| *** sdake_ has quit IRC | 00:00 | |
| *** nijaba_ has joined #tripleo | 00:10 | |
| *** nijaba_ has quit IRC | 00:10 | |
| *** nijaba_ has joined #tripleo | 00:10 | |
| *** nijaba has quit IRC | 00:10 | |
| *** chlong has joined #tripleo | 00:25 | |
| *** gchamoul- has quit IRC | 00:44 | |
| *** cwolferh has quit IRC | 00:50 | |
| *** lazy_prince has quit IRC | 01:13 | |
| *** killer_prince has joined #tripleo | 01:14 | |
| *** killer_prince is now known as lazy_prince | 01:14 | |
| *** pradeep has joined #tripleo | 01:16 | |
| *** Marga_ has quit IRC | 01:19 | |
| *** Marga_ has joined #tripleo | 01:20 | |
| *** stendulker has joined #tripleo | 01:23 | |
| *** yamahata has quit IRC | 01:28 | |
| SpamapS | looks like CI is toast | 01:37 |
|---|---|---|
| SpamapS | Ironic issues still going on | 01:37 |
| *** stendulker has quit IRC | 01:46 | |
| mordred | greghaynes, SpamapS: ^^ _WOW_ | 01:46 |
| greghaynes | hai | 01:47 |
| greghaynes | so, that bug is amazing, but its also not the cause of our friend colonel panic AFAICT | 01:47 |
| *** adrianopetrich has quit IRC | 01:48 | |
| *** adrianopetrich has joined #tripleo | 01:52 | |
| *** lazy_prince has quit IRC | 02:06 | |
| *** pradeep has quit IRC | 02:07 | |
| *** Marga_ has quit IRC | 02:26 | |
| *** Marga_ has joined #tripleo | 02:26 | |
| *** adrianopetrich has quit IRC | 02:28 | |
| *** Marga_ has quit IRC | 02:31 | |
| *** adrianopetrich has joined #tripleo | 02:31 | |
| *** yamahata has joined #tripleo | 02:40 | |
| *** noslzzp has joined #tripleo | 02:50 | |
| *** untriaged-bot has joined #tripleo | 03:00 | |
| untriaged-bot | No untriaged bugs so far! \o/ | 03:00 |
| *** untriaged-bot has quit IRC | 03:00 | |
| *** noslzzp has quit IRC | 03:03 | |
| *** pradeep1 has joined #tripleo | 03:50 | |
| *** yamahata has quit IRC | 03:55 | |
| *** yamahata has joined #tripleo | 03:55 | |
| *** pradeep has joined #tripleo | 04:03 | |
| *** pradeep1 has quit IRC | 04:03 | |
| openstackgerrit | Martin André proposed openstack/diskimage-builder: Fix init system detection on Docker images https://review.openstack.org/152785 | 04:07 |
| *** gchamoul has joined #tripleo | 04:10 | |
| *** gchamoul has quit IRC | 04:15 | |
| *** gchamoul has joined #tripleo | 04:16 | |
| *** masco has joined #tripleo | 04:23 | |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Add functional smoke test for disk-image-create https://review.openstack.org/169167 | 04:24 |
| *** tzumainn has quit IRC | 04:29 | |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Fix image size to fit filesystem journal https://review.openstack.org/169166 | 04:30 |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Add functional smoke test for disk-image-create https://review.openstack.org/169167 | 04:31 |
| openstackgerrit | greghaynes proposed openstack/diskimage-builder: Support building ACIs https://review.openstack.org/155484 | 04:31 |
| *** gchamoul has quit IRC | 04:41 | |
| *** gchamoul has joined #tripleo | 04:42 | |
| *** noslzzp has joined #tripleo | 04:44 | |
| *** noslzzp has quit IRC | 04:46 | |
| *** chlong has quit IRC | 05:09 | |
| *** pradeep has quit IRC | 05:17 | |
| *** pradeep has joined #tripleo | 05:20 | |
| *** hewbrocc` has joined #tripleo | 05:24 | |
| *** crinkle has quit IRC | 05:24 | |
| *** adrianopetrich has quit IRC | 05:24 | |
| *** crinkle has joined #tripleo | 05:25 | |
| *** hewbrocca has quit IRC | 05:26 | |
| *** adrianopetrich has joined #tripleo | 05:27 | |
| *** jcoufal has joined #tripleo | 05:59 | |
| *** spredzy|afk is now known as spredzy | 06:01 | |
| *** adrianopetrich has quit IRC | 06:02 | |
| *** adrianopetrich has joined #tripleo | 06:04 | |
| *** yamahata has quit IRC | 06:16 | |
| *** mrunge has joined #tripleo | 06:18 | |
| *** mrunge has quit IRC | 06:18 | |
| *** mrunge has joined #tripleo | 06:18 | |
| *** yamahata has joined #tripleo | 06:20 | |
| *** sdake_ has joined #tripleo | 06:32 | |
| *** yog__ has joined #tripleo | 06:43 | |
| *** dasm|afk is now known as dasm | 06:48 | |
| *** yamahata has quit IRC | 06:48 | |
| *** jprovazn has joined #tripleo | 06:50 | |
| *** mmagr has joined #tripleo | 07:04 | |
| *** ukalifon1 has joined #tripleo | 07:04 | |
| *** gchamoul_ has joined #tripleo | 07:05 | |
| *** sdake_ has quit IRC | 07:05 | |
| *** maufart has joined #tripleo | 07:07 | |
| *** gchamoul_ has quit IRC | 07:09 | |
| *** gchamoul has quit IRC | 07:12 | |
| *** jistr has joined #tripleo | 07:15 | |
| *** gchamoul has joined #tripleo | 07:16 | |
| *** dtantsur|afk is now known as dtantsur | 07:17 | |
| *** sdake has joined #tripleo | 07:18 | |
| *** gchamoul has quit IRC | 07:19 | |
| *** sdake_ has joined #tripleo | 07:25 | |
| *** sdake has quit IRC | 07:26 | |
| *** sdake_ has quit IRC | 07:30 | |
| *** pradeep2 has joined #tripleo | 07:32 | |
| *** pradeep has quit IRC | 07:32 | |
| *** hewbrocc` has quit IRC | 07:38 | |
| *** gfidente has joined #tripleo | 07:45 | |
| *** sdake has joined #tripleo | 07:52 | |
| *** jprovazn has quit IRC | 07:55 | |
| *** gchamoul has joined #tripleo | 07:57 | |
| *** jprovazn has joined #tripleo | 07:59 | |
| *** gchamoul has quit IRC | 08:01 | |
| openstackgerrit | Yanis Guenane proposed stackforge/puppet-tripleo: Enable access to HAProxy stats page https://review.openstack.org/171957 | 08:01 |
| *** gchamoul has joined #tripleo | 08:05 | |
| *** pblaho has joined #tripleo | 08:10 | |
| *** derekh has joined #tripleo | 08:13 | |
| gfidente | so CI is blocked on https://bugs.launchpad.net/tripleo/+bug/1441875 | 08:14 |
| openstack | Launchpad bug 1441875 in tripleo "Neutron fails to start in the seed with a PermissionDeniedError" [Critical,Triaged] | 08:14 |
| gfidente | I am trying to reproduce locally to see if I can get to bottom of it | 08:15 |
| derekh | gfidente: ok, give me a shout if you hit a brick wall and I'll take a look aswell | 08:23 |
| *** lucasagomes has joined #tripleo | 08:24 | |
| gfidente | derekh, I am not entirely sure how to reproduce seed building steps with same params defined in CI | 08:25 |
| gfidente | derekh, is everything I need defined http://git.openstack.org/cgit/openstack-infra/tripleo-ci/tree/toci_gate_test.sh ? | 08:26 |
| *** mrunge has quit IRC | 08:27 | |
| *** athomas has joined #tripleo | 08:31 | |
| derekh | gfidente: yup, thats pretty much them, the other ones that are defined before that script is run are here http://git.openstack.org/cgit/openstack-infra/project-config/tree/jenkins/jobs/tripleo.yaml | 08:33 |
| derekh | gfidente: but they shouldn't change things much for this | 08:33 |
| gfidente | yep, thanks | 08:34 |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration https://review.openstack.org/167538 | 08:36 |
| *** rwsu has quit IRC | 08:41 | |
| *** stendulker has joined #tripleo | 08:42 | |
| jistr | gfidente: trying to reproduce that as well as it's blocking any work i could do | 08:46 |
| jistr | gfidente: i think i hit a different error previously with seed, now rebuilding on a clean machine | 08:46 |
| gfidente | yeah I have to rebuild as well | 08:47 |
| openstackgerrit | Yanis Guenane proposed stackforge/puppet-tripleo: Loadbalancer: Add support for Redis https://review.openstack.org/171121 | 08:50 |
| *** jtomasek has joined #tripleo | 08:53 | |
| openstackgerrit | Yanis Guenane proposed stackforge/puppet-tripleo: Loadbalancer: Add support for Redis https://review.openstack.org/171121 | 08:56 |
| *** hewbrocca has joined #tripleo | 08:58 | |
| *** dtantsur is now known as dtantsur|brb | 08:58 | |
| *** untriaged-bot has joined #tripleo | 09:00 | |
| untriaged-bot | No untriaged bugs so far! \o/ | 09:00 |
| *** untriaged-bot has quit IRC | 09:00 | |
| mark-os | ping jistr | 09:14 |
| jistr | mark-os: pong, hello | 09:14 |
| mark-os | jistr: Hello, sorry to bother you again today | 09:14 |
| mark-os | jistr: you got me to curl the dashboard yesterday to try and solve my ConnectionError problem | 09:15 |
| *** pcaruana has joined #tripleo | 09:17 | |
| jistr | mark-os: btw i don't think i understood your problem yesterday... you said you couldn't get "tuskar-api --config-file etc/tuskar/tuskar.conf" command to work, but then we successfully curled the API, so i'm wondering what is the issue you have? | 09:19 |
| jistr | (we were curling tuskar-api, not dashboard) | 09:19 |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration https://review.openstack.org/167538 | 09:23 |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration https://review.openstack.org/167538 | 09:30 |
| *** Goneri has joined #tripleo | 09:39 | |
