all the keys and other stuff zulul needed to do15:33
@sean-k-mooney:matrix.orgjangutter: zuul watches the tenent file15:33
@sean-k-mooney:matrix.orgi think using inotify or similar15:33
@jangutter:matrix.orgHang on.15:33
@sean-k-mooney:matrix.orgso if it changes on disk it will pick it up and retirger15:33
@jim:acmegating.comsean-k-mooney: are you using zuul-operator?15:34
@sean-k-mooney:matrix.orgyes15:34
@sean-k-mooney:matrix.orgwith a patch or two but yes15:34
@fungicide:matrix.orgdeployment of the file is probably coupled with some automation to send a signal to the scheduler daemon15:34
@sean-k-mooney:matrix.orgim using fluxcd to manage the zuul operator15:34
@jim:acmegating.comsean-k-mooney: the operator did the smart-reconfigure for you15:34
@jangutter:matrix.orghttps://zuul-ci.org/docs/zuul/latest/releasenotes.html#relnotes-11-3-0-upgrade-notes <---15:35
@sean-k-mooney:matrix.orgi just pushed https://github.com/SeanMooney/zuul-homelab/commit/e04bb9996e86a70659923f6801b496d370bcd8f1 and trigered the flux reconsole to have it be appied15:35
@jangutter:matrix.orgin the past you could trigger it by restarting the scheduler, but that's no longer a thing anymore.15:35
@sean-k-mooney:matrix.orgya so that what i used to do hwen i deployed it under systemd15:36
@sean-k-mooney:matrix.orgbut i tought zuul got the ablit to trigger this iself at somepoint in the past15:36
@jangutter:matrix.orgI don't think it ever had.15:36
@jangutter:matrix.org(nodepool had, but zuul tenant reconfig has always been manual in my patchy memory)15:37
@sean-k-mooney:matrix.organyway im likeing the work flow of jsut pushing the commit and waiting 10 mintues for flxu to pick it up and apply it. (or being inpatiant and telling it to do it now)15:37
@sean-k-mooney:matrix.orgah ok ya it may have been nodepool i was thinking of15:37
@fungicide:matrix.orgto be clear, we have something similar in opendev, though triggered through ansible when we deploy a new version of that file15:38
@jangutter:matrix.orgYeah, external automation triggering the event, but not something internal to zuul-scheduler.15:39
@fungicide:matrix.orgin opendev we do it with a handler: https://opendev.org/opendev/system-config/src/branch/master/playbooks/roles/zuul-scheduler/handlers/main.yaml15:39
@sean-k-mooney:matrix.orgya when iu manually deployed zuul with ansible in the past i used a notifyer and handeler to triger the restart on chagnes to the config file15:39
@sean-k-mooney:matrix.orgso i have alwasy been used to this being automated15:39
@jangutter:matrix.orgI still like the sheer number of people who co-authored this: https://opendev.org/zuul/zuul-operator/commit/a9858a5b77b7c2b2219950bc457de1d79d3cd7b515:40
@jangutter:matrix.org^--- but that's the bit that makes zuul-operator uptake changes in the tenant config.,15:40
@sean-k-mooney:matrix.orgack ya that changed since i previously deployed zuul15:41
@sean-k-mooney:matrix.orgso since i mentioned i had to fork the zuul operator imiadaly to be able to use it15:41
@sean-k-mooney:matrix.orghwo shoudl i file bugs and upstream some of those changes15:41
@fungicide:matrix.orgfiling bugs is probably optional if you are going to push up fixes anyway, so long as you explain in the commit message the reason it's needed15:42
@sean-k-mooney:matrix.orgwell the first fix is trivial https://github.com/SeanMooney/zuul-operator/commit/d2dc4be75a8402c8159764dc24423487add72fe4 15:43
@sean-k-mooney:matrix.orgif i dont set the file mode to 400 or similar on the executor key it was not trusted for me adn woudl not work15:44
@sean-k-mooney:matrix.orgthe other change is propably not the correct way to fix it but it what i need to make it work https://github.com/SeanMooney/zuul-operator/commit/89ce1ae82de739c25155c3262578b841efd975eb15:44
