[08:51:31] morning [08:56:44] o/ [08:57:25] o/ [08:57:56] high five back! \o [08:57:57] xd [09:01:56] o/ [09:04:30] dcaro: a 2nd pipelinerun on vagrant/bookworm ran in an all-time record time of 48s (1st run: 1min30s). No logs timeout either time vs always on lima/bookworm [09:04:58] 🎉 [09:04:58] nice [09:05:31] would be interesting to find out what's the difference between the VMs setup, as both internally run the same OS and such [09:06:06] does it use a different virt hypervisor system or something? (libvirt/virtual box/xen/...) [09:08:15] I've uninstalled virtualbox completely, so at least we know no one is using that xd [09:15:57] are you running it on linux or macos? (I'm guessing linux?) [09:21:26] you guessed right (for vagrant)! with lima, I've tried it on both and found that there are no significant differences [09:25:27] hmm, is lima on linux using kvm? ('ps aux | grep accel=kvm' should show the VM qemu process running) [09:26:04] https://usercontent.irccloud-cdn.com/file/2qKaKOSj/image.png [09:26:44] nope [09:27:25] and afaik, kvm is not available for macos [09:28:06] I mean in linux xd [09:29:14] ahahha, in my head it was "of course kvm is running on linux, he must be meaning mac?" xd [09:31:11] (on linux yes, both my currently running lima vms show up in the ps aux output) [09:31:52] okok, it was just in case (I'm the kind of person that would spend an hour trying to figure out why my changes to a file don't work until I notice I'm editing the wrong file) [09:32:54] uhh, me too xd (generally, editing the right file but on the wrong machine) [09:46:42] what is a good equivalent on linux for mac's keychain, e.g. to store your git credentials and such? [09:52:07] I think seahorse is a common one (comes with gnome iirc) [10:13:15] dcaro: there are so many open lima-kilo MRs that I don't know where to best make changes anymore 😅 is there anything that could be merged already? [10:14:07] I think that the basic lima-vm MR can be merged (even if it's not perfect, it kind of works), then the removal of things on top of it can go in too [10:15:00] I think so too, we can throw on fixes later [10:15:33] I'm waiting for Raymond_Ndib.e though as he was using ith install-unistall stuff still (though for what I understood, it was not really working for him anyhow), and wanted to help him setting up a VM instead [10:17:22] ok [10:18:56] there are some lima-vm changes that are enmeshed with the remove-stuff mr though so I think I'll keep on building on top of that [10:31:48] to move stuff from the setup scripts to ansible it help to have removed the non-vm support (and uninstall), as it becomes way simpler [13:05:20] with toolforge deploy, is it at all possible to specify which branch to deploy a component from? [13:08:56] not directly, it does not clone any repository, it uses helm releases [13:09:14] but for each MR, we generate a helm release, that you can use [13:10:28] you can check the last step in the pipeline for an MR, called `publish-devchart-toolsbeta` and there you'll see the version for the chart, and put that in the component//values/local.yaml and you get that version [13:10:33] example of job https://gitlab.wikimedia.org/repos/cloud/toolforge/builds-builder/-/jobs/191695 [13:10:59] the chart version there is `0.0.89-dev-mr-30` [13:11:49] if nobody does it, one of this days I'll add a script to be able to just pass the MR number, like `./deploy.sh builds-builder --mr 48` [13:14:27] this is helpful, thanks! [13:15:51] there's a script to download the .deb packages for clis already on lima-kilo, that is quite helpful to test cli MRs [13:36:21] is there a way to get the chart version of the mr with gitlab's cli? [13:38:43] you cat try getting the artifacts from the first task (generate versions or something like that), or getting the logs from the last one and extracting it from there [13:39:30] the toolforge_download_package.py on lima-kilo is simple enough, and can give you hints on what APIs to call [14:45:28] blancadesal: Nicholas mentioned last week that catalyst has been having issues with openstack API stability -- can you tell me more about that? Or is there a ticket? [14:48:56] andrewbogott: this was mentioned to me in passing and I didn't want to bring it up before having more details, but we've still not had a proper sync post-pilot phase because of offsite/holidays/sickness [14:49:09] when I know more, I will check in with you [14:49:35] works for me! If it matters please express my interest in understanding & improving whatever it is. [14:49:36] thanks [14:50:11] absolutely [15:02:14] Do y'all have thoughts about the memory alert on clouddb1015? I'm tempted to just restart maria on that host but maybe someone wants to try a subtler approah [15:03:17] I did not look at it, that's not the one that was failing a couple weeks back no? [15:05:05] I don't think it's the same, let me check... [15:05:25] * andrewbogott remembers to use the sal toolforge tool [15:06:13] the sal has taught me nothing [15:12:01] komla, I was just re-reading T319883 and realized the developer hand-built mono / tried to use apt directly in a toolforge environment. I think they are really missing the idea of a buildpack. Can you help them out? They are another .NET user, and using lighthttpd [15:12:02] T319883: Migrate mbh from Toolforge GridEngine to Toolforge Kubernetes - https://phabricator.wikimedia.org/T319883 [15:15:15] balloons: we have https://gitlab.wikimedia.org/toolforge-repos/sample-dotnet-buildpack-app now :) [15:15:30] (they seem to be asking for an example of mono app) [15:15:36] dcaro, oh excellent. That answers of one their questions! [15:17:02] I'm drafting a basic response to the questions raised, but wanted to alert you that more support may be needed [15:18:23] it's ok, I have been focusing on getting people unblocked (including adding that buildpack + sample app), I can pitch in directly if you want too/have more specifics. Getting their code in a public repository helps a lot on reproducing the issues btw. [15:23:46] andrewbogott: I see some more alerts on clouddb1015, let me know if you want help (I'm guessing you are rebooting it) [15:24:33] I restarted maria, now trying to figure out if those are transient alerts from restarting or something uglier [15:24:54] might need to start slave and such [15:25:02] (or similar) [15:25:07] bah, it's not automated? [15:25:27] * andrewbogott starts slave [15:25:38] hm is etherpad down? [15:27:12] does not seem to work for me neither [15:28:09] I think etherpad is owned by https://office.wikimedia.org/wiki/Team_interfaces/SRE_-_Collaboration_Services/Team, let's see who's around [15:28:35] it's back [15:28:57] or almost xd, loaded the UI, but not the content yet [15:29:02] i restarted it [15:29:03] ah, there you go [15:29:07] thanks! [15:29:11] that seemed to work [15:29:35] oops, I just pinged in -sre [15:29:54] dcaro, yes, I keep asking for a public code repo.. I agree it can be much easier to simply get it working and "show" someone that way [15:30:05] uhoh now taavi is an honorary member of sre-collab [15:30:25] hahahahah, we'll add a column with 'etherpad' to the skill matrix for him [16:45:09] * taavi off [16:47:04] * dcaro off [16:47:06] cya tomorrow!