Tuesday, 2023-08-29

fungiyep, i received a copy00:11
liuc49Hi,everyone, I create an instance and assign a floatingip. But the ip cannot ping successfully and ssh enable. what's the problem?08:19
fricklerliuc49: this channel is about the infrastructure used to develop OpenStack. if you could stick around in the #openstack-qa channel where you already asked your question, you might get an answer there08:30
fricklerinfra-root: I'm currently reviewing some openstack docs. do you agree that we should replace references to eavesdrop with meetings.opendev.org consistently?09:40
fungifrickler: i agree that eavesdrop.openstack.org is now a cname and permanent http redirect to meetings.opendev.org, so that's an accurate change at least (i still think hosting channel logs under a "meetings" domain name is confusing, but it's our current reality)11:19
fricklerwell those are some kind of asynchronous permanent 24x7 meetings, if you want to stretch it a bit. I don't think changing the name once again or splitting up into something like chatlogs.o.o would be helpful11:23
*** NeilHanlon_ is now known as NeilHanlon13:11
fungiinterestingly, the change to stop mirroring i686 packages made minimal impact on the amount of storage consumed on the centos and centos-stream volumes: https://grafana.opendev.org/d/9871b26303/afs?orgId=1&from=now-24h&to=now13:27
TheJuliaGreetings, Can I get a hold for job ironic-tempest-ipa-partition-uefi-pxe-grub2 on change id # 885087 for the purpose of debugigng OVN GRUB Network Boot Failures13:27
fungizuul-client autohold --tenant=openstack --project=opendev.org/openstack/ironic --job=ironic-tempest-ipa-partition-uefi-pxe-grub2 --ref='refs/changes/87/885087/.*' --reason='TheJulia debugigng OVN GRUB Network Boot Failures'13:30
fungii've added that13:30
TheJuliaMuch appreciated!13:30
fungiyou bet!13:30
fungii've also deleted my mailman3 autohold now that all the changes we were demoing on there are fully in production13:31
fungilooks like we reached 557 concurrent builds a few minutes ago14:00
*** ykarel_ is now known as ykarel14:01
opendevreviewClark Boylan proposed opendev/system-config master: Update Gitea images to bookworm  https://review.opendev.org/c/opendev/system-config/+/89307314:54
clarkbinfra-root ^ the commit message calls out my biggest concern with that update which is replication over ssh using rsa keys since openssh-server updates past the 8.8 release14:55
clarkbwe test git push over ssh in our test jobs, but not from a MINA client. So there is some risk there14:56
clarkbI think it should all work, but open to feedback on how much testing we want to do. We can hold nodes and set them up to replicate between gerrit and gitea as a complete test if we feel that is necessary14:56
opendevreviewJulia Kreger proposed openstack/diskimage-builder master: Deprecate legacy deployment elements  https://review.opendev.org/c/openstack/diskimage-builder/+/89307715:13
opendevreviewJulia Kreger proposed openstack/diskimage-builder master: Remove deprecated deployment elements  https://review.opendev.org/c/openstack/diskimage-builder/+/89307815:13
opendevreviewSlawek Kaplonski proposed opendev/irc-meetings master: Make Yatin new chair of the neutron-ci meeting  https://review.opendev.org/c/opendev/irc-meetings/+/89308115:28
opendevreviewMerged openstack/diskimage-builder master: Update the shebang to python3  https://review.opendev.org/c/openstack/diskimage-builder/+/88964015:44
opendevreviewMerged opendev/irc-meetings master: Make Yatin new chair of the neutron-ci meeting  https://review.opendev.org/c/opendev/irc-meetings/+/89308115:53
clarkbthe git push to gitea failed on the bookworm change. Maybe this is more of a problem than I anticipated. But testing caught it so yay?16:35
frickler/etc/ssh/sshd_config line 36: no argument after keyword "AllowUsers"16:37
fricklerbut yay for testing, though some more basic issue than expected16:37
clarkbah you found it quicker than I did. Was that in the ssh container log?16:38
clarkbAllowUsers doesn't actualyl show up in our system-config git repo16:39