@sean-k-mooney:matrix.orgim deploying on talos and to allow bubblewrap to be able to create its onw user namespaces 15:45
@sean-k-mooney:matrix.orgi need to not set the runas runAsUser and runAsGroup15:45
@sean-k-mooney:matrix.orgbasically i dont think we can do nested user namespace and there is apprenlty a know incompatiblity with bublewarp and k8s with user namespaces15:46
@sean-k-mooney:matrix.orgwith those two tweaks it worked fine on https://www.talos.dev/15:47
@sean-k-mooney:matrix.orghttps://github.com/containers/bubblewrap/issues/505 is kind of the upstream issue tracker but its a problem for any env k8s env that has user namespece enbaled as far as i can tell15:51
@fungicide:matrix.orgsean-k-mooney: what bubblewrap version are you using? zuul should be executing with --disable-userns if bw is 0.8 or later15:52
@fungicide:matrix.orghttps://zuul-ci.org/docs/zuul/latest/releasenotes.html#relnotes-8-3-1-security-issues15:53
@sean-k-mooney:matrix.orgim just using the upstream images form whatever the zuul operator is installing but i can check15:53
@fungicide:matrix.orgshould be plenty new enough then15:53
@sean-k-mooney:matrix.organy idea where its installed15:54
@fungicide:matrix.orgin the zuul-executor image15:54
@sean-k-mooney:matrix.orgah found it15:54
@jangutter:matrix.orgIn an offtopic one, anyone willing to review some overambitious upgrades to the ensure-helm role in zuul-jobs? https://review.opendev.org/c/zuul/zuul-jobs/+/962794 (been tested with things like openstack-helm and it seems to work)15:54
@sean-k-mooney:matrix.orgbubblewrap 0.8.015:55
@sean-k-mooney:matrix.orgso quay.io/zuul-ci/zuul-executor@sha256:30579d4c56d8522d1fe3a33266031c8fe4c15b8833475fccb77b282b499d5fee15:55
@fungicide:matrix.orgsean-k-mooney: i wonder if the nesting problem isn't specifically user namespaces but some other capability that also requires a privileged container15:55
@sean-k-mooney:matrix.org startedAt: "2025-10-23T14:09:56Z"15:57
@sean-k-mooney:matrix.orgso that was apprently latest like 3 weeks ago15:57
@sean-k-mooney:matrix.orgperhap15:57
@sean-k-mooney:matrix.orgi dont have 8.3.1 however only 8.0.015:57
@sean-k-mooney:matrix.orgso maybe the zuul exector iamg ejust had not been updated with it yet or i was in the time window before this was fixed?15:58
@sean-k-mooney:matrix.orgi deploy this like the week after the summit and before the ptg15:58
@sean-k-mooney:matrix.orgso i didnt spend a lot of tiem digging into the root cause ocne i foudn a fix15:59
@sean-k-mooney:matrix.orgwell workaround15:59
@fungicide:matrix.orgsean-k-mooney: 8.3.1 was the zuul version where that security note was covered, not the bubblewrap version (which indicates support in bw 8.0.0)16:00
@sean-k-mooney:matrix.orgoh16:00
@sean-k-mooney:matrix.orgok16:00
@sean-k-mooney:matrix.orgthen yes it could be another cap issue that is resolve. to be faire  i think i also change teh defautl security context on the namespace at some point too16:01
@fungicide:matrix.orgthough it's worth noting that bw is up to a 0.11.0 release as of last year. i wonder why we're sticking with an older version in our images16:02
@fungicide:matrix.orgoh, we install the distro package for it16:03
@sean-k-mooney:matrix.orgone of the things i did was make the namespace prvidaged isntead of restricted when deployign https://termbin.com/9emm16:03
@sean-k-mooney:matrix.orgso combined with running the executor now as root its works but as i said proably not the correct thing to do outside of a home lab :)16:04
@fungicide:matrix.orgyeah, our debian bookworm base image gets a bubblewrap 0.8.0 package, when we update to debian trixie we'll have bubblewrap 0.11.016:04
