=== zz_natorious is now known as natorious === natorious is now known as zz_natorious === zz_natorious is now known as natorious === natorious is now known as zz_natorious === gamename is now known as gamename[away] === gamename[away] is now known as gamename [10:06] Hey all, I'm having an odd issue. Does Cloud-Init usually run as the root user or a sub-user? [10:07] I'm strace-ing a run and I see it downloading the chef omnibus installer into /tmp/ with permissions 0700, but then getting a permission denied on the script a second later and failing [10:20] Seth_Karlo: Hmm, that is strange; does the file end up with the expected permissions? [10:20] File is deleted before I can take a look at it [10:22] This is what I see: [pid 11304] execve("/tmp/tmpfpIQua/chef-omnibus-install", ["/tmp/tmpfpIQua/chef-omnibus-install"], [/* 21 vars */]) = -1 EACCES (Permission denied) [10:32] Seth_Karlo: What version of cloud-init are you using (and on what distro)? [10:32] cloud-init 0.7.5 on CentOS 7.103 [10:35] Seth_Karlo: So the code that is running is in cloudinit/config/cc_chef.py; if you add a sleep at around line 114 wherever that file is installed, then you'll have some time to check the file looks sensible. [10:36] Odd_Bloke: Understood, testing now [10:36] Seth_Karlo: "import time; time.sleep(N)", if you aren't a Python person. :) [10:38] Odd_Bloke: Perfect, thanks! [10:43] Odd_Bloke: Is N in seconds or ms? [10:43] Seth_Karlo: Seconds. [10:44] Odd_Bloke: Seemed to completely ignore it at line 114 in /usr/lib/python2.7/site-packages/cloudinit/config/cc_chef.py [10:45] Seth_Karlo: Can you pastebin your modified bit of the file? [10:45] Odd_Bloke: https://gist.github.com/Seth-Karlo/96e129c004e63d5cc331 [10:47] Seth_Karlo: Ah, put it a line up. [10:47] Seth_Karlo: util.subp is what's running the file. [10:47] Odd_Bloke: In between the util. ones? [10:48] Seth_Karlo: Yep. [10:48] Done, testing now :) [10:49] Odd_Bloke: Aaaaaah, I see the issue! [10:49] Odd_Bloke: My /tmp is mounted noexec! [10:51] Odd_Bloke: Now working, thank you very very much for your help. Kudos and my gratitude [10:52] Seth_Karlo: I'm always happy to help when it turns out there isn't a bug for me to fix. ;) [10:52] Odd_Bloke: Haha, I'll let you know if I find more! :P [10:52] :D [11:51] Claudiu Popa proposed stackforge/cloud-init: Add a draft spec for the parallel discovery of data sources https://review.openstack.org/220095 [11:54] Claudiu Popa proposed stackforge/cloud-init: Add a draft spec for the parallel discovery of data sources https://review.openstack.org/220095 === alexpilotti_ is now known as alexpilotti [14:34] claudiupopa: I think you might have done an incomplete naming change in the "FacadaParallelSearch" example in that spec. [14:34] claudiupopa: (Also Facada is a typo :p) [14:34] Oh, yep. :p [14:44] claudiupopa, are you going to tokyo ? [14:44] for openstack su mmit ? [14:44] Nope. [14:46] Odd_Bloke, didnt you do something wrt testing bin/cloud-init in 0.7 ? === zz_natorious is now known as natorious [14:47] smoser: I did, yeah. Why do you ask? [14:48] where is it ? [14:49] smoser: http://bazaar.launchpad.net/~cloud-init-dev/cloud-init/trunk/view/head:/tests/unittests/test_cli.py [14:49] thanks [14:49] :) [15:08] Odd_Bloke, do you happen to know... [15:09] http://paste.ubuntu.com/12263593/ [15:10] probably easier to read [15:10] http://paste.ubuntu.com/12263608/ [15:13] smoser: In a meeting right now, will look in a bit. :) === alexpilotti_ is now known as alexpilotti [15:27] hey smoser i sent you a response :D [15:27] j^2, thnks [15:27] we can sync later today? [15:27] sure. ping me here is fine [15:27] perfect [15:28] where'd you send rsponse ? [15:32] smoser: Given that those are all in the HTTP request, I would assume that they can all be logged. [15:32] Because the NSA has them anyway. :p [15:32] But I don't know that for sure. [15:32] thats what i thought too, Odd_Bloke [15:33] they're its supposed to work over http/ untrusted. [15:33] so you'd think it'd bad if there was secrets there [15:34] Yeah. [16:06] j^2, stuff for cloud-init 0.7.X is still on launchpad [16:06] so your review was at the correct place [16:15] ah, which is the one that shipped “in general” now-a-days? [16:16] j^2: 0.7.x is still shipped in general, 2.0 is still in early development. [16:16] Odd_Bloke: ah cool, then i’ll keep with 0.7.0 then [16:17] thanks [17:38] smoser whatever happened to enabling the launchpad + git stuff [17:38] is that possible still? [17:39] harlowja, yes. still possible. [17:39] +2 :) [17:39] can u press a button somewhere to make that happen ;) [17:40] ye-olde button [17:43] the button i'd rather push is 'make harlowja work enough on cloud-init 2.0 that he stops caring about 0.7' [17:43] do you have one of those buttons? [17:43] smoser ya, mainly this is for the y! CI system, that still pulls from 0.7, but really only knows how to interface with git :-P [17:53] it doesn't quite know bzr, and nobody seems willing to add bzr support :-P [20:01] claudiupopa in regard to https://review.openstack.org/#/c/220095/ let me know if u have any questions on how this got pulled off in taskflow (the parallel running based on dependencies ...) [20:01] because its awfully similar i think, ha [20:05] So they're similar? [20:05] * harlowja makes an example, but yes [20:05] Looking right now in the link. [20:06] k [20:09] claudiupopa http://paste.openstack.org/show/444643/ [20:09] output from running that [20:10] http://paste.openstack.org/show/444646/ [20:10] soooo it does something like u want i think, ha [20:10] and it can run in parallel as well :-P [20:10] u can run that by just cloning https://github.com/openstack/taskflow and making a venv and installing its requirements, then running it locally... [20:11] Nicee, why don't we use it for cloud-init v2 then? [20:11] could be done :-P [20:11] i am top maintainer/creator of that lib, so maybe we could, ha [20:11] but depends on others thoughts, ha [20:12] bb [20:12] The capabilities are represented by default_provides? [20:54] claudiupopa correct, or they could be [20:55] just something to think about, taskflow might be to big of a dependency, idk [22:04] smoser: So my HTTP friend tells me that in theory we should be able to store all the OAuth info in the log, but that assumes a good OAuth server implementation. [22:04] smoser: So we're probably better off not logging it. === natorious is now known as zz_natorious === zz_natorious is now known as natorious