Wednesday, 2025-10-01

tonybclarkb: I feel that.00:11
tonybI need the output of `find /afs/openstack.org/mirror -ls` I know I can run that on any system with the /afs/openstack.org mounted but in terms of I/O and runtime and strain on the server is there a "best" place to run that?00:52
tonybI'm trying to generate some tests for my new repomd-check and repomd-filter tools for reducing the files we mirror and verifying that we have a complete mirror00:53
opendevreviewOpenStack Proposal Bot proposed openstack/project-config master: Normalize projects.yaml  https://review.opendev.org/c/openstack/project-config/+/96255702:17
opendevreviewMichal Nasiadka proposed opendev/zuul-providers master: Add Alma Linux 10 image build jobs  https://review.opendev.org/c/opendev/zuul-providers/+/96256405:09
opendevreviewMichal Nasiadka proposed opendev/zuul-providers master: Add Alma Linux 10 image build jobs  https://review.opendev.org/c/opendev/zuul-providers/+/96256405:11
mnasiadkacorvus: thanks, squashed05:28
mnasiadkaclarkb: updated05:28
mnasiadkaclarkb: regarding the Rocky10 issue - I noticed that Rocky mirror ,,propagation’’ is not the best, in most cases I just use dl.rockylinux.org which has the latest content05:29
mnasiadkaclarkb: now iptables has some issue - I’ll recheck in the afternoon in hope that it fixes itself...05:41
*** tkajinam_ is now known as tkajinam10:54
*** darmach47 is now known as darmach13:01
mnasiadkaWhat is funny, arm64 build is passing14:35
clarkbmnasiadka: still a package thing then?14:43
mnasiadkaclarkb: yup14:44
mnasiadkaI don’t really know which mirror we’re using, so it’s probably hard to report anything.14:44
mnasiadkaI’m also seeing some Debian libssl-dev issues - probably debian-security mirror is desynched from the regular repo - works when not using the openmetal mirror (https://64a54baaf22851aff2b6-ebad28ee85ce63233d9ad88daf84e2e7.ssl.cf2.rackcdn.com/openstack/646c294c4a604fe0aa86518870954d58/kolla/build/000_FAILED_kolla-toolbox.log)14:48
mnasiadkaI’ll wait until tomorrow ;-)14:48
fungimnasiadka: debian just issued a dsa about openssl, so that sounds likely14:52
mnasiadkaActually I see mirror.debian got release 12 minutes ago, so maybe I’ll just retry - https://grafana.opendev.org/d/9871b26303/afs?orgId=1&from=now-6h&to=now&timezone=utc14:52
fungiyeah, it looks like our debian-security and debian mirrors are getting successfully updated on schedule14:52
opendevreviewMerged openstack/project-config master: Revert "Temporarily remove release docs semaphores"  https://review.opendev.org/c/openstack/project-config/+/96247715:10
mnasiadkafungi: yup, the problem is gone, so it’s properly in sync now15:28
fungiawesome15:28
opendevreviewClark Boylan proposed opendev/zuul-providers master: Revert "Disable rax-dfw due to API errors"  https://review.opendev.org/c/opendev/zuul-providers/+/96269715:34
clarkbI mentioned yesterday that I thought ^ was something we could do after the openstack release. I believe the openstack release is largely complete at this point (fungi can confirm) so we should be able to proceed with testing that whenever we like15:35
fungiyes, let's. approved15:35
opendevreviewMerged opendev/zuul-providers master: Revert "Disable rax-dfw due to API errors"  https://review.opendev.org/c/opendev/zuul-providers/+/96269715:36
clarkbhttps://grafana.opendev.org/d/fd44466e7f/zuul-launcher3a-rackspace?orgId=1&from=now-30m&to=now&timezone=utc&var-region=$__all is what I'm watching to see how dfw does15:43
clarkbideally that building number decreases and the ready/in-use numbers increase15:43
opendevreviewMichal Nasiadka proposed opendev/zuul-providers master: Add Alma Linux 10 image build jobs  https://review.opendev.org/c/opendev/zuul-providers/+/96256415:44
clarkbearly indications look good. That region has done exactly what I wanted. Building is down to ~12, ready is 3 and in use is 3715:50
fungiawesome15:52
clarkbit continues to look good. I'm going to context switch back to laptop prep and debugging now.16:32
clarkbI'm discovering that I need to consider better ergonomics when laptoping. Using a laptop atop a lap is actually terrible for my neck16:32
fungineck, elbows, wrists...16:37
fungiadding a lapdesk can help16:37
clarkbok disabling panel self refresh may be the ticket. That probably increases power draw though but I'll live with that until they fix it17:13
clarkbthe last major thing I need to do is set up wireguard. Which by the way is pretty awesome for setting up a VPN home17:17
clarkbI have found taht NetworkManager's wireguard support just doesn't work. But the wg-quick script in wireguard-tools is simple and works great17:18
corvuswg is also available for android17:29
clarkbyup I run it on my phone too. Bonus is I get all the ad blocking firewall and dns magic that way too17:38
clarkbthe one thing I'm finding is lacking is on my router/host side. I wish there was a way to do some basic IP tracking in the peer config so that I don't try and remember or tracke elsewhere which IPs I'm assigning to peers17:40
clarkbI'm using opnsense17:41
opendevreviewJeremy Stanley proposed opendev/system-config master: Document paste deletion from the Lodgeit database  https://review.opendev.org/c/opendev/system-config/+/96270918:45
clarkbI can't remember the difference between single backtick and double backtick quoting but I went ahead and approved that18:48