clarkbI should be able to fetch the container image and work backward from there though16:39
clarkbit is just AllowUsers on a line with nothing else which seems to line up with the error16:44
clarkb/usr/bin/entrypoint calls /etc/s6/openssh/setup which runs envsubst < /etc/templates/sshd_config > /etc/ssh/sshd_config16:45
clarkbthat template has AllowUsers ${USER} in it16:46
clarkbprod has the same problem but I guess older ssh doesn't care about that16:47
TheJuliao/ Looks like my hold is ready for me! who wants my ssh pub key? :)16:51
clarkbTheJulia: I can do ti16:55
TheJuliamuchas gracias16:56
opendevreviewClark Boylan proposed opendev/system-config master: Update Gitea images to bookworm  https://review.opendev.org/c/opendev/system-config/+/89307317:01
fricklerI haven't found any mention about such a change in the openssh releasenotes, might have been an unintentional fix17:07
fricklergit blame finds this: "Config parsing will be a little stricter wrt ... directives appearing without arguments." https://github.com/openssh/openssh-portable/commit/a10f929d1ce80640129fc5b6bc1acd9bf689169e17:13
fricklerso yes, not in 8.4. just "a little stricter" ;)17:14
clarkbthats probably it then17:14
TheJuliaYou guys can reclaim that held node, but any chance you can hold any success or failure of the same job name, any change id so I can compare? We appear to have a breaking bug but need to identify what tickles the issue and it is just the easiest way to do so18:29
clarkbTheJulia: we can only hold failures automatically iirc. You can opush a change with an artifical failure after the point it current breaks then sift through for a "success" there 18:30
TheJuliaokay, I'll do that after I take a nap then18:31
TheJuliaI mean, any failure on it with the default job config would be fine too :)18:31
TheJuliajust... super weird behavior :(18:31
*** melwitt_ is now known as melwitt18:34
TheJuliaoh, you guys can reclaim the node held on 000000024620:43
clarkbTheJulia: will do.20:44
clarkbfungi: I'm ready for meetpad testing whenever you are20:44
fungiokay, just finished eating20:58
fungilemme grab my wreath and clicker20:58
fungiclarkb: https://meetpad.opendev.org/MisleadingJazzesHostSolely21:06
clarkbfungi: we are already in isitbroken21:07
clarkbbut we need a third person to debug if bandwidht saving is on my end or corvus' side21:07
fungioh, omw21:08
fungiit was also a good opportunity to test out my new webcam ahead of the actual rehearsal call for the openinfra.live zuul episode21:25
fungiand glad that this bluetooth bone conduction headset has started working much more reliably after i got around to switching form pulseaudio to pipewire/wireplumber21:26
JayF\o/ I've heard that from a lot of folks (about pipewire having better support for misc. perhipherals) 21:27
clarkbthe headset I use is like 16 years old. But it keeps working so I'm refusing to replace it21:27
clarkbbut I did switch to an external USB DAC because whatever sound device is on this motherboard doesn't work at all last I tried it. I should test it again though21:28
clarkbhowever supposedly using the dac will make my voice come through much better for everyone else so meh21:28
fungiJayF: in particular, bluetooth two-way audio is very dodgy with anything other than pw/wp21:28
JayFYeah, AIUI PA is borderline unmaintained and was for a while; the pipewire migration took longer than folks thought21:29
clarkbre pipewire I'm running it but also pipewire-pulse and xfce uses the pulseaudio widget for volume management and muting so I don't actually know where my audio is being sent21:29
fungisupposedly pipewire inherited bt audio subsystems from the ofono telephony project21:29
JayFclarkb: so pipewire implements the pulseaudio api21:30
JayFclarkb: so clients all still use libpulse, any utilities that talked to pulseaudio still work, etc -- just the backend changed21:30
fungiyeah, i'm still using pavuctl as a client to enable/disable sinks and sources21:31
JayF(A common breaky case in gentoo is people getting excited they migrated to pipewire, set USE=-pulseaudio, and are then confused when their audio stops working ... they sense reworked how the use flags worked)21:31