@sean-k-mooney:matrix.orgso i was debating pushing up a patch to make setting  runAs* configurable so you could opt out 16:06
@sean-k-mooney:matrix.orgbut i did nto have time to figure out how the templating system works in the operator so just hardcoded for my local install for now16:06
@jangutter:matrix.orgI just find it deeply comforting to know that every k8s cluster is in some or other way a snowflake. The amount of customisations applied for different reasons means that your standard install mechanism basically needs to be a turing-complete scripting language.16:58
@fungicide:matrix.orgone of the reasons in opendev we still just use ansible to orchestrate docker-compose on cloud virtual machines17:00
@jangutter:matrix.orgBah, clearly you folks are not enterprisey enough. I mean all your logs are available publicly, where any competitor can read them for free!17:24
@sean-k-mooney:matrix.orgjangutter:  in this case talos is enablign strictor policies by default the upstream k8s ships out of the box17:34
@sean-k-mooney:matrix.orgout of the box k8s or kind does not enforce the resticted SSC context17:34
@jangutter:matrix.orgyep - bare k8s is kinda like the Linux kernel17:35
@jangutter:matrix.orgyou need some sort of k8s distro for it to be useful - even kind and minikube are k8s distros.17:36
@mnaser:matrix.orghave you met openstack? =P17:37
@sean-k-mooney:matrix.orghttps://docs.siderolabs.com/kubernetes-guides/security/pod-security17:37
@sean-k-mooney:matrix.orgmnaser: openstack is more standard out of the box from my experince17:37
@jangutter:matrix.orgwhich is why I find it comforting!17:38
@sean-k-mooney:matrix.orga vanila k8s isntall does not even have auth configured or networking17:38
@mnaser:matrix.orgmaybe from a nova perspective, but stuff like cinder or glance can be fun to discover how different it operates17:38
@sean-k-mooney:matrix.orgso the firest ting you have to do is decied how authenticationa and networkign and storage is going to work17:38
@mnaser:matrix.orgbut yes to be fair, you need a csi/cni and none of those operate in the same ways either17:39
@sean-k-mooney:matrix.orgopenstack's approch is there is one servce for each compoented and it can internally have external backend. k8s approch is there is an interenface for each logical thing that anyone can impelente with no default implemetion out of the box17:40
@sean-k-mooney:matrix.orgya so opentack's approch is the impleation defined an openstack wehre as k8s's appoch is its the api17:40
@sean-k-mooney:matrix.orgboth have pros and cons17:40
@jangutter:matrix.orgThe k8s folks lean in very heavily towards https://12factor.net/ and view anything not strictly following these principles as pathological.17:45
@sean-k-mooney:matrix.orgwell they fail on point 1 17:46
@sean-k-mooney:matrix.orgOne codebase tracked in revision control, many deploys17:46
@sean-k-mooney:matrix.orgbut yes they do17:47
@jangutter:matrix.orgThey also tend to make the assumption that some resources should be regarded infinite and considering small scale usage is pathological.17:47
@sean-k-mooney:matrix.orgwell that a tenant of clouds :)17:48
@jangutter:matrix.orgClearly you're not enterprisey enough!17:49
@sean-k-mooney:matrix.orgi am ok with that in all aspcts of my life17:51
-@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed: [zuul/zuul] 966623: Add support for zk watch zxids https://review.opendev.org/c/zuul/zuul/+/96662317:51
@jangutter:matrix.orgI for one refuse to eat anything until the proper JIRA has been approved by management. Done wonders for my weight, but I'm starting to see spots.17:52
@mnaser:matrix.orgi think the xterm-helper-textarea seems to tell me that this might be https://github.com/xtermjs/xterm.js so maybe bringing that up to date might land us improvements17:54