| mark-os | jistr: I can connect to the dashboard, I thought there was an error with "tuskar-api --config-file etc/tuskar/tuskar.conf" but I think you told me it was fine | 09:41 |
| *** lynxman has quit IRC | 09:43 | |
| *** lynxman has joined #tripleo | 09:44 | |
| * jistr -> lunch, biab | 09:44 | |
| *** sdake_ has joined #tripleo | 09:46 | |
| *** sdake_ has quit IRC | 09:49 | |
| *** sdake has quit IRC | 09:50 | |
| *** dtantsur|brb is now known as dtantsur | 09:52 | |
| *** pelix has joined #tripleo | 09:55 | |
| jprovazn | is there a simple way how to add (register) more nodes/VMs once UC is deployed? | 10:00 |
| *** bvandenh has joined #tripleo | 10:00 | |
| jprovazn | dtantsur: ^ maybe you might know? | 10:00 |
| dasm | jprovazn: you can do this via tuskar-ui | 10:01 |
| dasm | but if you're not planning to install horizon, etc. you should use ironic | 10:01 |
| openstackgerrit | Derek Higgins proposed openstack-infra/tripleo-ci: Revert "Switch Delorean jobs to F21" https://review.openstack.org/171989 | 10:03 |
| *** jcoufal has quit IRC | 10:04 | |
| jprovazn | dasm: thanks, yes I think I need a combination of create-nodes+instack-ironic-deployment calls, though it seems that the json file used to pass all the info around complicates this :/ | 10:06 |
| StevenK | register-nodes, not create-nodes | 10:06 |
| StevenK | create-nodes actually calls libvirt to create VMs | 10:06 |
| jprovazn | StevenK: yep, and I need/want create them too (which I did not mention) | 10:07 |
| *** akrivoka has joined #tripleo | 10:08 | |
| StevenK | jprovazn: create-nodes should have created 15, with the default params | 10:08 |
| jprovazn | StevenK: yes, but instack defaults to 2 | 10:09 |
| jprovazn | or to exact count you set | 10:09 |
| openstackgerrit | Marios Andreou proposed openstack/tuskar: Wires up constraints and type into returned plan parameters https://review.openstack.org/171713 | 10:20 |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration https://review.openstack.org/167538 | 10:25 |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Enable use of coordination_url in ceilometer https://review.openstack.org/171996 | 10:25 |
| *** jcoufal has joined #tripleo | 10:29 | |
| *** jcoufal has quit IRC | 10:36 | |
| derekh | gfidente: I guess the problem is in a change to the neutron packaging | 10:38 |
| derekh | [root@localhost ~]# ls -l /usr/share/neutron/neutron-dist.conf /etc/neutron/neutron.conf | 10:38 |
| derekh | -rw-r-----. 1 root root 1322 Apr 9 10:36 /etc/neutron/neutron.conf | 10:38 |
| derekh | -rw-r-----. 1 root root 469 Apr 9 01:26 /usr/share/neutron/neutron-dist.conf | 10:38 |
| derekh | neutron can't read those files | 10:38 |
| *** bvandenh has quit IRC | 10:40 | |
| *** bvandenh has joined #tripleo | 10:44 | |
| *** hewbrocca has quit IRC | 10:45 | |
| derekh | gfidente: the neutron package maintainer is looking into fixing the file permissions in the package | 10:55 |
| gfidente | derekh, thanks, I still haven't got a fresh seed to try it out but comparing to the OC nodes, I see those are supposed to be 0600 but root:neutron owned | 10:57 |
| gfidente | actually 0640 root:neutron | 10:57 |
| derekh | gfidente: yup | 10:58 |
| gfidente | thanks! | 10:58 |
| jistr | derekh: thanks | 10:58 |
| gfidente | shall we update the bug with the news? | 10:58 |
| derekh | gfidente: jistr: no prob | 10:58 |
| derekh | gfidente: will do | 10:58 |
| jistr | do we take that package from RDO directly? i see we have it forked in rdo-management but are we actually using that? https://github.com/rdo-management/neutron-packaging | 10:59 |
| jistr | ATM i don't understand the web of deloreans and rpm spec forks, i wonder if we have some docs on that | 11:02 |
| derekh | jistr: we're pulling in the last successfully buit master package, the packaging is here https://github.com/openstack-packages/neutron | 11:03 |
| *** hewbrocca has joined #tripleo | 11:04 | |
| derekh | jistr: this is what will go into RDO at milestone/release time | 11:04 |
| jistr | derekh: thanks for the info :) | 11:05 |
| jistr | mark-os: can you please describe the symptoms of the problem you're hitting again? i think i still don't follow what's the actual problem. You mentioned ConnectionError -- where is that printed? | 11:05 |
| derekh | jistr: the rdo/delorean relationship is explained here https://www.rdoproject.org/packaging/rdo-packaging.html | 11:05 |
| derekh | jistr: rdo-management is a fork that was created outside of that | 11:06 |
| mark-os | jistr: So I issue this command: tools/with_venv.sh ./manage.py runserver 0.0.0.0:8080 | 11:09 |
| mark-os | jistr: follow this I go to the address of the dashboard which shows me the login screen, I login and then I'm given this ConnectionError | 11:09 |
| jistr | ah ok... unfortunately i don't know much about Horizon and its tooling, but it could be a more general issue. Is there some error message or backtrace printed or something? Maybe in console where you run the runserver command? | 11:12 |
| jistr | mark-os: ^ | 11:12 |
| *** rhallisey has joined #tripleo | 11:13 | |
| mark-os | jistr: here's a paste of the error http://paste.openstack.org/show/lBUcVbWWKl1RAKtypstl/ | 11:14 |
| *** rbrady has quit IRC | 11:14 | |
| *** mmagr has quit IRC | 11:16 | |
| *** subscope has quit IRC | 11:17 | |
| jistr | mark-os: ok, i wonder if this could be related to a change of URL prefixing mechanism in tuskarclient some while ago... | 11:19 |
| *** subscope has joined #tripleo | 11:20 | |
| jistr | mark-os: probably not, but just for completeness: https://github.com/openstack/python-tuskarclient/commit/284c47b0c8582dd8d5fb2e8d7974513d88d9eb40 so if the package you're using is newer than that change, i think you'll need to make sure that the endpoint URL in Keystone for the Tuskar service has a "/v2/" at the end of it. But i think if this was the problem, it would give you 404 and not connection refused... | 11:26 |
| *** lucasagomes is now known as lucas-hungry | 11:27 | |
| jistr | maybe the Tuskar endpoint in Keystone points to something totally wrong then, or maybe isn't there, so that might be a thing to check | 11:28 |
| *** mmagr has joined #tripleo | 11:29 | |
| dasm | jistr: it doesn't seem to be a problem with tuskarclient, because urls look good. | 11:29 |
| dasm | mark-os: quick test: run tuskar plan-list | 11:29 |
| dasm | mark-os: this test run on server with horizon | 11:31 |
| dasm | jistr: misread paths as urls -- shame on me. | 11:33 |
| dasm | but question to mark-os is still valid | 11:33 |
| *** nijaba_ has quit IRC | 11:36 | |
| *** bvandenh has quit IRC | 11:37 | |
| *** bvandenh has joined #tripleo | 11:39 | |
| *** nijaba has joined #tripleo | 11:42 | |
| *** nijaba has quit IRC | 11:42 | |
| *** nijaba has joined #tripleo | 11:42 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Pass in libvirt_rbd_secret_key for nova compute https://review.openstack.org/170518 | 11:43 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Add support for configuration of Glance RBD store https://review.openstack.org/172016 | 11:43 |
| *** stendulker has quit IRC | 11:44 | |
| gfidente | jistr, ^^ that was a rebase because the other depends on it | 11:47 |
| *** lucas-hungry is now known as lucasagomes | 11:48 | |
| jistr | gfidente: ack :) | 11:49 |
| *** bvandenh has quit IRC | 11:53 | |
| *** jdob has joined #tripleo | 11:55 | |
| *** masco has quit IRC | 12:04 | |
| *** mmagr has quit IRC | 12:05 | |
| *** rbrady has joined #tripleo | 12:06 | |
| *** dprince has joined #tripleo | 12:07 | |
| *** ukalifon1 has quit IRC | 12:14 | |
| *** morazi has joined #tripleo | 12:17 | |
| *** tzumainn has joined #tripleo | 12:19 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Add support for configuration of Glance RBD store https://review.openstack.org/172016 | 12:20 |
| *** pradeep2 has quit IRC | 12:21 | |
| *** adrianopetrich has quit IRC | 12:24 | |
| *** adrianopetrich has joined #tripleo | 12:27 | |
| dprince | derekh, should we just ping to the earlier Delorean from Monday/Tuesday (which passed CI) | 12:28 |
| dprince | derekh: pin, not ping | 12:28 |
| derekh | dprince: mightn't be a bad idea while the neutron package is fixed, let me see if there is a patch up to fix it yet | 12:29 |
| derekh | dprince: https://review.gerrithub.io/#/c/229858/2 | 12:30 |
| dprince | derekh: Still, I think we should seriously consider always pinning Delorean | 12:31 |
| derekh | dprince: yup | 12:31 |
| derekh | dprince: or use a symbolic link that is changed by CI ... | 12:32 |
