Perl 6 - the future is here, just unevenly distributed

IRC log for #puppet-openstack, 2014-01-14

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
23:07 rwsu joined #puppet-openstack
23:10 dtalton2 joined #puppet-openstack
23:20 prad_ left #puppet-openstack
23:24 openstackgerrit joined #puppet-openstack
23:39 pzimmer1 left #puppet-openstack
00:04 hogepodge nice
00:04 hogepodge Thanks
00:14 tnoor joined #puppet-openstack
00:17 tnoor joined #puppet-openstack
00:23 tnoor1 joined #puppet-openstack
00:46 bodepd joined #puppet-openstack
01:37 xingchao joined #puppet-openstack
01:45 otherwiseguy joined #puppet-openstack
01:48 hogepodge joined #puppet-openstack
01:48 xingchao_ joined #puppet-openstack
01:51 rongze joined #puppet-openstack
01:59 rharrison joined #puppet-openstack
02:19 badiane_ka joined #puppet-openstack
02:27 xingchao joined #puppet-openstack
13:41 ilbot3 joined #puppet-openstack
13:41 topic for #puppet-openstack is now Place to collaborate on Puppet/OpenStack tools: logs at http://irclog.perlgeek.de/puppet-openstack/today
13:45 EmilienM mgagne: good morning, I'm reading https://review.openstack.org/#/c/63978/6/manifests/compute/libvirt.pp and I don't understand. I thought I had to remove deprecation support and remove the param instead.
13:46 EmilienM mgagne: when I read your previous review "Removing old configs is important" - maybe I misunderstood, if you could confirm me what I should do on this one =)
13:54 finnx joined #puppet-openstack
14:04 dprince joined #puppet-openstack
14:10 EmilienM fc__: also the ones of sbadia are good for review
14:52 djschaap joined #puppet-openstack
15:04 dmsimard joined #puppet-openstack
15:04 prad joined #puppet-openstack
15:14 pzimmer joined #puppet-openstack
15:15 otherwiseguy joined #puppet-openstack
15:45 openstackgerrit A change was merged to stackforge/puppet-ceph: correct option name in class ceph  https://review.openstack.org/66380
15:46 rongze joined #puppet-openstack
16:00 dtalton joined #puppet-openstack
16:18 mgagne EmilienM: configs = nova_config, not params.
16:19 EmilienM mgagne: good morning :-)
16:19 mgagne EmilienM: hi
16:20 mgagne EmilienM: I don't have a technical solution yet for what I had in mind. I therefore suggest to only deprecate the parameter and throw a warning.
16:21 ken_barber joined #puppet-openstack
16:22 mgagne EmilienM: Parameter libvirt_type is deprecated, use virt_type instead.
16:22 EmilienM mgagne: what I did at the first patchsets :(
16:22 mgagne EmilienM: yes but what we suggested is finding a way to remove the old nova_config to avoid leaving garbage
16:23 mgagne EmilienM: but we (I) don't have a technical solution yet for it.
16:23 mgagne EmilienM: without overloading the manifest with a bunch of ensure => absent
16:24 EmilienM pfiou what a mess
16:25 openstackgerrit François Charlier proposed a change to stackforge/puppet-ceilometer: Remove use of deprecated include_class matcher  https://review.openstack.org/66622
16:26 openstackgerrit François Charlier proposed a change to stackforge/puppet-ceilometer: Add API binding options  https://review.openstack.org/66624
16:26 dmsimard Oh, maybe fc__ has an opinion on my ceilometer middleware issue :)
16:27 fc__ dmsimard: yes ? what issue ?
16:28 dmsimard fc__: A swift proxy with the ceilometer middleware (as installed by puppet-ceilometer, puppet-swift) - swift does not have sufficient permissions in /etc/ceilometer and /var/log/ceilometer
16:28 dmsimard So it fails, horribly.
16:28 fc__ oh, yes … well, I don't think there's another choice than 1/add the swift user to the ceilometer group
16:29 EmilienM mgagne: ok thx. I'll work on it asap.
16:29 dmsimard fc__: swift::proxy::ceilometer already does that, but the folders are 750
16:29 fc__ and 2/ change the permissions on /etc/ceilometer & /var/log/ceilometer to be group readable and group writable
16:29 dmsimard fc__: So you're comfortable with 770 ?
16:29 fc__ dmsimard: at least for /var/log/ceilometer
16:30 fc__ is /etc/ceilometer required to be group writable ?
