[00:12] oh apart from the fact that you can't chattr a symlink [00:12] oh well [00:13] bummer. I liked that idea. [00:13] are you at a point that you could install auditd and use auditctl to add a file watch? [00:14] it also appears that there is no debug shell on tty2 during oem-config [00:14] which makes everything else harder [00:14] ow [00:14] i'm sure i fix that by mounting the image and screwing around in /lib/systemd though [00:15] +can [00:26] mwhudson: is_resolved_used() in netcfg-common.c is suspect, IMO. [00:27] char *path = realpath(RESOLV_FILE, NULL); [00:27] resolved_used = !strcmp(path, RESOLVED_FILE); [00:27] RESOLVED_FILE "/run/systemd/resolve/stub-resolv.conf" [00:28] yeah i saw some stuff in there that didn't quite make sense [00:28] Oh, no. That should work. [00:28] realpath() should resolve ../run to /run [00:30] mwhudson: I think it's something more subtle. Like maybe it's being put in static config mode halfway through. [00:30] mwhudson: In mine, I noticed that after DHCP, it then asked me for a nameserver (which is very suspect). Did you get the same there? [00:30] And I just hit enter, so then it had nothing to write to resolv.conf, thus giving me a blank one. [00:30] yeah istr that === amurray` is now known as amurray [00:38] er [00:38] Okay, I've stopped caring for today. [00:38] infinity: is the fact that /etc/resolv.conf is a _broken_ symlink when oem-config comes up likely to be relevant? [00:38] is it just that systemd-resolved isn't running, i wonder [00:39] mwhudson: Yeah, that's potentially relevant. [00:40] yeah that seems to have stopped it asking for a nameserver [00:41] (running systemctl start systemd-resolved in another tty) [00:41] Oh, is it just a unit ordering issue? [00:41] I imagine the oem-config systemd integration isn't heavily tested. [00:41] yeah [00:41] not so much an ordering as just nothing causing it to be started [00:42] afaict anyway [00:43] Well, yeah. Same thing. [00:43] As in, the target needs some more stuff in Wants, probably. [00:43] yeah [00:43] If that's all it is, easy fix. [00:49] infinity: https://code.launchpad.net/~mwhudson/ubiquity/+git/ubiquity/+merge/352792 ? [00:51] mwhudson: Tested and works? [00:51] infinity: no [00:51] infinity: lunch has a higher priority than that [00:52] infinity: but if it looks ok in spirit i can get onto that [00:52] mwhudson: Heh. I'm in the middle of a fresh test install ehre. I'll test it. [00:52] infinity: ok thanks! [00:52] * mwhudson lunches [01:12] oh ffs [01:16] quick go grab a second lunch [01:17] * tsimonq2 passes mwhudson a Lunch™ [01:19] heh [01:21] Oh FFS is right. Apparently, one shouldn't reboot after installing oem-config if you don't prepare first.. [01:24] Or maybe there's some other reason my boot is stalling... [01:24] Man, I hate computers. [01:24] Especially fake ones. [01:25] my xubuntu VM is very confused by virtualbox display settings [01:26] which apparently fails to set a display size so every time I log into it I get a System Error [01:26] I'm very confused by people using virtualbox. [01:26] well, when your employer is like "plz install this rootkit to connect to the corporate VPN" and you're like "no", it's an option [01:33] infinity: earlier on, i installed, ran oem-prepare, shut the instance down.. and then promptly deleted the image [01:33] infinity: are you getting anywhere or is it worth me trying to test too? [01:34] mwhudson: I'm redoing an install (nearly done) to figure out why my last one decided it didn't like booting anymore. [01:38] uh ubiquity git tip ftbfs [01:39] mwhudson: Excellent. [01:39] because of lint [01:39] the best kind of ftbfs [01:40] That moment when you've set your VM's login to ubuntu/ubuntu and it takes you FOUR FRIGGIN TRIES TO TYPE IT CORRECTLY. [01:41] mwhudson: Pastebin the linty sadness? [01:41] ./tests/run-pyflakes [01:41] bin/ubiquity:100: 'termios' imported but unused [01:41] bin/ubiquity:100: 'termios' imported but unused [01:41] man these ken thompson attacks are getting annoying.. time was you only got the extra prompts once [01:41] mwhudson: That's not tip. [01:41] mwhudson: Get current. [01:42] infinity: best is when you typo ubuntu twice the same way in the installer [01:42] ubnutu is how I usually (mis)type it. [01:42] In fact, the last commit to ubiquity is a fix of that. :P [01:42] ah wait [01:42] i'm on the bionic branch [01:42] Oh gross, was that committed to bionic too? [01:43] yeah and not reverted from there [01:43] * mwhudson has read git log now [01:44] infinity: I was going to say... :P [01:46] mwhudson: Confirmed that the fix is full of tasty fixy goodness. Merging. [01:46] infinity: thanks [01:49] infinity: how are ubiquity SRUs handled? [01:50] mwhudson: Expand that question? [01:50] infinity: well we want this fixed in bionic i presume [01:51] mwhudson: We do indeed. [01:51] do i need to do things to the bionic branch in git (like reverting that suspect commit?) or do i just get the source package from bionic, splat the fix in and upload it to unapproved? [01:51] and fiddle with the bug [01:52] mwhudson: Or submit an MP. [01:52] mwhudson: But I'm doing that revert first. [01:53] ah ok [01:54] mwhudson: There, bionic branch should look less poop now. [01:57] indeed [01:58] infinity: https://code.launchpad.net/~mwhudson/ubiquity/+git/ubiquity/+merge/352793 [01:59] mwhudson: Ta. [01:59] mwhudson: Was there some sense of OMG urgency when this card landed in your lap, or can we let it sit there while we roll up a few more fixes? [02:00] infinity: that sounds like a question for someone who is on vacation or someone who is sick [02:01] mwhudson: Right, I'm going to call it not super urgent, then. [02:01] infinity: seems fair [02:02] mwhudson: iz merged, though. [02:02] infinity: thanks === ledeni_ is now known as ledeni === ledeni_ is now known as ledeni === debfx_ is now known as debfx === s1aden is now known as sladen [12:41] is launchpad rendering weirdly/without images/formatting for others or is it just me? [12:42] seb128: Rollout issue, Colin's on the job. [12:42] k, thanks infinity [12:43] I forgot about the weird way that code and DB deployments both use the same intermediate tree so doing both at the same time results in bdaness [12:43] *badness [12:48] Ouch. === osx is now known as Toyo [12:58] LP fixed [13:00] cjwatson, thanks! [16:34] Hi. where does if_link.h come from? [16:35] stgraber@castiana:~$ find /usr/include/ | grep if_link.h [16:35] /usr/include/linux/if_link.h [16:35] stgraber@castiana:~$ dpkg -S /usr/include/linux/if_link.h [16:35] linux-libc-dev:amd64: /usr/include/linux/if_link.h [16:37] Thank you. [16:38] Can I just backport those headers or do I need to backport something else for them to work properly? [16:41] Nevermind, thought they were a source package on their own. [19:39] xnox, can libzstd be synced now? looks like the delta you added was only for xenial upgrades?