| dprince | derekh: exactly, once RDO CI implements a "Stable" symlink we could use it instead | 12:32 |
| derekh | dprince: ack, | 12:34 |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Add support for Glance RBD backend https://review.openstack.org/172032 | 12:43 |
| dprince | gfidente: there is my patch :) | 12:43 |
| mark-os | jistr: I've got the /v2/ at the end of my endpoint url | 12:43 |
| dprince | gfidente: I took a slightly different approach although there are some similarities with regards to the extra ceph configuration | 12:43 |
| mark-os | dasm: $ tuskar plan-list | 12:43 |
| mark-os | dasm: is that the command you want me to run? | 12:44 |
| dasm | yes | 12:44 |
| dasm | mark-os: ^ | 12:44 |
| mark-os | dasm: not recognise | 12:44 |
| mark-os | dasm: actually just ran it through venv | 12:44 |
| mark-os | dasm: going to test it out | 12:45 |
| dasm | if you have tuskar-api running, look at logs and check if there are double /v2's | 12:45 |
| dprince | gfidente: the biggest difference is the fact that I'm using GlanceBackend as a single parameter to control all the backends (swift, file, rbd). This seemed to work better with puppet-glance because of the default_store settings http://git.openstack.org/cgit/stackforge/puppet-glance/tree/manifests/backend/rbd.pp#n45 | 12:45 |
| dasm | mark-os: right now I have: 192.168.122.1 - - [09/Apr/2015 12:45:49] "GET /v2/plans HTTP/1.1" 200 27412 │+--------------------------------------+-----------+-------------+---------------------+ | 12:45 |
| dprince | gfidente: functionally it seems to work fine, although when I tested it locally I hit something similar to this: https://bugs.launchpad.net/cinder/+bug/1409012 | 12:46 |
| openstack | Launchpad bug 1409012 in Cinder "Volume becomes in 'error' state after scheduler starts" [High,Fix released] - Assigned to Michal Dulko (michal-dulko-f) | 12:46 |
| mark-os | dasm: Just ran that command, I'm getting the same error as the error which occurs when running the server | 12:46 |
| dasm | and what is happening in tuskar-api logs? | 12:46 |
| *** rlandy has joined #tripleo | 12:47 | |
| mark-os | dasm: where can I find these logs? | 12:47 |
| dasm | how are you running tuskar-api? | 12:48 |
| dasm | i'm starting it with: tuskar-api --config-file etc/tuskar/tuskar.conf | 12:48 |
| dasm | and in the same terminal, i have output, with errors, debug infos, etc. | 12:48 |
| dasm | so similar to this, you should have running instance of API | 12:48 |
| *** adrianopetrich_ has joined #tripleo | 12:49 | |
| dasm | started by yourself, or by automatic script, or sth. | 12:49 |
| *** adrianopetrich has quit IRC | 12:49 | |
| dasm | so, first of all: check if you have tuskar-api running: ps aux | grep tuskar-api | 12:49 |
| dasm | mine: root 6055 1.4 0.5 206784 47428 pts/26 S+ 12:45 0:03 /usr/bin/python /usr/local/bin/tuskar-api --config-file=etc/tuskar/tuskar.conf | 12:50 |
| dasm | for me, it's running in undercloud | 12:50 |
| mark-os | dasm: I start tuskar-api the same way as you. Just noticed that when I started the api and left it running and went to the terminal where I executed $ tuskar plan-list , a message appeared in the tuskar-api terminal | 12:51 |
| mark-os | dasm: do I have to just leave tuskar-api running on that terminal? | 12:51 |
| *** tserong has quit IRC | 12:52 | |
| dasm | you need to have tuskar-api running all the time | 12:52 |
| mark-os | dasm: /infrastructure/ appeared now | 12:54 |
| dasm | +1 | 12:54 |
| mark-os | dasm: can't believe that was the reason why I was getting the error | 12:54 |
| mark-os | dasm: I wasn't leaving the API running | 12:54 |
| shardy | derekh: Hey, how long does delorean take to pick up changes e.g https://review.gerrithub.io/#/c/229858/? | 12:54 |
| shardy | is there somewhere I can check status for current builds? | 12:54 |
| dasm | mark-os: /infrastructure tries to read an info from tuskar-api | 12:55 |
| dasm | so if there is no API, it receives an error. | 12:55 |
| *** weshay has joined #tripleo | 12:55 | |
| dasm | and that's the whole mistery ;) | 12:55 |
| derekh | shardy: that patch hasn't merged | 12:55 |
| dasm | s/mistery/mystery | 12:55 |
| derekh | shardy: it will appear here once built http://trunk.rdoproject.org/f21/report.html (The report will be better later today) | 12:56 |
| shardy | derekh: gah, that gerrithub ui is really confusuing | 12:57 |
| slagle | dprince: hi, it looks like the passthrough support didnt make it into the set of -puppet templates. e.g., ControllerPassthroughDeployment/ControllerPassthroughSpecificDeployment. | 12:57 |
| slagle | dprince: i was going to work on adding them | 12:57 |
| *** athomas has quit IRC | 12:57 | |
| shardy | I saw apevec's Verified +1 and equated that to approved | 12:57 |
| slagle | dprince: unless it was intentional? | 12:57 |
| derekh | shardy: its merged now, so it shouldn't be long | 12:58 |
| shardy | derekh: Ok, thanks, just keen to get my local environment un-broken ;) | 12:58 |
| derekh | shardy: yup | 12:58 |
| shardy | slagle: have you seen https://review.openstack.org/#/c/170137/ ? | 13:00 |
| dprince | slagle: looking | 13:00 |
| derekh | dprince: reverting the patch that moved us onto f21/current failed CI, error loogs different, rechecking https://review.openstack.org/#/c/171989/1 | 13:00 |
| shardy | we were dicussing adding a similar post-deploy hook, which could run any software config in an operator-provided template | 13:00 |
| shardy | using parameter_defaults would avoid the need for the pass-through parameters | 13:00 |
| shardy | slagle: not that I'm opposed to reinstating the interfaces you mention, it's just an alternative approach | 13:01 |
| dprince | derekh: it will fail because of the Ironicclient breakage issue | 13:01 |
| slagle | shardy: cool, i'll take a look at that. we were previously using the ExtraConfig parameters to enable the rhn/satellite registration | 13:01 |
| dprince | derekh: I would like to see us just pin Delorean to something on Tuesday | 13:01 |
| slagle | shardy: i think i could make that work with this other patch though | 13:01 |
| slagle | shardy: while i have you though, we also need something to unregister overcloud nodes on a stack-delete or scale down | 13:02 |
| mark-os | dasm: yeah I honestly can't believe that was the reason why it wasn't working haha | 13:02 |
| slagle | shardy: i was thinking I could use the softwaredeployment Delete lifecycle for this, but i wasnt sure where that would get wired in | 13:02 |
| derekh | dprince: ahh, I didn;t know about a ironicclient breakage, ok will try something from tuesday | 13:02 |
| dprince | derekh: http://trunk.rdoproject.org/f21/ca/06/ca06d64d6131de1f3a8bfca87f0620b27c746976_657408d4 | 13:02 |
| dprince | derekh: try that one | 13:03 |
| mark-os | dasm: I think it would be good to amend the documentation so that the users know that it has to be left running. I thought it was similar to a linux service where you just start it the once and it'd work | 13:03 |
| shardy | slagle: yeah you can probably use a SoftwareDeployment with an action that's only for DELETE | 13:03 |
| dprince | derekh: that is from my overcloud images (built on Tuesday) which have been working fine on F21 | 13:03 |
| mark-os | dasm: Is there any way I can confirm that Tuskar UI is fully working for me now? | 13:03 |
| shardy | slagle: I imagined we could have a firstboot dir for the cloud-init extra stuff mapped to e.g OS::TripleO::NodeUserData | 13:04 |
| *** athomas has joined #tripleo | 13:04 | |
| shardy | slagle: then we could also have, say, OS::TripleO::NodeExtraConfig | 13:04 |
| shardy | which defaults to a no-op template, but takes the server id, and can be used to hook in arbitrary additional config/deployment resources | 13:05 |
| shardy | e.g including anything that only happens on delete potentially | 13:05 |
| slagle | shardy: alright, that sounds like it would work | 13:05 |
| shardy | slagle: let me hack out a patch to illustrate the latter approach | 13:06 |
| dasm | mark-os: tuskar-ui is just an extension to horizon. so if you have already "infrastructure" there, it means that's running | 13:06 |
| openstackgerrit | Derek Higgins proposed openstack-infra/tripleo-ci: Pin to a known good delorean repository https://review.openstack.org/172039 | 13:06 |
| dasm | ofc, if you've run all the additional work connected to: templates, creating plans, creating roles, etc. | 13:06 |
| dprince | slagle: re your question about pass-thru I basically agree with shardy's approach | 13:06 |