16:31 dmsimard fc__: I'm not sure, I need to do some more tests, I don't believe so .. but I had an issue untill I chown'd -R swift.swift /etc/ceilometer (for testing). The issue was "Jan 13 22:10:48 proxy01 proxy-server STDOUT: 2014-01-13 22:10:48 CRITICAL [ceilometer] coercing to Unicode: need string or buffer, NoneType found"
16:33 dmsimard I'll report back when I do some more tests
16:34 openstackgerrit François Charlier proposed a change to stackforge/puppet-ceilometer: Only add ceilometer user to nova group if using compute agent  https://review.openstack.org/66628
16:52 openstackgerrit A change was merged to stackforge/puppet-nova: Fix vncproxy package/service name on Debian  https://review.openstack.org/65828
16:54 openstackgerrit A change was merged to stackforge/puppet-nova: Support network_device_mtu of nova managed interfaces  https://review.openstack.org/63588
16:55 openstackgerrit A change was merged to stackforge/puppet-glance: Make sure swift backend in-file documentation matches actual code.  https://review.openstack.org/65527
16:57 openstackgerrit A change was merged to stackforge/puppet-ceilometer: Fix package ceilometer-alarm type error on Debian  https://review.openstack.org/63935
16:59 openstackgerrit A change was merged to stackforge/puppet-neutron: L3 schedulers support  https://review.openstack.org/63974
17:03 dtalton2 joined #puppet-openstack
17:05 openstackgerrit A change was merged to stackforge/puppet-cinder: Update GlusterFS driver  https://review.openstack.org/63979
17:06 dmsimard fc__: Ah, dang
17:07 dmsimard fc__: /var/log/ceilometer is ceilometer:adm
17:07 fc__ :-/
17:07 dmsimard fc__: Hence why adding swift to ceilometer group isn't enough
17:07 fc__ dmsimard: on what distro ?
17:07 dmsimard Ubuntu
17:07 dmsimard 12.04
17:07 openstackgerrit A change was merged to stackforge/puppet-swift: Remove redundant user ownership declaration and ensure spec test coherency  https://review.openstack.org/65560
17:07 dmsimard and swift-proxy-server.log is root:root
17:08 rmoe joined #puppet-openstack
17:09 * fc__ off: time to go fetch the kids already
17:12 mgagne anyone avoid issues using qemu64 cpu model on intel processors?
17:13 mgagne looks like libvirt had the great idea to add svm flag to qemu64, this flag is only supported by amd :-/
17:28 angdraug joined #puppet-openstack
17:35 dmsimard fc__: problem looks a bit more complex than I initially believed - it looks like it could be a bug
17:36 dmsimard fc__: trying to dig deeper
17:39 csschwe joined #puppet-openstack
17:55 dmsimard fc__: I added some information to an existing bug.. https://bugs.launchpad.net/ceilometer/+bug/1262264. One of your colleagues tagged it as "won't fix" :) Maybe you or EmilienM can also have a look.
17:58 dmsimard Ah, oops, Julien isn't at enovance anymore :)
18:01 xarses joined #puppet-openstack
18:03 pzimmer joined #puppet-openstack
18:04 bauzas joined #puppet-openstack
18:04 pzimmer1 joined #puppet-openstack
18:05 fc__ dmsimard: does swift-proxy-server start when executed 'manually' as swift user from su or sudo ? if yes, it might be a problem with the init/upstart script
18:10 rongze joined #puppet-openstack
18:18 dmsimard fc__: hmm, good point, let me see
18:21 dmsimard fc__: yup, starting /usr/bin/swift-init proxy-server start as swift works.
18:23 sputnik13net joined #puppet-openstack
18:25 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-neutron: Isolated network support  https://review.openstack.org/63952
18:25 EmilienM this patch was ready to be merged and failed because of conflict. When Jenkins +1, I'll merge it by myself
18:26 EmilienM oops I did wrong in the rebase. Let's do it again
18:31 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-neutron: Isolated network support  https://review.openstack.org/63952
18:32 openstackgerrit Emilien Macchi proposed a change to stackforge/puppet-neutron: Isolated network support  https://review.openstack.org/63952
18:37 openstackgerrit A change was merged to stackforge/puppet-neutron: Isolated network support  https://review.openstack.org/63952
18:40 hogepodge joined #puppet-openstack
18:49 hogepodge joined #puppet-openstack
18:51 dmsimard fc__: I narrowed down the issue. I don't understand why the error message changes when I change the /etc/ceilometer permissions (this is what lead me in the wrong direction..). The issue is really about /var/log/ceilometer, /etc/ceilometer is fine
19:21 hogepodge ping fc__
19:21 fc__ hogepodge: _o/
19:22 hogepodge re https://github.com/puppetlabs/puppetlabs-mongodb/pull/82
19:22 fc__ hogepodge: i was just leaving (have to feed the kids)
19:22 hogepodge Ah, ok, we can talk later. Quick thing, though.
19:22 hogepodge The copyright is going to be a problem. :-/
19:23 fc__ ok, I think I'll be able to remove the copyright thing
19:23 hogepodge Thanks. :-)
19:23 hogepodge I'll probably try to figure out a way to system test it, so no need to worry about that.