@mnaser:matrix.orgI am going to do some work on zuul-web to bump the dependencies such as xterm and see if we can improve the sitaution :)18:00
@jangutter:matrix.orgWant me to send up a hamfisted attempt?18:01
@jim:acmegating.commnaser: if you can bump xterm alone, great.  if you run into the deadlock we currently have with cra, openapi, etc, i suggest avoiding it.  there is a complex untangling that needs to be done, and it's on my list.18:02
@mnaser:matrix.orgyeah i will try to work my way around not playing with too many of those more complicated pieces18:02
@jangutter:matrix.orgyarn thinks it might be as simple as18:02
- "xterm": "^4.12.0",
+ "xterm": "^5.3.0",
@mnaser:matrix.orgthey moved to @xterm/xterm and addons moved to another namespace, so addressing that / fixing imports / etc18:03
@jangutter:matrix.orgoooh, yeah, I didn't even read the warnings so many warnings.18:03
@sean-k-mooney:matrix.orgah interesting its alwasy nice if you can just bump a dep and they have already fixed your issue18:16
-@gerrit:opendev.org- Jan Gutter proposed: [zuul/zuul] 966625: Test new xterm.js https://review.opendev.org/c/zuul/zuul/+/96662518:17
@jangutter:matrix.orgyeah, xterm.js 5.x has made a ton of breaking changes. Maybe we're lucky and 4.19 solves your problem. https://github.com/xtermjs/xterm.js/releases18:26
-@gerrit:opendev.org- Mohammed Naser proposed: [zuul/zuul] 966626: Upgrade Xterm.js to version 5.x https://review.opendev.org/c/zuul/zuul/+/96662618:45
@mnaser:matrix.orgi'll have some more stylistic-ish changes to follow up that may help the looks but splitting those out since they are more of an opinion :)18:46
@mnaser:matrix.orghttps://imgur.com/a/QBtbAss18:56
@mnaser:matrix.orgi would say an improvement :) but maybe in my opinion.. i will push up this follow up change and let people be the judges in their own browsers18:56
-@gerrit:opendev.org- Mohammed Naser proposed: [zuul/zuul] 966628: Improve console text readability https://review.opendev.org/c/zuul/zuul/+/96662818:58
-@gerrit:opendev.org- Mohammed Naser proposed:19:02
- [zuul/zuul] 966626: Upgrade Xterm.js to version 5.x https://review.opendev.org/c/zuul/zuul/+/966626
- [zuul/zuul] 966628: Improve console text readability https://review.opendev.org/c/zuul/zuul/+/966628
@sean-k-mooney:matrix.orgmonospace helps as does increase the ling hight19:06
@sean-k-mooney:matrix.orgbut ideally it would scale dynmaicly 19:06
@sean-k-mooney:matrix.orgah you update dit to 1.219:07
@sean-k-mooney:matrix.orgnot 12 px19:07
@sean-k-mooney:matrix.orgya that shoudl generally help as it now in theory relitive to the point size of the text19:08
@mnaser:matrix.orghttps://storage.gra.cloud.ovh.net/v1/AUTH_dcaab5e32b234d56b626f72581e3644c/zuul_opendev_logs_9ed/zuul/9ed9641f4a08448fa982e2d44cfe6f6e/npm/html/ preview for those who are curious :)19:14
@sean-k-mooney:matrix.orgah nothing liike testing in psudo productoin19:18
@jangutter:matrix.orgImpressive!19:18
@sean-k-mooney:matrix.orgseems to work up to 200% scale fine for me19:19
@jangutter:matrix.orgMan, imagine needing to run this across all the myriad combinations of browsers, hardware and OSes to spot the breakdowns.19:20
@sean-k-mooney:matrix.org<div style="width: 1682px; height: 18.3333px; line-height: 18.3333px; overflow: hidden;"><span style="letter-spacing: 0.00013412px;">2025-11-10 18:15:28.180863 | controller | {0} tempest.api.image.v2.admin.test_images_metadefs_resource_types.MetadataResourceTypesTest.test_basic_meta_def_resource_type_association [0.579411s] ... ok</span></div>19:21
@sean-k-mooney:matrix.orgso it looks like the way its represented has changed a lot19:21