| mark-os | dasm: I can't see the other sections for Horizon though. I only see infrastructure | 13:06 |
| dprince | slagle: we can somewhat wire in the ExtraConfig stuff to puppet, but it isn't going to be intuitive I'm afraid | 13:07 |
| slagle | dprince: yea, i'm going to give it a go with his patch instead | 13:07 |
| *** lifeless has quit IRC | 13:07 | |
| slagle | i like this model better | 13:07 |
| dprince | slagle: what exactly are you trying to customize? | 13:07 |
| *** trown|outttypeww is now known as trown | 13:07 | |
| dprince | slagle: is it just a one off config value? or something more? | 13:07 |
| slagle | dprince: need to pass in the data for rhn registration. it's 5 or so values | 13:08 |
| *** spzala has joined #tripleo | 13:08 | |
| dasm | mark-os: if you've followed this instruction: http://tuskar-ui.readthedocs.org/en/latest/install.html#install-the-management-ui one of the steps means: "Then disable the other dashboards:" | 13:08 |
| dasm | so, everything apart "Infrastructure" is disabled by you. | 13:08 |
| dprince | slagle: I see, so yeah. ExtraConfig with puppet would only give you access to set things for OpenStack services. | 13:08 |
| dprince | slagle: it would be clunky (with puppet). Hoenstly ExtraConfig is already a bit clunky as is | 13:09 |
| mark-os | dasm: why would they suggest disabling the other dashboards? Just for testing purposes? How would I go about enabling them again? | 13:09 |
| dprince | slagle: I'd sort of like to see it go away I think | 13:09 |
| openstackgerrit | Ryan Hallisey proposed stackforge/kolla: Update error message for build-docker-images https://review.openstack.org/171822 | 13:09 |
| dasm | mark-os: because you don't need it in tuskar. | 13:09 |
| *** yamahata has joined #tripleo | 13:10 | |
| mark-os | dasm: why so? Does Tuskar cover everything which was covered by the previous 3 dashboards? | 13:10 |
| dprince | slagle: the best we could do is to map the ExtraConfig stuff into the various puppet config providers (nova_config, glance_config, etc.) which would allow you to set some arbitrary values only if they don't override things already being managed by the puppet modules themselves. | 13:10 |
| dprince | slagle: so it isn't ideal. With TIE we essentially duplicated config sections and rely'd on the fact that the last setting wins | 13:11 |
| dasm | mark-os: because you're trying to configure undercloud. you don't need rest of dashboards in the undercloud. | 13:11 |
| mark-os | dasm: So tuskar configures undercloud, what does Horizon configure? | 13:12 |
| slagle | dprince: agreed. i'm going to try it this other way instead. plus i like the idea of being able to checkin templates for the different SoftwareDeployments you want to include in your stack to t-h-t directly | 13:15 |
| dprince | slagle: right, like we do for the basic net- templates | 13:17 |
| dprince | slagle: you can easily choose between them within the Heat environment | 13:18 |
| *** jprovazn has quit IRC | 13:20 | |
| *** jprovazn has joined #tripleo | 13:20 | |
| *** noslzzp has joined #tripleo | 13:20 | |
| *** jprovazn has quit IRC | 13:21 | |
| openstackgerrit | Yanis Guenane proposed openstack/tripleo-heat-templates: Add support for Redis configuration https://review.openstack.org/167538 | 13:26 |
| openstackgerrit | Steven Hardy proposed openstack/tripleo-heat-templates: Enable passing optional first-boot user-data https://review.openstack.org/170137 | 13:29 |
| openstackgerrit | Steven Hardy proposed openstack/tripleo-heat-templates: Add hooks for extra post-deployment config https://review.openstack.org/172050 | 13:29 |
| shardy | slagle: ^^ | 13:29 |
| shardy | that second example is untested, but hopefully illustrates one way we could do it | 13:29 |
| slagle | shardy: awesome! | 13:29 |
| shardy | it's be nice to remove the puppet/element template duplication, but otherwise hopefully it gets the idea across | 13:30 |
| shardy | slagle: one disadvantage of the user_data pre-deploy hook is lack of signalling, so e.g if you fail to reach satellite it's going to be hard to signal failure | 13:31 |
| shardy | so one alternative would be to copy the second approach and have e.g extraconfig/pre_deploy/foo.yaml | 13:32 |
| shardy | which runs after firstboot but before we start configuring things | 13:32 |
| shardy | Or, we can just do everything post-deploy in the fat-image case | 13:32 |
| dprince | derekh: the issue with ironicclient was that tripleo-incubator was pip installing the latest client, but we were using an older Ironic version in the seed | 13:35 |
| spredzy | dprince, ping. IF I wanted to use a custome fact in t-h-t, where could I put it ? Does it have to be in puppe-tripleo or is there another mechanism to provide a custom fact ? | 13:35 |
| derekh | dprince: ahh, ok | 13:36 |
| dprince | spredzy: so the puppet hook we use supports using factor instead of (or in addition to) hiera | 13:36 |
| dprince | spredzy: http://git.openstack.org/cgit/openstack/tripleo-heat-templates/tree/puppet/controller-post-puppet.yaml#n18 | 13:37 |
| slagle | shardy: so this is the current rhn registration o-r-c script: https://github.com/openstack/diskimage-builder/blob/master/elements/rhel-common/os-refresh-config/pre-configure.d/06-rhel-registration | 13:37 |
| slagle | shardy: if i wanted the equivalent in this other way, i would more or less have that script run as part of a SoftwareDeployment? | 13:37 |
| spredzy | dprince, so where should I put my new fact if I create one ? | 13:37 |
| dprince | spredzy: you'll see those options around in several places. If you enable factor then then when 'puppet apply' runs for that SoftwareConfig it will set FACTOR_ variables for all of the heat inputs | 13:37 |
| dprince | spredzy: if you enable factor and hiera you'll get them in both places (hieradata files and FACTOR_ variables) | 13:38 |
| dprince | spredzy: the best thing to do is try it. Just set enable_facter: True in your controller-post-puppet.yaml template and then look at the effect on the os-collect-config log file in your controller node | 13:40 |
| spredzy | dprince, will do that | 13:40 |
| *** prad has joined #tripleo | 13:40 | |
| dprince | spredzy: you could even do this with a heat stack-update if you've already got a running overcloud | 13:40 |
| spredzy | dprince, nop I have not. I am about to start one | 13:40 |
| dprince | spredzy: you should see the 'step' variables then getting set with FACTOR_ variables | 13:40 |
| shardy | slagle: yes, exactly, you'd run that via a SoftwareConfig/SoftwareDeployment | 13:41 |
| dprince | spredzy: where as without the change it just uses a custom hiera datafile for the step variable | 13:41 |
| shardy | slagle: optionally providing the REG_ variables as inputs to the deployment/config | 13:41 |
| shardy | slagle: sec, let me push an example doing that.. | 13:42 |
| shardy | slagle: does this need to happen pre-deploy so we can pull packages? | 13:42 |
| slagle | shardy: it probably should eventually | 13:44 |
| slagle | shardy: although, initially all the package are preinstalled | 13:44 |
| shardy | slagle: Ok I'll do the post-deploy example for now | 13:44 |
| shardy | the pre-deploy will look identical apart from where we hook it in | 13:44 |
| openstackgerrit | Marios Andreou proposed openstack/tuskar: Wires up constraints and type into returned plan parameters https://review.openstack.org/171713 | 13:44 |
| slagle | yea, that'll be enough for me to go on | 13:44 |
| marios | jtomasek: updated | 13:47 |
| jtomasek | marios: looking | 13:48 |
| *** jcoufal has joined #tripleo | 13:48 | |
| *** ccrouch has joined #tripleo | 13:52 | |
| *** openstackgerrit has quit IRC | 13:53 | |
| *** openstackgerrit has joined #tripleo | 13:53 | |
| *** jprovazn has joined #tripleo | 13:56 | |
| openstackgerrit | Merged openstack/diskimage-builder: No markdown docs for elements https://review.openstack.org/169992 | 13:59 |
| *** dasm is now known as dasm|afk | 14:00 | |
| openstackgerrit | Victor Lowther proposed openstack/os-cloud-config: Add pxe_drac driver support to os-cloud-config https://review.openstack.org/171774 | 14:02 |
| *** pradeep has joined #tripleo | 14:03 | |
| openstackgerrit | Steven Hardy proposed openstack/tripleo-heat-templates: Example of post-deploy hook doing rhel registration https://review.openstack.org/172065 | 14:04 |
| shardy | slagle: ^^ | 14:04 |
| *** jerryz has joined #tripleo | 14:04 | |
| *** lblanchard has joined #tripleo | 14:04 | |
| shardy | note, that uses an additional environment file, environment_rhel_reg.yaml | 14:04 |