clarkbthat makes sense to not have to rewrite all the frontend volume management tools21:31
fungithough i rewrote the mute toggle my bt button calls to use the wpctl utility rather than pactl (or was it pacmd?)21:32
JayFyeah, pipewire itself also is what powers new-style wayland screen sharing on linux21:32
clarkbI'm going to be sad when I have to switch to sway21:32
fungi`wpctl set-mute @DEFAULT_SOURCE@ toggle` will switch my mic on if it's off or vice versa, and then i just follow it with a `wpctl get-volume @DEFAULT_SOURCE@` and echo that as a notification to my desktop21:33
clarkbfungi: my headset has a physical inline mute slider. I really like the feature21:33
fungihave xbindkeys configured to recognize the button from my clicker and run that script in the background every time i press it21:34
fungithere is no "inline" for this headset since it's all wireless and super low-profile21:34
JayFit's probably a bug if you can't software-mute the mic using pa apis21:34
fungiso i just wanted something else wireless i could hold in my hand to mute and unmute21:34
JayFoooooh, I see what you're saying now21:35
fungiyeah, there's a "smart button" on this bone conduction wreath but it involves me blindly reaching up to my temple and is also a noticeable action on camera anyway21:36
clarkbinsecure-ci-registry02.opendev.org is the only host with that name prefix in inventory and resolving insecure-ci-registry.opendev.org in DNS gives the 02 name back21:37
clarkbI'm going to delete insecure-ci-registry01.opendev.org now (finally)21:37
fungithe bluetooth "button" i used is sold primarily as a virtual camera shutter bulb, but it just shows up in the kernel as a bluetooth keyboard (granted it's a one-key keyboard, but whatevs)21:37
JayFI have like, two of those.21:38
JayFNice to know they are generally useful21:38
clarkb#status log Deleted insecure-ci-registry01.opendev.org (e8753b25-4743-402e-8ec5-8987d11202eb) as it has been replaced by a newer server.21:39
opendevstatusclarkb: finished logging21:39
fungiJayF: install xbindkeys, stick two lines in your .xbindkeysrc and have it call a two-line script... bingo you have a remote mic toggle button21:40
JayFI'm on wayland, but I know the waylandy ways to do that, too21:40
JayF(xbindkeys still might work, too?)21:40
fungii used xev to work out what key scancode the button was emitting, but it's not rocket surgery21:41
clarkbone really great things about the DAC is it has an output volume wheel on it21:41
JayFoh for sure I've done this before :D 21:41
clarkbso I have pulse/pipewire set up to always output at 100% then control volume via an "analog" device whcih is so nice21:41
JayFoh holy crap, I just realized  I can use solaar to script my G{1..5} buttons on the left side of my logitech keyboard to do whatever I want, including input switching like you do21:42
clarkbI think you can do what fungi does with a second regular keyboard too21:43
clarkbor like with a 10key pad21:43
fungithough in those cases the trick would be using keys you don't normally need21:45
clarkbI think X differentiates the sources too21:45
clarkbso you could filter on that. But maybe that isn't stable when you unplug and replug them in21:45
fungithis clicker emits what xorg sees as the volume down key, which i don't normally use on my actual keyboards21:46
fungithe "smart button" on my headset emits the play audio key, so i guess multimedia keys are common for these sorts of things to masquerade as21:47
JayFyeah, you can basically configure solaar to pass any keypress when it intercepts a keypress for a logitech-keyboard-control-key that usually isn't passed thru to the OS21:48
clarkbI haven't heard any concerns for fedora-latest removal23:33
clarkbI'm going to approve that change now23:33
clarkbif things go unexpectedly wrong I have no issues with reverting it23:34
opendevreviewMerged opendev/base-jobs master: Remove fedora-latest nodeset  https://review.opendev.org/c/opendev/base-jobs/+/89238023:43

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!