[03:00] babbageclunk: 1 line PR? (plus small test) https://github.com/juju/juju/pull/11485 [03:00] wallyworld: yup yup [03:02] wallyworld: lgtm! [03:02] tyvm [03:12] do we have a convenient way of combining multiple errors into one? [03:12] superglue! [03:12] hmm, not sure [03:12] babbageclunk: not from memory [03:12] or gorilla tape [03:13] babbageclunk: the problem is the Cause is ambiguous [03:13] go to bed rick_h, you're drunk! [03:13] so I think using Annotatef is your best bet [03:13] babbageclunk: am I? crap when did that happen [03:13] chose one error, then have the other in the annotation [03:13] we've all been there! [03:14] hpidcock: yeah, but it's not just 2 (although I guess I could combine them pariwise) [03:14] pairwise [03:15] I think multiple errors is a symptoms of bad design anyway :P [03:15] I mean, I'm running a script on multiple machines and collecting the errors [03:15] babbageclunk: then they are results, not the error [03:16] sure... but I want to fail if any of them failed [03:16] and report the failure as an error [03:16] yeah well then you should combine them [03:16] into one new error [03:16] implement your own error type if introspection is needed [03:17] he didn't like what I had to say [03:17] sorry computer crashed [03:17] "yeah well then you should combine them [03:17] into one new error" [03:17] "implement your own error type if introspection is needed" [03:18] my computer didn't really crash [03:18] rage quit? [03:18] irc needs a /flounce [03:19] I guess string concatenation it is! Ah well [03:19] thanks hpidcock [03:20] babbageclunk: string concat makes sense unless you need to allow the caller to introspect each error [03:20] then it should be results [03:20] my 2c [03:20] yeah definitely, but in this case I just want to report it to the user. [03:21] to the CLI or just into a log? [03:23] cli [03:24] well if the user specified multiple entities, then the errors are results are they not? [03:24] the error would be "I failed to talk to juju api or something else" [03:25] the results are each of the individual operations [03:25] I'll shutup now leave it to you hah [03:31] This is in juju-restore, so there's no api involved. But the user doesn't specify the controller machines individually, they just say "restore this backup" and we say whether it worked or not. [05:29] tlm: hey, any issues with my comments on the PR? [05:29] haven't looked wallyworld pulling my hair out over something else [05:30] tlm seems to find the fun weird stuff [05:30] no worries, reach out if you need help [05:30] to do with the model operator stuff? [05:30] think I have got it thanks to hpidcock just throwing the tool box at it now [05:31] angry music helps [05:33] * tlm fixed his problem, drinks babbageclunk rum [05:35] ok some bash help please? why doesn't this give any output? `ssh ubuntu@10.77.116.85 -- sudo bash -c "echo hey"` [05:35] in exchange for my rum tlm [05:35] "hey" "hey" "hey" "hey" [05:35] um [05:35] works for me :-D [05:36] heeeeeeeeyaaaaaaaa [05:36] i am not good at the ssh foo, [05:36] works without the bash bit [05:37] Actually my problem is more that I'm not seeing stderr from the command if I run `sudo bash somescript` over ssh [05:37] hpidcock is a bash wizard [05:37] uhoh [05:38] I mean I'm reluctant to bug him because he's busy and also his advice is annoying [05:38] ;) [05:38] lmao [05:38] i know a guy I can ask babbageclunk [05:38] not hpidcock [05:38] but for reals I would appreciate it this time hpidcock not like before [05:39] just need to whack '' around your ssh command [05:40] * babbageclunk tries that [05:40] ssh ubuntu@10.77.116.85 'sudo bash -c "echo hey"' [05:40] yay, hpidcock is a genius! [05:40] totally works [05:40] thanks [05:40] I take back the mean thing I said just before [05:41] hah [05:43] ahhhhh - it's because without the quotes the bash command is just taking the command to be "echo", and then the next arg is $0 for the command [06:08] wallyworld: pushed up some changes, just taking the dogs for a run so they stop bugging me [06:22] ok :-) [06:31] tlm: reviewed, see if i make sense or you think i am on crack.... [09:10] stickupkid: found a small problem in my lxd PR (channel not propagated correctly when you install the snap instead of refreshing it). Will amend the last commit to address this [09:26] achilleasa, I've just popped out of a meeting, so I've got time to do the full PR test now [09:27] stickupkid: give me a few min to double-check something. [09:30] stickupkid: pushed the missing bit. btw, I just tried to use my PR to do nested-lxd on eoan host with lxd 3.0/stable (I see 3.0.3 installed in my bionic box) but still doesn't work [09:31] debug logs for bionic and eoan look more or less the same. There isn't actually any output after the "no profile changes required" message and the agent starting. I also compared the profiles in the host boxes to make sure they match [09:31] any idea where I should be looking next for the agent copy bits? [09:31] also manadart ^ [09:32] achilleasa, i'm in daily [09:32] (sidenote: downgrading to something like 3.0.3 prevents lxd from starting due to schema incompat) [12:12] stickupkid: is this what I am supposed to be seeing? https://pastebin.canonical.com/p/HjQKkXG84m/ [12:13] stickupkid: I still get apparmor's "are you running confined" logs in the nested container [12:50] stickupkid: one step closer: https://pastebin.canonical.com/p/pX8H7fntsM/ [12:51] this is after the unconfined patch [12:51] hope we don't have to remove apparmor via cloudinit... [12:53] there is a possibly related bug here: https://bugs.launchpad.net/snapcraft/+bug/1822372 [12:53] Bug #1822372: Snapcraft cleanbuild on Arch Linux with LXD - AppArmor permission error [13:34] stickupkid: https://pastebin.canonical.com/p/C6rwgyxgg4/. from juju edge snap on serverstack. - wonder if its related to my port security enabled bit? dunno [13:35] hml, but it should block the destroying of the controller, maybe I should just say there will be an orphaned set of os interfaces [13:35] stickupkid: the controller was destroyed [13:35] stickupkid: checking the interfaces [13:35] hml, that's how it should work [13:37] stickupkid: yup, i have an orphaner interface. thought we fixed this? [13:37] hml, you can't fix it if the server stack doesn't support it [13:52] achilleasa, ping [13:52] hml, get me? [13:53] stickupkid: yeah yeah. [13:53] hml, better error message maybe? [13:53] stickupkid: yes. especially if we want to let users know to remove them [13:53] we’ve done that a few times. [13:53] yes, I'll add a card [13:54] stickupkid: looks like we get a NotFound which is helpful [13:54] nice nice [13:57] stickupkid: pong [13:57] achilleasa, was in daily [15:05] stickupkid: got it working with nesting: https://pastebin.canonical.com/p/PV2MdNDgCn/ [15:05] (+unconfined) [15:05] achilleasa, in daily [15:55] stickupkid: thanks for the review. I will need to !!build!! until it gets green (having issues with windows box for the last few attempts)