fungii just spotted a typo anyway18:48
opendevreviewJeremy Stanley proposed opendev/system-config master: Document paste deletion from the Lodgeit database  https://review.opendev.org/c/opendev/system-config/+/96270918:50
fungisorry about that18:50
fungii also reordered the sentences in one paragraph18:50
fungiargh, another typo18:51
opendevreviewJeremy Stanley proposed opendev/system-config master: Document paste deletion from the Lodgeit database  https://review.opendev.org/c/opendev/system-config/+/96270918:52
fungii should do a better job of proofing my changes before i push them18:52
clarkband reapproved18:52
clarkbfwiw I did read it too but my brain must have skimmed right over those issues18:53
fungiyeah, i mean, i read it several times after i wrote it, but it's always easier to spot errors after i push it for review18:53
* fungi shakes fist at elusive laws of reality18:54
opendevreviewMerged opendev/system-config master: Document paste deletion from the Lodgeit database  https://review.opendev.org/c/opendev/system-config/+/96270919:02
clarkbthis is a fun issue. This laptop is the first one where I'm trying to run with everything 2x scaled. This seems to work great for the native display but as soon as I plug into a 1920x1080 stand in for a conference projector the scale no longer works. And I'm not sure I can control this per display21:56
JayFWhat DE?21:57
clarkbJayF: xfce21:57
JayFHmm. I will note that the underlying tech will do it, it's not great (the scaling is per-window, so if it's over the border of two different scales it picks one), but it does it21:57
clarkbits fine for projecting I'll just have to switch back to 1x scaling and run the buitl in display at 1920x1080 which wroks well enough for things like that21:57
JayFI know how to do it in sway, but I know basically nothing about xfce21:57
clarkbsway is also wayland not X21:58
clarkband I think they both deal with this stuff differently21:58
JayFX does this better than wayland fwiw21:58
JayFbut they both have the basics down21:58
* JayF learned way more about this than he ever wanted to getting jetbrains ides to work sanely on his system21:58
clarkbeventually everything will be 4k and it won't matter anymore21:58
clarkbbut for conference projectors I can just mirror idsplays and run at 1920x1080 across the baord with no scaling21:59
JayFI have a dedicated user with a bog-standard gnome for projectioning22:01
JayFI have it all working properly in sway, but for some reason the bigger AV systems at conferences hate it (including the SCALE magic box that works with everything)22:01
clarkbhuh I've never had any problems with xmonad specifically when projectoring. Its always just another display22:02
clarkbI do find xmonad multimonitor to be weird so I typically do try to mirror22:02
clarkb(there are other people who disagree on this vehemently and say xmonad is the one system that handles multimonitor well. I dunno my mental model for 3d space isn't good enough I guess)22:02
clarkbI've even got the key combo to extend the tiling space over my status bar so that I can properly full screen22:03
clarkbthe other fun I've just discovered is that dmenu doesn't seem to repsect the old school font specification any more -*-source-code-pro-medium-*-*-*-*-p-*-* or whatever and I've had to convert to xft:SourceCodePro-Medium:style=bold:pixelsize=XY but that is a million times more readable so I'll take it22:05
JayFI just generally blame sway? It's a pretty thin layer which is good for day-to-day but not great for things like projectoring22:05
JayFI use fuzzel for dmenu-ish stuff22:05
JayFI think this is a parallel ecosystem thing though, where my tools are going to msotly be wayland-only22:05
clarkbyes, this is why I haven't switched and its really the great failing in the wayland transition I think22:06
clarkblike its great you can nest X in wayland but we almost needed a wayland is an X server transition or something so that we don't have to chgne tools across the board22:07
JayFI switched once I realized there was a perceptible speed increase when using wayland-only (this was back when I used gnome)22:07
clarkbmakes ripping the bandaid off difficult22:07
JayFonly x clients I have running are the slack electron app and gsd-xettings 22:07
JayFand given I went straight to sway from gnome, I basically have been wayland-first in all my seutp22:08
clarkbonce upon a time I would use x forwarding a lot too, but these days even "thin" clients tend to be zoomy enough to not need that sort of functionality22:08
JayFwaypipe supposedly does that for wayland, I've never got it fully working22:09
JayFbut I generally hate remote desktop and all that with a fury. UI latency is king.22:10
fungiratpoison works great great for me with multiple monitors running different resolutions, even attaching and detaching them on the fly22:10
clarkbya the primary reasons for it were when you actually got less latency because running utilities on beefy linux servers was faster than on your local sun workstation22:10
clarkbbut it was also a college campus with a zoomy lan22:11
fungithough to be clear, i configure my displays from the commandline using xrandr22:11
clarkbfungi: ya I think this problem would exist for you too if you were using scaling with x too fwiw22:11
clarkbits inherent to the problem of resolutions not all being within a close range of each other anymore22:12
