[01:52] jgrimm, here now [07:42] Hello all ! [08:33] Is there any way to "rsync" in cloud-init ? === shardy is now known as shardy_lunch === shardy_lunch is now known as shardy === shardy is now known as shardy_lunch === shardy_lunch is now known as shardy [17:21] Hey folks, I'm seeing cloud-init-local crash on boot when it tries to read /sys/class/net/eth0/carrier. It gets an EINVAL and blows up. It looks like reads of this file will always return EINVAL when the link is down, and since cloud-init-local runs *before* networking, the link is always going to be in that state. [17:22] Has anyone else seen this behavior? Simply running 'ip link set eth0 down' followed by 'cat /sys/class/net/eth0/carrier' will reproduce the problem. [18:19] On that topic: https://bugs.launchpad.net/cloud-init/+bug/1646919 [18:38] larsks, https://code.launchpad.net/~harlowja/cloud-init/+git/cloud-init/+merge/305882 [18:38] i think [18:39] Looks likely. [18:39] i marked yours a dupe. un-dupe if you like. [18:39] if you want to give harlowja's branch a kick, go for it. [18:39] supp [18:39] lol [18:40] i've just been kind of weary of pulling it as it seems like it could possibly go awry [18:40] my code is perfect [18:40] lol [18:40] that, and harlowja has not been responsive ;) [18:40] oops [18:40] smoser, agreed, looks like a dupe. [18:40] harlowja, kick. [18:41] kick [18:41] lol [18:41] soccer time [18:45] larsks ya, i'll fix it [18:45] \o/ [18:50] @smoser do u recall https://gist.github.com/harlowja/eb6f6688b3d6cefcaf302e588e034d9f [18:50] switching to building in a docker container [18:50] and hit that [18:51] on 0.7.7 [18:51] let me see if bumping up the cloud-init tag works/helps [18:51] that is fixed i think [18:51] k [18:52] harlowja, 9972d246947f1a6ec102b978b99b26acc43133ec [18:52] k [18:52] switching up to 0.7.8 [19:06] https://gist.github.com/harlowja/fa0ee4aa83acf8d7a35bf3b3742330e6 poop [19:06] lol [19:06] suck [19:07] harlowja, well, that is also fixed. [19:07] ya, prob, ha [19:07] seems like maybe we need an 0.7.9 for you, eh? [19:07] https://gist.github.com/harlowja/6c403b87249c84b914e3d60dc4d388a8 (my jenkinsfile that is running all this) [19:08] can we get a 0.7.9 release with all those fixes :-P [19:09] http://paste.ubuntu.com/23569017/ [19:09] woah [19:10] u the man [19:10] lol [19:10] let me see if master works, ha [19:11] please do [19:11] https://gist.github.com/harlowja/ed6dc735af22a491edf021690a43a49f [19:11] poopie [19:11] lol [19:11] two instances of new-style syntax :-P [19:11] do you create gists manually ? [19:11] ya [19:12] this is me copying from a jenkins console file [19:12] to a gist, lol [19:12] :-( [19:12] ok. [19:12] i can fix that for you quick. or you can and i'll pull [19:12] i can [19:12] i'd really like the other branch you have to come in [19:12] i got it covered boss [19:12] it scares me though.. [19:12] lol [19:12] the io errors one that larsks wants [19:12] its xmas, not halloween man [19:12] lol [19:12] don't be scared [19:13] be all santa like [19:13] lol [19:13] * smoser is blamed for that set notation [19:13] :-p [19:13] i didnt know [19:13] oh. thats 2.6 even. wow. [19:13] ya [19:14] :-( [19:25] harlowja, i'm fix and push the py26 [19:25] oh ok [19:25] http://paste.ubuntu.com/23569084/ [19:25] right ? [19:25] i was nearly there, ha [19:25] that passes in my cent6 container now [19:25] except i just got `fatal: remote error: DNS lookup failed: address 'xmlrpc.lp.internal' not found: [Errno -2] Name or service not known.` [19:25] is there something more ? [19:25] when git pushing [19:25] which was weird, ha [19:25] you work on the other one. [19:25] yeah, that is wierd. [19:25] :-( [19:26] ha [19:26] if you want to 'git show' [19:26] and paste [19:26] i can grab your credit too [19:26] my credit card is in there [19:26] gittttttt [19:26] lol [19:27] git config user.routing_number 482959929943 [19:27] bah. [19:27] did i type that in irc ? [19:27] whats the security code [19:27] lol [19:27] harlowja, in your sys-io-errors branch, should read_sys_net just have a on_einval parameter (since EINVAL is expected for down interfaces)? [19:27] larsks i'm gonna go with yes [19:28] :-P [19:28] ...because that would simplify a lot :) [19:29] smoser paste looks ok to me [19:29] u can do https://code.launchpad.net/~harlowja/cloud-init/+git/cloud-init/+merge/312390 also [19:29] back way when i didn't know those links expired, ha [19:29] oh well [19:30] oh. they do ? i didnt know that. [19:30] :-( [19:30] well or amazon changed it [19:30] so