[08:54:01] morning ☕️ [08:59:28] o/ [09:15:41] morning [10:16:27] morning [10:18:52] dcaro: what would be the best way to try out all the lima-kilo changes? [10:20:11] what I've done is scratch all my VMs, then generate it again with the changes, and right away do a build, a docker run of that image to see it works, and a job run with a script to check it works too [10:24:28] I think that the most problematic parts is the connectivity between the parts (user <-> API, API <-> harbor, k8s <-> API, ...), so running something that exercises those things [10:35:21] I'm confused... when checking out https://gitlab.wikimedia.org/repos/cloud/toolforge/lima-kilo/-/merge_requests/87, 9aced9cc is shown as the latest commit. Where does ba443ae9 fit in? [10:35:35] https://gitlab.wikimedia.org/repos/cloud/toolforge/lima-kilo/-/merge_requests/87/diffs?commit_id=ba443ae97bf15d8acbeb86ac63fdfcd9334f4f61 [10:40:12] that commit is from the branch on top of that one (!90), probably I pushed with that one at some point, and then pushed back without it, but gitlab does not show 'removed 1 commit' [10:43:48] note that if you are testing with wm-lol --ref with_everything, I'm playing with the locales buildpack, so that might fail [10:45:40] 👍 [10:54:10] blancadesal: for you the home directory of the user inside the VM is the same as the one for your user? [10:55:16] my user, with the addition of .linux [10:56:49] okok, same as me, I suspect that dhinus ends up with the same, so ansible tries to setup stuff in a read-only mount [10:57:09] (as in with the same exact path inside and outside the lima-vm for the user home) [10:57:43] did you remove the bits where the vm user's home dir was being set? [11:00:32] not that I know of, I did not touch the bookworm.yaml file [11:05:21] I mean, ansible is installing in the env dir, which is in the vm user's home dir and not the mounted home dir. Does that work differently for dhinus? (I'm not sure I understood the problem) [11:15:06] https://gitlab.wikimedia.org/repos/cloud/toolforge/lima-kilo/-/merge_requests/87#note_66255 [11:15:43] I think that the issue is that the vm user home dir, and the mounted external user home dir, end up in the same path? (so when trying to do anything on the vm user's home dir, it fails with read-only mount) [11:16:05] I see [11:19:19] probably we can just mount it in /mnt/home_dir or similar instead, to ensure we don't get issues [11:19:32] but I'd like to verify it first with dhinus [11:30:35] in a meeting, I'll check later! [11:39:12] np [11:54:56] * dcaro lunch [12:47:30] just remembered, today is a holiday in the US [14:16:58] I'm testing the mounts in lima-vm, it mounts my home dir to /Users/fran as read-only, but $HOME is set to /home/fran.linux which is read-write [14:17:27] and both env and .toolforge-lima-kilo are in /home/fran.linux [14:17:39] interesting [14:20:04] is anyone doing anything unusual on cloudvirt1060? it just alerted T355061 which seems to be a new alert, but possibly problematic regardless [14:20:05] T355061: MaxConnTrack Netfilter: Maximum number of allowed connection tracking entries alert on cloudvirt1060:9100 - https://phabricator.wikimedia.org/T355061 [14:20:28] dcaro: I'm confused by the fact that when I run "limactl shell bookworm" I end up in /Users/fran/wmf/lima-kilo, maybe limactl is saving that path somewhere [14:24:14] dhinus: if you are in a dir, and that dir is mounted to the lima vm, that's where you end up. So if you run the command from your home dir, you would end up there instead of inside of your lima-kilo folder [14:24:53] right. I wonder if ansible also runs in that dir, and that becomes the default dir in ansible for relative paths [14:25:59] hmm, this seems to be the case even if the dir is not mounted. How does that work? [14:26:11] do we need to mount ~ at all? I don't like that the VM can access all of my files (even if just read-only), do you use it for testing? [14:26:29] I removed the mount with "limactl edit bookworm" and now it defaults to /home/fran.linux [14:27:02] I don't think we need to mount it [14:27:17] I would remove the "mounts" section in bookworm.yaml then [14:27:47] maybe leave them but comment them out? It can still be nice for testing sometimes [14:28:17] the lima tmp dir might be needed [14:28:45] we can leave it with a comment yes, but anyway uncommenting is not enough, you need to delete the VM and recreate it or the change to the yaml won't be read [14:29:05] or you can use "limactl edit bookworm" to edit the yaml of an existing VM [14:29:20] which will show the commented lines, so it's handy to uncomment [14:29:51] yup, I meant comment it out in the template, then those who want to mount it still can do that easily [14:30:00] makes sense yes! [14:30:09] re: lima tmp dir, why is that needed? [14:30:57] * dhinus reads the docs [14:32:08] taavi: not me [14:32:54] blancadesal: I can't find any clear explanation in the docs, but I suspect it's just mounted as a way to quickly transfer a file from the VM to the host, which can be helpful so I'm ok with keeping it [14:33:38] dhinus: it's so you can run the ansible-playbook from inside the vm [14:33:56] we could copy the playbooks instead I guess (or mount only the repository folder) [14:35:01] right, the playbooks are coming from that dir! [14:35:58] mounting the repository folder would probably be more convenient, so you can edit a playbook without having to copy it over again [14:37:17] yep, you don't really need the whole of your home dir xd [14:38:15] unfortunately "mount" seems to require an absolute path, so you can not just mount "." https://github.com/lima-vm/lima/issues/893 [14:54:14] hmpf... yep, though that might be the case xd [14:55:59] brb [17:43:33] I've just stumbled on some sort of phab mirror [17:43:34] https://blickforce.com/?_=%2Fproject%2Fview%2F525%2F%23KJWqMdlUlBn7I%2BwXQhTjhp%2F6as93ElSopuY0J2bYYEUn