UbuntuIRC / 2014 /10 /15 /#cloud-init.txt
niansa
Initial commit
4aa5fce
=== daxroc_ is now known as daxroc
=== harlowja is now known as harlowja_away
=== Guest90815 is now known as rcj
[09:54] <flozano> any idea?
[09:54] <flozano> any hint about how to debug the issue?
[12:16] <smoser> flozano, example of the multipart?
[12:17] <flozano> I found the issue - it seems if the name of the attachment contains “/“s, it’s not properly executed
[12:17] <flozano> I made a clean-up and now it seems fime
[12:17] <flozano> *fine
[12:17] <smoser> if the filename has a '/' ?. thats possible. run-parts gets called on the directory, and wouldnt recurse.
=== Hazelesque_ is now known as Hazelesque
=== zz_gondoi is now known as gondoi
[14:25] <flozano> maybe it’d be good if cloud-init would clean-up these attachment names
[16:25] <lauris> hi, have a problem with puppet module
[16:25] <lauris> Oct 15 16:22:45 ip-10-0-110-91 cloud-init: 2014-10-15 16:22:45,280 - util.py[WARNING]: Running puppet (<module 'cloudinit.config.cc_puppet' from '/usr/lib/python2.7/site-packages/cloudinit/config/cc_puppet.pyc'>) failed
[16:41] <JayF> harlowja_away: https://review.openstack.org/#/c/99235 relevant to your interests
[16:41] <JayF> smoser: https://review.openstack.org/#/c/99235 possibly also relevant to you (ironic-spec for support for configdrive)
[16:42] <JayF> lauris: hey, if you look in /run/cloud-init/ you should have two json files which might have more information about the failure
[16:42] <smoser> lauris, there is probably a stack trace somewhere or information in /var/log/cloud-init-output.log
[16:42] <smoser> flozano, i'm not opposed to that. file a bug if you'd like. its not a unreasonable request.
[16:58] <lauris> JayF, there's no such folder /run/cloud-init
[16:58] <lauris> smoser, this is the only information I get from the log: http://pastie.org/9650295
[16:58] <lauris> not much of the details why it fails
[17:00] <smoser> you'll need console log i think. not sure what that is that i'm seeing output of.
[17:01] <smoser> in 0.7.2 command output (such as yum install) probably went to /dev/console.
[17:02] <smoser> so ec2-get-console-output might have it.
[17:02] <lauris> i can check the journal
[17:02] <lauris> i think it went there
[17:04] <lauris> no additional information there
[17:11] <smoser> i really do not understand how pip can be so prevelant.
[17:11] <smoser> and i even less understand how openstack c-i can rely upon it.
[17:11] <smoser> i see pip-timeout related failures in at least 1/3 devstack startups.
[17:12] <JayF> Openstack-infra runs a mirror
[17:13] <smoser> what garbage. i even looked at doing that, and *that* timed out over and over again.
[17:13] <smoser> but at least that indicates i was going down the right path.
[17:13] <smoser> :)
[17:13] <JayF> I mean, ask in -infra
[17:14] <JayF> I bet they'd be willing to tell you how they do theirs (or point you at code)
[17:14] <smoser> lauris, https://tickets.puppetlabs.com/si/jira.issueviews:issue-html/MODULES-1315/MODULES-1315.html maybe ?
[17:14] <smoser> JayF, oh, there is a general "pip mirror" code
[17:14] <JayF> I think they use something like bandersnatch?
[17:15] <JayF> b/c it used to be they'd only mirror requirements and requirements requirements
[17:15] <lauris> smoser, can't see how it is related?
[17:15] <lauris> it is cloud-init who fails
[17:15] <smoser> http://aboutsimon.com/2012/02/24/create-a-local-pypi-mirror/ thats what i looked at. seemed reasonable
=== harlowja_away is now known as harlowja
[17:15] <lauris> if I start puppet agent manually it works fine
[17:15] <JayF> but now they do the whole thing to support stackforge folks who are outside global-req
[17:15] <smoser> lauris, no. cloud-init registers failure of something else
[17:15] <smoser> that printed out to stderr:
[17:15] <smoser> Failed to issue method call: No such file or directory
[17:16] <lauris> so? that doesn't tell much about the problem
[17:16] <smoser> theres probably a fuller stacktrace in your log. and it is possible that cloud-init ran 'systemctl' , and that caused it. but i'm not sure.
[17:16] <lauris> if I run systemctl manually it works fine, both enable and start
[17:16] <harlowja> JayF thx, i'll forward that to some folks
[17:17] <JayF> harlowja: make sure to look at the referenced nova spec as well
[17:18] <JayF> harlowja: since this touches the virt driver
[17:18] <harlowja> yup yup
[17:19] <lauris> smoser, I checked console log as well, same information as i cloud-init.log
[17:20] <harlowja> btw, smoser and i updated https://etherpad.openstack.org/p/cloud-init-next with more goodies
[17:20] <harlowja> JayF harmw ^
[17:21] <lauris> smoser, it's fedora, by the way
[19:32] <harlowja> smoser https://code.launchpad.net/~harlowja/cloud-init/no-prettytable/+merge/238491 let me know what u think :-P
[19:32] <harlowja> example @ http://paste.ubuntu.com/8566656/
=== gondoi is now known as zz_gondoi