| shardy | it may be better if we establish a standard "extra" environment file somewhere, or if the values are put into the default (global) registry on the seed | 14:05 |
| *** sdake has joined #tripleo | 14:10 | |
| *** mark-os has quit IRC | 14:11 | |
| *** pdb has quit IRC | 14:16 | |
| *** pdb has joined #tripleo | 14:17 | |
| *** pcaruana has quit IRC | 14:17 | |
| *** radez_g0n3 is now known as radez | 14:18 | |
| *** lifeless has joined #tripleo | 14:18 | |
| slagle | shardy: cool, i'll take a look | 14:19 |
| gfidente | dprince, so here is my take, given we use multibackend in cinder already, we have params to configure glance with swift already, and glance supports multiple stores natively, I would like to use provide it as a feature instead of taking the route of making the backends a selection | 14:22 |
| gfidente | dprince, but there are two issues, 1) my submission could get there, but it doesn't implement it correctly now, you end up with just the rbd backend working 2) puppet-glace should NOT do the defaul_store thing, that is crazy | 14:23 |
| gfidente | dprince, what do you think? | 14:23 |
| *** radez is now known as radez_bbl | 14:23 | |
| dprince | gfidente: That is the reason I choose GlanceBackend. It works as is, quite nicely with the glance backend modules. | 14:24 |
| gfidente | yeah | 14:24 |
| gfidente | but it really is puppet-glance doing it wrong | 14:24 |
| dprince | gfidente: I do agree that the puppet modules should probably not be setting default_store there. However, I'm not sure there is a strong use case for multiple Glance backends to be used at the same time | 14:25 |
| dprince | gfidente: using a glance backend which isn't the default can get clunky on the CLI | 14:25 |
| *** jcoufal has quit IRC | 14:26 | |
| *** jcoufal has joined #tripleo | 14:27 | |
| gfidente | dprince, okay I'll be honest, I will not push for the multiple backends in glance because puppet-glance won't help us there, causing instead lot of troubles | 14:28 |
| *** jcoufal has quit IRC | 14:29 | |
| gfidente | but if you ask me, we're giving up the feature because of puppet-glance | 14:29 |
| openstackgerrit | Merged openstack/tuskar-ui: Remove duplicated MAC addresses from multi MAC field https://review.openstack.org/168318 | 14:29 |
| *** mark-os has joined #tripleo | 14:31 | |
| dprince | gfidente: Understood that puppet-glance may have had some effect our our decision here. People are using it however... and products are already built on it. If nobody has yet found a need to simultaneously use multiple glance backends at once is it really a feature we need? | 14:32 |
| dprince | gfidente: infinite configurability is cool... but if it costs us extra work and the "feature" is only ficticious then I'm not sure it is worth it | 14:33 |
| dprince | gfidente: all that said, we should ask around. If people want to do this then like you said we can put in the work to make it happend | 14:33 |
| gfidente | dprince, indeed this is a fact I can't oppose to, but the actual tool (glance) is meant to work with multiple backends ... it is only the specific puppet module which prevents you from using it easily | 14:34 |
| dprince | gfidente: yes, we agree | 14:34 |
| mark-os | Can anyone tell me why Tuskar UI is quite slow compared to Horizon? | 14:35 |
| spredzy | dprince, I added enable_facter: True, but all I see are data that comes from Heat itself. So a custom fact I want to create is one that check if a psecific file is present or not on the system. | 14:37 |
| spredzy | Should that go to puppet-tripleo ? | 14:38 |
| *** bvandenh has joined #tripleo | 14:38 | |
| dprince | spredzy: perhaps, I'd have to see it first :) | 14:39 |
| dprince | spredzy: push a quick patch maybe? | 14:39 |
| spredzy | dprince, https://review.openstack.org/#/c/171555/11/lib/facter/galera_bootstrapped.rb,cm | 14:39 |
| spredzy | I want to add this fact in the installer | 14:40 |
| spredzy | just not sure about where it belongs | 14:40 |
| derekh | Can we get a second +2 on this to unblock puppet job https://review.openstack.org/#/c/172039/1 | 14:47 |
| *** jerryz has quit IRC | 14:48 | |
| *** noslzzp has quit IRC | 14:48 | |
| * jistr on it | 14:49 | |
| shardy | derekh: FWIW I applied that pin locally and Neutron still failed to start | 14:50 |
| shardy | it's possible I did something wrong though - did it work for you? | 14:50 |
| jistr | i did the same and it worked for me... | 14:50 |
| * jistr double checks | 14:50 | |
| shardy | http://trunk.rdoproject.org/f21/ca/06/ca06d64d6131de1f3a8bfca87f0620b27c746976_657408d4 | 14:51 |
| shardy | that's in my environment, let me re-test | 14:51 |
| jistr | shardy: do you have USE_CACHE=0 ? | 14:51 |
| derekh | shardy: jistr it worked in CI, hasn't reported back yet https://jenkins02.openstack.org/job/check-tripleo-ironic-overcloud-f20puppet-nonha/140/ | 14:51 |
| jistr | shardy: also you might check the logs that the neutron installed into the image is indeed dev276 and not dev296 | 14:53 |
| jistr | shardy: and you might need to rm -rf ~/.cache/image-create | 14:53 |
| shardy | jistr: No, I've not got USE_CACHE set at all | 14:53 |
| shardy | I'll try removing the cache and all the qcow's | 14:53 |
| jistr | yeah i'd say that should do it | 14:54 |
| *** jcoufal has joined #tripleo | 14:54 | |
| *** cwolferh has joined #tripleo | 14:57 | |
| *** untriaged-bot has joined #tripleo | 15:00 | |
| untriaged-bot | No untriaged bugs so far! \o/ | 15:00 |
| *** untriaged-bot has quit IRC | 15:00 | |
| *** maufart has quit IRC | 15:03 | |
| openstackgerrit | Yanis Guenane proposed stackforge/puppet-tripleo: Add $::galera_bootstrapped fact https://review.openstack.org/172107 | 15:04 |
| openstackgerrit | Merged openstack-infra/tripleo-ci: Pin to a known good delorean repository https://review.openstack.org/172039 | 15:05 |
| openstackgerrit | Yanis Guenane proposed stackforge/puppet-tripleo: Add $::galera_bootstrapped fact https://review.openstack.org/172107 | 15:06 |
| *** jcoufal has quit IRC | 15:06 | |
| mark-os | jistr: any idea why tuskar ui is so slow? | 15:09 |
| *** yamahata has quit IRC | 15:10 | |
| jistr | mark-os: i don't work with UI usually, but my guess would be that if you're running Horizon via the command you posted earlier, it runs in dev mode, checking on every request what changed in the source, or maybe even handling each request with a freshly loaded code. I'm just guessing though, i've never worked with django. | 15:12 |
| mark-os | jistr: Sounds like that's what could well be the reason. Do you know who would have a better idea about performance of Tuskar UI? | 15:15 |
| *** jistr is now known as jistr|mtg | 15:15 | |
| *** jprovazn has quit IRC | 15:20 | |
| jistr|mtg | mark-os: maybe jtomasek? In general these people i think: https://review.openstack.org/#/q/status:merged+project:openstack/tuskar-ui,n,z | 15:24 |
| *** jistr|mtg is now known as jistr | 15:27 | |
| *** spredzy is now known as spredzy|afk | 15:32 | |
| mark-os | jistr: thanks! | 15:33 |
| mark-os | jtomasek: ping | 15:34 |
| jtomasek | mark-os: pong, yes it is slow at the moment, especially in development mode. It is mostly caused by quite lot of api requests done to load data for the page and quite often these requests aren't very well optimized. I'd call it work in progress :) | 15:37 |
| *** jistr has quit IRC | 15:40 | |
| *** jtomasek has quit IRC | 15:47 | |
| *** alop has joined #tripleo | 15:53 | |
| *** Marga_ has joined #tripleo | 15:53 | |
| *** Marga_ has quit IRC | 15:53 | |
| *** Marga_ has joined #tripleo | 15:54 | |
| *** rhallisey has quit IRC | 15:55 | |
| *** dtantsur is now known as dtantsur|afk | 15:57 | |
| *** rhallisey has joined #tripleo | 15:57 | |
| *** yamahata has joined #tripleo | 16:02 | |
| *** rwsu has joined #tripleo | 16:02 | |
| *** sirushti has left #tripleo | 16:03 | |
| *** radez_bbl is now known as radez | 16:05 | |
| openstackgerrit | Jeff Peeler proposed openstack/tripleo-heat-templates: Make all default values match overcloud defaults https://review.openstack.org/171411 | 16:10 |
| *** derekh has quit IRC | 16:12 | |
| *** bvandenh has quit IRC | 16:12 | |
| *** lucasagomes has quit IRC | 16:16 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/diskimage-builder: Updated from global requirements https://review.openstack.org/166431 | 16:17 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/os-cloud-config: Updated from global requirements https://review.openstack.org/157776 | 16:24 |
| *** radez is now known as radez_g0n3 | 16:30 | |