19:23 bauzas joined #puppet-openstack
19:24 hogepodge Have fun with the kiddos!
19:24 fc__ thx !
19:59 ianw joined #puppet-openstack
20:26 xarses joined #puppet-openstack
20:46 sputnik13net joined #puppet-openstack
20:50 dmsimard hogepodge: poke
21:13 radmiraal left #puppet-openstack
21:26 hogepodge dmsimard: pike
21:26 hogepodge poke
21:28 dmsimard hogepodge: Been trying to do something as clean as possible for puppet-swift today - with the ceilometer middleware. Related bug is https://bugs.launchpad.net/ceilometer/+bug/1262264
21:28 dmsimard hogepodge: tldr of that bug is that the puppet-ceilometer and puppet-swift (with ceilometer middleware) in their current state do not deliver something working out of the box
21:29 hogepodge That's good, because both are still rc. :-)
21:29 hogepodge I'm running into ceilometer problems for nova too. I may just be configuring it wrong.
21:29 hogepodge Thanks for the heads up.
21:30 dmsimard hogepodge: Yeah hence the poke. I really want to do this as clean as possible but the reality is that we'd need to 755 /etc/ceilometer and 644 /etc/ceilometer/ceilometer.conf as part of ::proxy::ceilometer AND do something about /var/log/ceilometer as well
21:31 dmsimard hogepodge: A bit sad that the official response to that bug is that "it is a deployment problem"
21:33 hogepodge Yeah. As I'm working on my deployment I'll look in to it. The answer might be to say that for Havana Ceilometer is a tech preview. How is core handling it? What's in their deployment docs?
21:33 hogepodge We should try to follow those. Maybe a config variable to allow for more permissive permissions.
21:36 dmsimard hogepodge: Official installation instructions say that swift should have write access to /var/log/ceilometer
21:36 dmsimard hogepodge: That's it, though, it doesn't mention anything about ceilometer.conf but it's obvious that it needs access to it.
21:37 hogepodge Then that's what we should do for the module. Option to add swift to ceilometer group, then make /var/log/ceilometer 0764
21:38 dmsimard hogepodge: ::proxy::ceilometer already adds swift to the ceilometer group, it's not sufficient (that's the bug I cited above)
21:38 mgagne hogepodge: nothing tells you there won't be any secret data in /var/log/ceilometer
21:38 hogepodge 0760 then?
21:38 mgagne hogepodge: I would much better like /var/log/ceilometer to be owned by ceilometer.ceilometer instead of ceilometer.amd than open up /var/log/ceilometer to everyone
21:38 hogepodge for files in the directory.
21:38 dmsimard hogepodge: A possible solution (compromise?) I saw is that instead of logging to /dev/log for swift by defaut, log to /var/log/swift instead and have ::proxy::ceilometer change the log destination in ceilometer.conf to /var/log/swift
21:39 hogepodge Yeah, that's another option.
21:39 mgagne hogepodge: other solution would be to move ceilometer.log to /var/log/swift when used on swift-proxy
21:39 mgagne dmsimard: ^
21:39 dmsimard mgagne: beat you to that one :)
21:39 mgagne dmsimard: my thinking is too loud
21:40 dmsimard I think moving everything to /var/log/swift would make sense.
21:40 mgagne dmsimard: logrotate would need to be configured too
21:40 hogepodge Documenting it and throwing in a notify would help prevent surprises.
21:40 dmsimard mgagne: Yeah. Not like ceilometer adds a logrotate for /var/log/ceilometer currently, though
21:40 mgagne dmsimard: shhhhh
21:41 hogepodge brb
21:41 dmsimard mgagne: Hey come on, I'm expecting stuff to be working fine :P
21:41 mgagne dmsimard: or we create a swift folder in /var/log/ceilometer with appropriate permissions
21:43 mgagne dmsimard: we should ask #openstack-metering or #openstack-swift about it
21:44 fc__ just a quick question: could it be possible that the issue has something to do with the way the service is launched or some apparmor/selinux configuration ?
21:45 * fc__ back to family
21:45 dmsimard mgagne: Asked #openstack-ceilometer, didn't get a response
21:45 fc__ anyway I'll try to reproduce this setup tomorrow on some Debian & Ubuntu VMs
21:45 sanman joined #puppet-openstack
21:46 dmsimard fc__: Thanks for your time :)
22:01 xarses joined #puppet-openstack
22:33 xarses joined #puppet-openstack
22:41 rharrison joined #puppet-openstack
22:55 hogepodge joined #puppet-openstack
22:58 openstackgerrit Donald Talton proposed a change to stackforge/puppet_openstack_builder: Fixed bad args and bad chars in puppet master command  https://review.openstack.org/66726

| Channels | #puppet-openstack index | Today | | Search | Google Search | Plain-Text | summary