its dead anyway, lol [19:30] http://bit.ly/TyoUQs ---> nowhere now [19:34] ill hold off on thatone. ast here are other bit.ly broken links [19:34] k [19:34] i think its all just the actual links being dead [19:34] not bit.lky [19:34] ya, amazon [19:34] amazon fault [19:51] ok larsks smoser https://code.launchpad.net/~harlowja/cloud-init/+git/cloud-init/+merge/305882 updated [19:56] harlowja, lgtm. [19:58] larsks u are the person who could possibly say update the package for cloud-init on epel right :-P [19:58] i'm sure mdorman_ would like that :-P [19:58] *epel 7 and 6 :) [19:59] wink wink [19:59] harlowja, sort of. I'm working on an updated package in RHEL. [19:59] That should work it's way into CentOS. The one in EPEL should probably just go away. [19:59] But I will look into it. [20:00] u will get a cookie [20:19] harlowja, oops, your patch has a bug (line 239) [20:19] ...which shows up when you run the tests :) [20:36] ah [20:38] good catch, was running the test_net test, guess another one hit it or something [20:39] you might as well fix larsks's comment about the super set too. [20:39] names.extend(sorted(potential_interfaces)) [20:39] ... [20:39] for name in names: [20:39] Now I'm looking at a unicodedecodeerror exception. [20:39] if name not in potential_interfaces: [20:39] ... [20:40] k [20:40] done [21:08] harlowja, open to disagreement [21:08] you use _on_excp_false once [21:08] well, 3 times, one function [21:09] any reason not to have it namespaced inside that function [21:11] ?? [21:12] ah [21:12] i can namespace [21:13] http://paste.ubuntu.com/23569503/ [21:14] ya [21:14] updated with that [21:15] powersj, magicalChicken http://paste.ubuntu.com/23569508 [21:15] maybe because i'm in a container ? [21:16] or because i'm root ? [21:16] smoser: I think so [21:16] you can run as non-root [21:16] I do everything as my own user or jenkins user [21:16] well, i was in a container [21:16] so i was non-[real]root [21:16] ah yeah [21:16] what branch should i try ? [21:17] start with mine as it is what I am going to push to merge https://git.launchpad.net/~powersj/cloud-init/?h=integration-testing [21:17] k [21:20] smoser, https://code.launchpad.net/~larsks/cloud-init/+git/cloud-init/+merge/312399 fixes a unicodedecodeerror I was hitting on centos 7. [21:21] powersj, http://paste.ubuntu.com/23569531 [21:22] container is already running... that's a new one [21:22] still doing this in a container? [21:22] larsks, is there a specific reason for utf-8 there ? [21:22] rather than just encode() [21:23] No specific reason, other than "make everyting utf-8". What does .encode() with no encoding do? [21:23] I see, uses the default encoding. [21:23] Sounds good to me. I will update. [21:24] powersj, no, that was outside now [21:25] harlowja, if theres a bug, try to remember to [21:25] LP: #XXXX [21:25] in the commit [21:26] ah [21:26] smoser, does LP: do something magical? Or is that just informative? [21:26] and larsks for you, not Closes-bug [21:26] but LP: # [21:26] launchpad does read that now. [21:26] Ah, okay. [21:26] smoser: https://paste.ubuntu.com/23569555/ [21:27] https://help.launchpad.net/Code/Git <-- "Linking to bugs" [21:27] larsks, and also, i've got stuff that scrapes that info for ubuntu changelog entries [21:28] I've updated the merge request w/ LP: and .encode() with no explicit encoding. [21:28] powersj, i'm on 16.04 on that system. [21:28] lxc 2.05 [21:28] is that relevant ? [21:29] smoser: might be, torkoal is running 2.6.2 and I'm running 2.4.1 [21:29] I didn't consider that [21:29] actually that is the output of lxc --version... the package version on torkoal is 2.0.5.... [21:29] powersj, from ? [21:29] hm... odd [21:30] well, that'd probably stricktly be the client [21:30] maybe [21:30] i dont know [21:31] larsks, maybe SysConf should just convert to a string ? [21:32] wait. mayb not [21:32] i'm just trying to think of fallout . somone getting a string now when they expected bytes. [21:33] hm.. now im' confused. [21:33] smoser, I'm pretty sure this is a safe change. [21:33] we're calling load_file with no 'decode' [21:34] so it should get the decode=True [21:34] Ah, that sounds nicer. [21:34] I have to take off, but I will look at that later this evening. [21:34] which means it should already return a string [21:35] unless, larsks 0439d8a17d181a2546f2f7cb2d71a04bbb13b186 already fixed it for you ? [21:35] and yeah, i'm turning into a pumpkin now too [21:35] i'll leaev that in the MP [21:37] man python2 is confusing. [21:39] i'm confused.