| greghaynes | bnemec: replied on https://review.openstack.org/#/c/169167/12/tests/run_tests.sh | 16:31 |
| *** athomas has quit IRC | 16:33 | |
| greghaynes | bnemec: I was hoping to get some jobs pushed up today :) | 16:38 |
| *** yog__ has quit IRC | 16:49 | |
| SpamapS | greghaynes: good job man | 16:54 |
| greghaynes | Yep, im still kind of bummed that we dont have a strat for testing that we can make an image that boots | 16:56 |
| *** Goneri has quit IRC | 16:58 | |
| *** bvandenh has joined #tripleo | 16:59 | |
| greghaynes | dib docs question - do we want to suggest pip installation at all or just stick with the "dib is installed by addind bin to your path" | 17:03 |
| greghaynes | technically, pip install does work, and it is kind of neat for a user to be able to just pip install diskimage-builder;disk-image-create <distro> vm | 17:03 |
| slagle | there was another review out there to make it runnable from the venv... | 17:04 |
| slagle | so no system modifications necessary | 17:04 |
| greghaynes | yea, good point that I should probably verify it works ;) | 17:05 |
| slagle | https://review.openstack.org/#/c/168002/ | 17:06 |
| greghaynes | oh! right | 17:06 |
| greghaynes | I get around that by sourcing venv | 17:06 |
| *** pradeep has quit IRC | 17:13 | |
| *** openstack has quit IRC | 17:13 | |
| *** openstack has joined #tripleo | 17:14 | |
| greghaynes | bnemec: we can, it will just sudo. We cant use the tox jobs infra has for doing it so its not really much of a benefit | 17:16 |
| greghaynes | the response I got is that generally they arent too fond of tox targets that sudo | 17:17 |
| *** radez_g0n3 is now known as radez | 17:17 | |
| bnemec | Except we need to sudo regardless of how we run it. :-/ | 17:17 |
| greghaynes | Yep | 17:18 |
| *** pelix has quit IRC | 17:18 | |
| greghaynes | I think the idea is that tox doesnt really provide you any value if youre not running a test that requires a specific venv, and if youre sudoing in your target then theres potential issues with files you end up creating | 17:19 |
| greghaynes | not an issue for us since we arent using a venv at all, so then the question is why tox | 17:20 |
| bnemec | So I guess my motivation here was twofold - consistency and the ability to just run "tox -e functional" without installing anything to the host. | 17:20 |
| bnemec | But since it sounds like running in a venv doesn't work right now anyway, maybe that's not realistic. | 17:21 |
| *** Marga_ has quit IRC | 17:21 | |
| greghaynes | The consistency is nice, the installation shouldnt be an issue since we dont have any python deps (only package ones) and tox doesnt buy us anything there | 17:21 |
| greghaynes | bnemec: what did you think about run_functests.sh? | 17:22 |
| greghaynes | for the name | 17:22 |
| *** spredzy|afk is now known as spredzy | 17:22 | |
| bnemec | greghaynes: I think I'm fine with that. | 17:22 |
| greghaynes | actually, it just occured to me that maybe we could run dib out ot the tox venv rather than bin/ so we actually test venv installation | 17:24 |
| greghaynes | in which case tox -e functional is useful | 17:24 |
| *** Marga_ has joined #tripleo | 17:24 | |
| greghaynes | clarkb: ^ will that offend ou horribly since were going to be sudoing in that target | 17:25 |
| bnemec | For the "able to boot" thing, could we set up a dsvm job that doesn't run tempest, but runs a functional test that builds, uploads, and boots an image in devstack? | 17:25 |
| *** akrivoka has joined #tripleo | 17:26 | |
| greghaynes | bnemec: the problem I ran into is that qemu booting is *way* slow, are there dsvm jobs that qemu boot somehow? | 17:26 |
| greghaynes | (when I say way slow, ssh would timeout to the qemu host before ssh could reply) | 17:27 |
| clarkb | greghaynes: if running `tox` tries to escalate privileges locally I pretty much never touch the project again | 17:27 |
| clarkb | greghaynes: but thats me | 17:27 |
| greghaynes | haha | 17:27 |
| greghaynes | tell us how you really feel | 17:27 |
| bnemec | greghaynes: I think they all do. | 17:27 |
| clarkb | we also actively fail tests that run tox and sudo | 17:27 |
| bnemec | tox -e trash-my-machine ;-) | 17:27 |
| greghaynes | nice | 17:27 |
| greghaynes | bnemec: hrm, so I should look into how they get away with that | 17:28 |
| clarkb | greghaynes: bnemec yes I think sdague ran numbers at one point and we boot a quarter million qemu VMs a week (or something in that range) | 17:29 |
| *** Marga_ has quit IRC | 17:29 | |
| clarkb | greghaynes: we boot cirros | 17:29 |
| clarkb | which runs busybox | 17:29 |
| greghaynes | aye | 17:29 |
| greghaynes | the problem with cirros is we need to install grub | 17:29 |
| clarkb | it actually works well enough that I just ssh'd between two VMs on different hypervisors in that setup | 17:29 |
| bnemec | greghaynes: We boot a nested fedora vm in the instack sanity check script, but we do have a bunch of wait_fors: https://github.com/rdo-management/instack-undercloud/blob/master/scripts/instack-test-overcloud#L46 | 17:29 |
| greghaynes | ok, so sounds like the path to sanity there is find a way to not boot a full on cloud iage | 17:30 |
| greghaynes | er, cloud image | 17:30 |
| bnemec | I'm wondering if maybe waiting for cloud-final to complete helps, because then hopefully a lot of the startup tasks have completed so you're not competing for CPU time. | 17:31 |
| greghaynes | bnemec: huh, thats a fedora? | 17:31 |
| greghaynes | and says 300 secs | 17:31 |
| bnemec | Because we even manage to attach and format a volume with no problem. | 17:31 |
| greghaynes | bnemec: this was me just testing manually on a cloud node | 17:32 |
| bnemec | greghaynes: Yeah, stock fedora image. | 17:32 |
| greghaynes | bnemec: I wonder if that works because youre testing on a much more performant cloud? | 17:32 |
| * SpamapS catches up | 17:32 | |
| bnemec | It's possible. We generally do this on dedicated baremetal boxes. | 17:32 |
| greghaynes | ah | 17:33 |
| clarkb | ya dedicated baremetal is likely to use kvm not qemu right? | 17:33 |
| greghaynes | so I wonder if we can just init to busybox for our ubuntu and fedora images and boot them | 17:33 |
| bnemec | clarkb: No, this is still running nested, it's just nested in a vm on a standalone box so there's not much resource contention. | 17:34 |
| greghaynes | just to get coverage on bootloader installation | 17:34 |
| *** bvandenh has quit IRC | 17:34 | |
| bnemec | Although we still only give the nova-compute vm 2 cpus, and the nested vm only gets 1, I believe. | 17:34 |
| SpamapS | greghaynes: so pass init=/bin/some-bash-script on kernel cmdline? | 17:34 |
| greghaynes | yep | 17:34 |
| SpamapS | greghaynes: I'm not sure what that's trying to accomplish. | 17:35 |
| clarkb | SpamapS: it checks that you have a bootloader and kernel | 17:35 |
| SpamapS | Sure it tests the boot loader, but I'm not sure that's worth the trouble. | 17:35 |
| greghaynes | ensure we install bootloader | 17:35 |
| clarkb | SpamapS: because as greghaynes discovered recently that isn't always the case | 17:35 |
| SpamapS | Oh? | 17:35 |
| greghaynes | yea, our -minimal's were failing there | 17:35 |
| greghaynes | and still are AFAICT | 17:36 |
| clarkb | they were/are so minimal they didn't have kernels | 17:36 |
| SpamapS | Well if that's the case, then we can make the script rather simple.. it can just do a few sanity checks and then halt the VM. | 17:36 |
| bnemec | That's one way to save space. :-) | 17:36 |
| SpamapS | honestly.. | 17:36 |
| SpamapS | I think that's good | 17:36 |
| SpamapS | chroots don't need kernels | 17:36 |
| SpamapS | containers don't need kernels | 17:37 |
| greghaynes | Yep | 17:37 |
| SpamapS | maybe we should rename to os-image-builder ;) | 17:37 |
| clarkb | the problem with it is this | 17:37 |
| clarkb | the error you get back from such an image when trying to boot a VM with it is almost useless | 17:37 |
| greghaynes | when writing these docs I keep thinking that the disk part is less relevant :) | 17:37 |
| clarkb | and clouds make it hard to debug | 17:37 |
| greghaynes | Sure, so IMO the poorly named vm element should error if you dont have a kernel | 17:37 |
| clarkb | especially if you don't have a Xen hypervisor laying around | 17:37 |
| SpamapS | clarkb: yeah it fails somewhere deep in the bowels of QEMU's BIOS I imagine. | 17:38 |
| greghaynes | or something along those lines | 17:38 |
| clarkb | also a chroot/container with a kernel won't hurt anything right? | 17:38 |