fungii don't bother to adjust scaling at the x level, i just scale within the application (my terminals, browser windows, whatever) as needed22:12
JayFfungi: it's crazy to me how much a keyboard-driven workflow makes using a laptop so much more pleasant. 22:12
JayFfungi: I obvs don't use ratpoison but my sway is setup to be anti-rodent :)22:12
clarkbJayF: I find tiling is just a better organizational method too even if you have a mouse22:13
fungii still don't use a login manager either, just log into a getty-type linux virtual console and startx, all my setup is driven from my ~/.xinitrc22:13
clarkbbefore I disabled panel self refresh the way to fix laggy display output was to switch to vty1 then back to vty722:14
JayFclarkb: I didn't wanna use sway at first, I switched after trying just about every tiling extension for gnome in existance22:14
clarkbmade me think of the days when I ran arch and used startx too22:14
clarkbJayF: xfce doesn't "support" alternative window managers so I actually have $HOME/bin/xfwm4 symlinked to /usr/bin/xmonad22:15
clarkbJayF: and it works22:15
fungii'm sure i'll eventually have to ditch ratpoison since it doesn't seem to be actively maintained these days, but it still works well and the keybindings are reminiscent of gnu screen, unlike a lot of other mouseless tiling wms22:15
JayFI'm happy to use sddm still :) but I also want all the dbus madness and such to work properly, and I'm not *sure* how to do that direct from cli22:15
JayFfungi: sway is mostly BYO-keybindings22:15
JayFor at least I'll say: I made all mine 100% custom22:15
fungiyeah, i could probably configure it with a profile that mostly matches rp22:16
JayFhttps://github.com/jayofdoom/dotfiles/tree/main/sway/conf.d22:16
JayFhttps://github.com/jayofdoom/dotfiles/blob/main/sway/conf.d/02-framework-output and https://github.com/jayofdoom/dotfiles/blob/main/sway/conf.d/15-keybindings being the most relevant to this chat22:16
clarkbTIL about https://docs.qtile.org/en/stable/22:17
clarkbmaybe this is what I should be switching to as ti can wayland and X then I can switch one thing at a time22:17
JayFI'll note wlroots-based compositors are missing a crucial feature for many folks22:18
JayFyou can only share the entire screen, not a single window/app22:18
JayF(and even that support is more recent than is reasonable)22:18
clarkbI find with X and meetpad/zoom at least it doesn't like to share windows in other workspaces or other workspaces entirely which si also annoying with a tiling wm22:22
clarkbI basically want to put the management for sharing in one workspace then the recording in another so they aren't getting in each others way as I interact with them separately22:23
fungialso possible to do that with two separate machines22:24
fungiif you really want them isolated22:24
clarkbits mostly that if you want to share say a terminal then letting it be whatever size you want independent of the chrome browser that is also getting tield is nice22:25
JayFAs an aside: https://gitea14.opendev.org:3081/ is showing up in some searches for me22:25
JayFI imagine you are not expecting an alternate-port version of gitea on google22:25
clarkbI tend to end up untiling the terminal and letting it float22:25
clarkbJayF: we are not, but we have noticed that ai bot crawlers also don't care about non standard port locations22:25
JayFI mean, this is google. So it's AI but not /AI/22:25
clarkbya I'm just pointing out its not new the crawlers are finding it. I think what is new is that google is apparently silly enough to index it for web search22:26
fungii wonder if we could serve up different robots.txt files depending on whether it's coming through haproxy vs straight to apache22:26
JayFyeah and it's only for the obscure thing I was searching, e.g. 'openstack ironic opendev' gives me a proper URL22:27
JayFI was searching for references to the old teeth-agent to try and timestamp my first openstacking of any kind22:27
JayF(Nov 2023 fwiw)22:27
JayF**1322:27
clarkbfungi: googling seems to indicate we should use canonical location tags22:27
fungithat's probably configurable in the apache vhosts, yep22:28
clarkbhowever, I don't know how we'd get gitea to selectively serve canonical tags based on requests coming through haproxy or direct22:28
clarkbactually I guess the canonical location doesn't change so we can just always serve it22:28
fungii assume it's just a header apache would add, then we wouldn't need to differentiate22:28
clarkbfungi: no its in the html22:29
fungioh, odd...22:29
fungiare http headers no longer cool?22:29
clarkbfungi: https://developers.google.com/search/docs/crawling-indexing/consolidate-duplicate-urls?hl=en&visit_id=638949544692725832-3028177890&rd=1#rel-canonical-link-method22:29
fungii wonder why that's at the html layer22:29
fungimaybe to support content mirroring22:30
clarkbI think this would require patchign gitea22:30
clarkbbut would be doable22:30
JayFI'm thinking in a shared hosting world, you'd want the client (uploading html) to be able to set that, not have to have the server (.htaccess or similar) send the header22:31
clarkbfungi: oh that doc says it can also supply it as a header22:31
clarkbso ya we could potentially have apache serve it for all content and just say the canonical location for any url is that url at https://opendev.org/22:32

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