@sean-k-mooney:matrix.orgi can see that the value now change when i increase and decrease the zoom19:21
@sean-k-mooney:matrix.orgwere as before it would not19:22
@jangutter:matrix.org"we don't trust the browser to do the right thing"19:22
@jangutter:matrix.orgAlso, insane to me that fractional pixels are now the reliable way to render.19:23
@sean-k-mooney:matrix.orgwell this is shoign the comptue values in teh debugger19:25
@sean-k-mooney:matrix.orgbut ya that how to get it to be constent for every one with the same DPI display :P19:26
@sean-k-mooney:matrix.orgmnaser: my solution to moving to a 4k display was to get 4k lg cx 40 inch old tv so i had the same pxl densitiy as my previous ultrawide 1440p display which had the same densitiy as my previoul 27inch 1080p display19:28
@sean-k-mooney:matrix.orgunfortually if i want to upgrade again i will also need a bigger desk...19:29
@mnaser:matrix.orgsean-k-mooney: this is what i ended up with https://rog.asus.com/monitors/27-to-31-5-inches/rog-swift-oled-pg27ucdm/ :) -- i used to run an lg cx display two, but i have two of those now19:33
@mnaser:matrix.orgoled + 4k + nice size (so can run scaled) + 240hz19:34
@fungicide:matrix.orghighest res my monitors go is 2k, but on a 7" diagnonal that can still get very pixel-dense19:36
@mnaser:matrix.orgseems like zuul-nox-py312 is failing on my changes but looks like its unrelated to my hcanges19:36
@mnaser:matrix.orgyeah that's neat, at the end of the day it comes down to ppi instead of size :)19:37
@fungicide:matrix.orgright, my newer netbooks are 2560x1440 resolution and on quite small panels19:39
@fungicide:matrix.orgi have one where that's the native res on a 5" diagonal panel even19:40
@jangutter:matrix.orgI have a theory that most of the drive to high ppi + refresh is driven by cats. Their vision system requires this otherwise things are a flickering mess. Plus their natural sociopathy means that they are open to manipulating the human servitor species.19:56
@fungicide:matrix.orgi'm owned by multiple cats, would believe the explanation20:00
-@gerrit:opendev.org- James E. Blair https://matrix.to/#/@jim:acmegating.com proposed:22:05
- [zuul/zuul] 966501: Ensure requests referred to in assignments are up to date https://review.opendev.org/c/zuul/zuul/+/966501
- [zuul/zuul] 966637: Add min_request_version to assignUnassignedNode method https://review.opendev.org/c/zuul/zuul/+/966637
- [zuul/zuul] 966638: Make ZuulTreeCache.waitForSync more robust https://review.opendev.org/c/zuul/zuul/+/966638
@mnaser:matrix.orgStrange failure on my xterm change -- `AttributeError: 'TestCmd' object has no attribute 'assertEquals'. Did you mean: 'assertEqual'?` .. i seem to be the first one hitting it22:20
@mnaser:matrix.orgtesttools 2.8.0 released 12 hours ago22:23
@mnaser:matrix.orghttps://github.com/testing-cabal/testtools/pull/424 this probably did it22:23
-@gerrit:opendev.org- Mohammed Naser proposed:22:29
- [zuul/zuul] 966626: Upgrade Xterm.js to version 5.x https://review.opendev.org/c/zuul/zuul/+/966626
- [zuul/zuul] 966628: Improve console text readability https://review.opendev.org/c/zuul/zuul/+/966628
- [zuul/zuul] 966643: Switch to using assertEqual() in tests https://review.opendev.org/c/zuul/zuul/+/966643
@mnaser:matrix.orgI put the wrong commit ID link and need to fix flake8, will do when near desktop again23:23
-@gerrit:opendev.org- Mohammed Naser proposed:23:56
- [zuul/zuul] 966643: Switch to using assertEqual() in tests https://review.opendev.org/c/zuul/zuul/+/966643
- [zuul/zuul] 966626: Upgrade Xterm.js to version 5.x https://review.opendev.org/c/zuul/zuul/+/966626
- [zuul/zuul] 966628: Improve console text readability https://review.opendev.org/c/zuul/zuul/+/966628

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!