| SpamapS | greghaynes: hm. Not sure I agree with that. Kernel could easily come post-pxe-boot | 17:38 |
| *** spredzy is now known as spredzy|afk | 17:38 | |
| SpamapS | Then again, so could bootloader. ;) | 17:39 |
| greghaynes | aye | 17:39 |
| SpamapS | greghaynes: anyway, I think we need to come up with a name for a wrapper which is minimal+stuff | 17:39 |
| SpamapS | we could rename -minimal to -just-userspace | 17:39 |
| greghaynes | im trying to think what the use cases are, A user wants to make an image that will 'just work' for most clouds / libvirt; a super minimal one they can build off of; is that it? | 17:43 |
| clarkb | as a user here, I want one image that I can just use in a variety of situations | 17:44 |
| greghaynes | yea, so thats the first one | 17:44 |
| clarkb | I would likely be most happy to have a bootloader and kernel in place even if I was just using the image for a container today | 17:44 |
| clarkb | because tomorrow I will boot it in $cloud | 17:44 |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Add support for Glance RBD backend https://review.openstack.org/172032 | 17:44 |
| greghaynes | I think the minimal is a lot more useful for someone writing their own elements | 17:44 |
| greghaynes | since theres not really a way to opt out of things that an element provides, if you dont have a super minimal os one then you have to either make your own element for that distro or use the everything and kitchen sink one | 17:46 |
| clarkb | so maybe the thing to do is to rename it as SpamapS suggests | 17:49 |
| clarkb | to me "minimal" implies "minimally viable" | 17:49 |
| greghaynes | yep | 17:50 |
| clarkb | then have a minimal element that tacks on kernel and bootloader to just-userspace | 17:50 |
| greghaynes | I kind of want to make that the ubunt element | 17:50 |
| SpamapS | haha | 17:51 |
| SpamapS | so another way to look at it | 17:51 |
| SpamapS | is that the composition should almost always happen at runtime | 17:52 |
| SpamapS | unless you have something you need to do to glue two elements together, then it's best to just let them stay as their own islands. | 17:52 |
| clarkb | except that as a user I hate that | 17:52 |
| SpamapS | so you want meta packages | 17:52 |
| clarkb | its confusing to know what I need to have a viable image | 17:52 |
| SpamapS | and I get that | 17:52 |
| clarkb | when really I just want to tell dib make me an ubuntu | 17:52 |
| clarkb | currently I end up reading a lot of shell | 17:53 |
| SpamapS | The README's aren't good enough then. | 17:53 |
| clarkb | thats part of the problem | 17:53 |
| greghaynes | thats part of it, wed also need recipes of "this set of elements gets you an ubuntu" | 17:53 |
| greghaynes | otherwise you have to read all the readmes | 17:53 |
| clarkb | greghaynes: right | 17:54 |
| SpamapS | 'ubuntu-minimal vm' would seem to be the right mix, but it won't currently get you what you want. | 17:54 |
| greghaynes | nope, that will get you a "what, why cant I ssh in" | 17:55 |
| SpamapS | clarkb: 'ubuntu' does give you want you want | 17:55 |
| greghaynes | 'ubuntu vm' | 17:55 |
| SpamapS | clarkb: it just doesn't give you what mordred wants: no cloud-init ;) | 17:55 |
| clarkb | SpamapS: no I have to add at least vm iirc | 17:55 |
| clarkb | and seems like there was one other element too | 17:55 |
| SpamapS | Yeah, which is because we can't remove stuff. | 17:55 |
| greghaynes | nah, ubuntu vm is your 99% use case | 17:55 |
| greghaynes | 99% ubuntu use case ;) | 17:55 |
| clarkb | ya its ubuntu vm | 17:56 |
| *** Marga_ has joined #tripleo | 17:56 | |
| greghaynes | ok, so the ubuntu-minimal element is kind of breaking that composition paradigm | 17:56 |
| greghaynes | sorry, how were usin git | 17:56 |
| SpamapS | so basically, what we want is a meta-element called 'ubuntu-minimal' which pulls in 'ubuntu-userspace kernel vm vm-kernel' | 17:56 |
| greghaynes | actually, maybe not, as long as we dont go down the path of adding element-deps | 17:57 |
| SpamapS | with vm-kernel being something I just invented to mean "install the kernel that this OS thinks is best for vms" | 17:57 |
| greghaynes | the fedora-minimal does this though | 17:57 |
| SpamapS | does what? | 17:57 |
| greghaynes | fedora-minimal depends on yum-minimal | 17:57 |
| SpamapS | thats fine | 17:58 |
| SpamapS | IIRC mordred has patches to make ubuntu and debian minimals depend on a debootstrap element | 17:58 |
| SpamapS | same thing really | 17:58 |
| *** spredzy|afk is now known as spredzy | 18:01 | |
| greghaynes | your kernel vs vm-kernel element... whats the difference there? | 18:03 |
| greghaynes | and then, can we rename vm to bootloader :) | 18:04 |
| SpamapS | greghaynes: well Ubuntu for instance has linux-image-virtual which has no hardware drivers | 18:04 |
| clarkb | greghaynes: some distros insist that you run a different kernel in a VM | 18:05 |
| greghaynes | ah | 18:05 |
| greghaynes | so, wouldnt we want one or the other? | 18:05 |
| *** sdake_ has joined #tripleo | 18:05 | |
| SpamapS | looking at them | 18:06 |
| SpamapS | clarkb: insist is the wrong word | 18:06 |
| SpamapS | clarkb: Ubuntu uses the virtual one to save space. | 18:06 |
| SpamapS | image-generic works fine in VMs | 18:06 |
| SpamapS | and has all the virt drivers | 18:06 |
| *** Marga_ has quit IRC | 18:07 | |
| SpamapS | actually | 18:07 |
| SpamapS | I'm full of crap | 18:07 |
| *** mestery has quit IRC | 18:07 | |
| SpamapS | http://paste.ubuntu.com/10784567/ | 18:08 |
| SpamapS | Ubuntu's linux-image-virtual just depends on Generic. | 18:08 |
| SpamapS | clarkb: lets just call it kernel | 18:08 |
| SpamapS | also does the yum-minimal end up pulling in a kernel anyway? | 18:08 |
| greghaynes | that is precisely what I am messing with | 18:08 |
| greghaynes | right now, no | 18:08 |
| greghaynes | that does need to happen somewhere though | 18:09 |
| *** sdake has quit IRC | 18:09 | |
| *** spredzy is now known as spredzy|afk | 18:09 | |
| *** Goneri has joined #tripleo | 18:10 | |
| *** mestery has joined #tripleo | 18:10 | |
| SpamapS | greghaynes: lets just make a kernel element then. | 18:12 |
| SpamapS | it's really just a package | 18:13 |
| SpamapS | linux-image-generic for ubuntu, 'kernel' for everything rpm | 18:13 |
| SpamapS | not sure for Debian | 18:14 |
| SpamapS | probably linux-image-generic | 18:14 |
| greghaynes | +1 | 18:14 |
| greghaynes | ill split it out of the ubuntu-minimal also | 18:14 |
| SpamapS | is it already in there? | 18:14 |
| greghaynes | I think I added it | 18:14 |
| SpamapS | ah yeah | 18:14 |
| SpamapS | you know what | 18:15 |
| SpamapS | leave it | 18:15 |
| SpamapS | greghaynes: the current ubuntu-minimal has a kernel already in it. So how did you end up without a kernel? | 18:15 |
| greghaynes | https://review.openstack.org/#/c/167003/35/elements/ubuntu-minimal/package-installs.yaml ? | 18:15 |
| greghaynes | I added that | 18:15 |
| SpamapS | greghaynes: oh right, ok | 18:15 |
| SpamapS | so I'm backing up from this | 18:16 |
| SpamapS | lets just let the minimal elements install kernels | 18:16 |
| SpamapS | if somebody wants to build chroots or containers w/ diskimage-builder, let them refactor | 18:16 |
| SpamapS | I just realized we're bending over backwards to optimize about 30MB out of something we're not using. | 18:16 |
| greghaynes | ok. that will likely be me in the not far future, but that actually works - if we use -minimal as the name of the meta element then when we split it out well maintain the same feature set for that element | 18:17 |
| greghaynes | (containers is probably going to be my next functest spreee) | 18:17 |
| SpamapS | greghaynes: right, and then you can use an 'ubuntu-userspace' element for your containery chrooty things | 18:17 |
| greghaynes | yep | 18:17 |
| SpamapS | alright, I think I've just earned some extra caffeine. Go me. | 18:18 |
| greghaynes | \O/ | 18:18 |
| *** Marga_ has joined #tripleo | 18:18 | |
| *** Marga_ has quit IRC | 18:20 | |
| *** Marga_ has joined #tripleo | 18:20 | |
| greghaynes | ok, interesting thought - what happens when you specify -t tar (or aci, etc) and the vm element | 18:21 |
| greghaynes | I think we make the fs, intall the bootloader there, then tar up the chroot? | 18:21 |
| greghaynes | so its just an expensive noop | 18:21 |
| greghaynes | which makes me think - is there ever a reason to not have the vm element | 18:21 |
| openstackgerrit | Daneyon Hansen proposed stackforge/kolla: Updates Heat template Compose and Kolla Repositories https://review.openstack.org/172166 | 18:23 |
| greghaynes | that noop is pretty easy to optimize out I think - if were not outputting to a format that needs a filesystem dont make the fs or run the phases that operate on the fs | 18:25 |
| *** sdake has joined #tripleo | 18:25 | |
| openstackgerrit | Daneyon Hansen proposed stackforge/kolla: Updates Docker bin in Heat template https://review.openstack.org/168645 | 18:27 |
| *** sdake_ has quit IRC | 18:29 | |
| *** pblaho has quit IRC | 18:37 | |
| *** jtomasek has joined #tripleo | 18:43 | |
| *** lblanchard has quit IRC | 18:46 | |
| openstackgerrit | Dan Prince proposed openstack/tripleo-incubator: Update puppet docs to match our CI https://review.openstack.org/172174 | 18:47 |
| *** lblanchard has joined #tripleo | 18:48 | |
| *** alop has quit IRC | 18:58 | |
| *** spredzy|afk is now known as spredzy | 18:58 | |
| *** Marga_ has quit IRC | 19:03 | |
| *** Marga_ has joined #tripleo | 19:05 | |
| openstackgerrit | Merged stackforge/kolla: Adds Initial Logging Support https://review.openstack.org/169878 | 19:05 |
| *** Marga_ has quit IRC | 19:06 | |
| *** alop has joined #tripleo | 19:06 | |
| *** Marga_ has joined #tripleo | 19:06 | |
| *** Goneri has quit IRC | 19:14 | |
| *** dprince has quit IRC | 19:20 | |
| *** weshay has quit IRC | 19:29 | |
| gfidente | this has 2 x +2 but didn't get +a because of CI issues, anyone willing to help? https://review.openstack.org/#/c/170518/ | 19:32 |
| *** pblaho has joined #tripleo | 19:33 | |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Pass in libvirt_rbd_secret_key for nova compute https://review.openstack.org/170518 | 19:36 |
| gfidente | slagle :D | 19:38 |
| *** gfidente has quit IRC | 19:39 | |
| *** Marga_ has quit IRC | 19:39 | |
| *** Marga_ has joined #tripleo | 19:40 | |
| *** Marga_ has quit IRC | 19:40 | |
| *** Marga_ has joined #tripleo | 19:40 | |
| *** Marga_ has quit IRC | 19:45 | |
| *** alop has quit IRC | 19:52 | |
| *** alop has joined #tripleo | 20:01 | |
| *** lblanchard has quit IRC | 20:02 | |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Wire in list of controller nodes as memcache_servers for Swift proxy https://review.openstack.org/171248 | 20:18 |
| openstackgerrit | Merged openstack/tripleo-incubator: Switch USE_MERGEPY default to 0 https://review.openstack.org/171413 | 20:20 |
| *** cody-somerville has quit IRC | 20:32 | |
| openstackgerrit | Merged stackforge/puppet-tripleo: Rethink the backup option for Galera https://review.openstack.org/170021 | 20:34 |
| slagle | https://www.google.com/maps/place/White+Spot+Triple+O%27s/@49.286238,-123.121313,17z/data=!3m1!4b1!4m2!3m1!1s0x0:0xe68e3fbe1a9efe65 | 20:35 |
| slagle | now i wish i had gone to the one in Hong Kong. | 20:35 |
| *** Marga_ has joined #tripleo | 20:36 | |
| * bnemec clicks | 20:37 | |
| * bnemec is suddenly hungry for poutine | 20:37 | |
| * bnemec blames slagle | 20:37 | |
| *** jtomasek has quit IRC | 20:37 | |
| *** Marga_ has quit IRC | 20:38 | |
| greghaynes | Hello my redhat enlightened friends, I have been trying to battle http://paste.ubuntu.com/10785667/ which is what happens when I boot a dib made image using fedora-minimal (which makes the chroot using rpm --installroot) | 20:42 |
| greghaynes | the image is getting booted with virtio, and ive verified the initrd has the virtio and virtio_blk drivers... | 20:42 |
| greghaynes | (btw, the combined microcode + initrd fs format is really annoying to unpack :p) | 20:43 |
| greghaynes | wondering if people have any ideas | 20:44 |
| greghaynes | mordred: ^ :( | 20:45 |
| slagle | are you soure the root device actually has the cloudimg-rootfs label? | 20:46 |
| greghaynes | No, but - I notice its only listing the sr0 as found partitions | 20:47 |
| greghaynes | (I have config drive turned on) | 20:47 |
| *** pcrews has quit IRC | 20:49 | |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Add support for Glance RBD backend https://review.openstack.org/172032 | 20:50 |
| *** Marga_ has joined #tripleo | 20:51 | |
| *** pcrews has joined #tripleo | 20:53 | |
| *** sdake has quit IRC | 20:53 | |
| openstackgerrit | James Slagle proposed openstack/tripleo-heat-templates: Example of post-deploy hook doing rhel registration https://review.openstack.org/172065 | 20:54 |
| *** untriaged-bot has joined #tripleo | 21:00 | |
| untriaged-bot | Untriaged bugs so far: | 21:00 |
| untriaged-bot | https://bugs.launchpad.net/tripleo/+bug/1442256 | 21:00 |
| openstack | Launchpad bug 1442256 in tripleo "mariadb fails to start with port in use error" [Undecided,New] | 21:00 |
| *** untriaged-bot has quit IRC | 21:00 | |
| openstackgerrit | Dan Prince proposed openstack/tripleo-incubator: Don't hard code the use-ephemeral element https://review.openstack.org/172209 | 21:03 |
| *** thrash is now known as thrash|g0ne | 21:03 | |
| openstackgerrit | Dan Prince proposed openstack-infra/tripleo-ci: Set root disk == node disk for puppet jobs https://review.openstack.org/172210 | 21:07 |
| openstackgerrit | Merged stackforge/kolla: Updates Docker bin in Heat template https://review.openstack.org/168645 | 21:11 |
| *** jcoufal has joined #tripleo | 21:12 | |
| bnemec | greghaynes: You didn't do anything funny with the minimal image where like the virtio_blk driver would have been removed, right? | 21:16 |
| greghaynes | nah, I even inspected it and the initrd post build to make sure its there | 21:17 |
| greghaynes | because thats what ive been suspecting too | 21:17 |
| bnemec | Yeah, it looks like it got past the ramdisk at least. | 21:18 |
| greghaynes | hrm | 21:21 |
| *** alop has quit IRC | 21:29 | |
| *** alop has joined #tripleo | 21:37 | |
| *** eghobo has joined #tripleo | 21:39 | |
| *** radez is now known as radez_g0n3 | 21:42 | |
| SpamapS | It does?/win 8 | 21:43 |
| SpamapS | doh | 21:43 |
| greghaynes | the whole libvirt serial -> file thing is awesome until I want to drop into a busybox | 21:45 |
| greghaynes | bnemec: what makes you say it got past the initrd? | 21:48 |
| bnemec | greghaynes: It died on a kernel panic, not a dracut error. | 21:48 |
| greghaynes | ah | 21:49 |
| *** trown is now known as trown|outttypeww | 21:50 | |
| *** weshay has joined #tripleo | 21:58 | |
| *** rhallisey has quit IRC | 21:58 | |
| *** adrianopetrich_ has quit IRC | 22:00 | |
| *** eghobo_ has joined #tripleo | 22:00 | |
| *** eghobo has quit IRC | 22:03 | |
| *** adrianopetrich has joined #tripleo | 22:03 | |
| greghaynes | haha, it is totally virtio | 22:04 |
| greghaynes | change the libvirt disk definition to ide and it now finds an sda | 22:05 |
| *** sdake has joined #tripleo | 22:07 | |
| *** sdake_ has joined #tripleo | 22:10 | |
| *** sdake has quit IRC | 22:14 | |
| *** jcoufal has quit IRC | 22:14 | |
| *** rhallisey has joined #tripleo | 22:18 | |
| *** cwolferh has quit IRC | 22:26 | |
| *** akrivoka has quit IRC | 22:27 | |
| *** rlandy has quit IRC | 22:30 | |
| *** rlandy has joined #tripleo | 22:30 | |
| *** cwolferh has joined #tripleo | 22:34 | |
| *** rlandy has quit IRC | 22:37 | |
| *** Marga_ has quit IRC | 22:43 | |
| *** sdake_ has quit IRC | 22:47 | |
| *** prad has quit IRC | 22:48 | |
| *** eghobo_ has quit IRC | 23:04 | |
| openstackgerrit | Dan Sneddon proposed openstack/os-net-config: Add static IP assignment via template tokens in config.yaml https://review.openstack.org/169937 | 23:11 |
| *** Marga_ has joined #tripleo | 23:14 | |
| *** bvandenh has joined #tripleo | 23:16 | |
| *** alop has quit IRC | 23:19 | |
| *** yuanying has joined #tripleo | 23:23 | |
| *** bvandenh has quit IRC | 23:29 | |
| *** jcoufal has joined #tripleo | 23:34 | |
| *** Marga_ has quit IRC | 23:35 | |
| *** Marga_ has joined #tripleo | 23:36 | |
| *** Marga_ has quit IRC | 23:38 | |
| greghaynes | bnemec: I was just following a blog of yours without realizing it :) | 23:39 |
| *** prad has joined #tripleo | 23:39 | |
| greghaynes | So, I can get the image to boot from ide, and the virtio_blk driver is not loaded already but I can load it | 23:40 |
| greghaynes | im wondering if I somehow need to force the driver to be loaded in the initrd? | 23:40 |
| *** panda|afk has quit IRC | 23:43 | |
| *** prad has quit IRC | 23:44 | |
| *** panda|afk has joined #tripleo | 23:44 | |
| *** noslzzp has joined #tripleo | 23:46 | |
| *** Marga_ has joined #tripleo | 23:47 | |
| *** Marga_ has quit IRC | 23:48 | |
| *** Marga_ has joined #tripleo | 23:48 | |
| *** eghobo has joined #tripleo | 23:53 | |
| *** eghobo has quit IRC | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!