| *** Ryjedo_ has joined #tripleo | 00:02 | |
| *** Ryjedo has quit IRC | 00:03 | |
| *** Ryjedo_ is now known as Ryjedo | 00:03 | |
| *** dmacpher has joined #tripleo | 00:11 | |
| *** jobewan has quit IRC | 00:13 | |
| *** limao has joined #tripleo | 00:46 | |
| *** fragatina has joined #tripleo | 00:48 | |
| *** bana_k has joined #tripleo | 00:56 | |
| *** Goneri has joined #tripleo | 01:00 | |
| *** bfournie has joined #tripleo | 01:08 | |
| *** gregwork has quit IRC | 01:10 | |
| *** NachoDuck has quit IRC | 01:11 | |
| *** NachoDuck has joined #tripleo | 01:24 | |
| *** gregwork has joined #tripleo | 01:26 | |
| *** Goneri has quit IRC | 01:46 | |
| *** Goneri has joined #tripleo | 01:52 | |
| *** bana_k has quit IRC | 01:56 | |
| *** yamahata has quit IRC | 02:14 | |
| *** Goneri has quit IRC | 02:30 | |
| *** apetrich has quit IRC | 02:38 | |
| *** apetrich has joined #tripleo | 02:39 | |
| *** limao has quit IRC | 03:11 | |
| *** rhefner has joined #tripleo | 03:42 | |
| openstackgerrit | Karthik S proposed openstack/puppet-tripleo: Configure the numvfs for SRIOV interfaces https://review.openstack.org/361350 | 03:45 |
|---|---|---|
| *** bana_k has joined #tripleo | 03:57 | |
| *** limao has joined #tripleo | 04:10 | |
| *** chlong has quit IRC | 04:12 | |
| *** limao has quit IRC | 04:15 | |
| *** chlong has joined #tripleo | 04:25 | |
| *** coolsvap_ has joined #tripleo | 04:28 | |
| *** limao has joined #tripleo | 04:28 | |
| *** coolsvap_ is now known as coolsvap_\ | 04:28 | |
| *** coolsvap_\ is now known as coolsvap | 04:28 | |
| *** bana_k has quit IRC | 04:29 | |
| *** links has joined #tripleo | 04:33 | |
| *** numans has joined #tripleo | 04:34 | |
| *** fragatin_ has joined #tripleo | 04:45 | |
| *** fragati__ has joined #tripleo | 04:48 | |
| *** fragatina has quit IRC | 04:48 | |
| *** fragatin_ has quit IRC | 04:50 | |
| *** jaosorior has joined #tripleo | 04:52 | |
| *** fragati__ has quit IRC | 04:53 | |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: Create entries for overcloud VIPs in /etc/hosts https://review.openstack.org/357765 | 05:08 |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: Add HAProxy TLS handled by certmonger as composable service https://review.openstack.org/356430 | 05:12 |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: DO NOT MERGE - TESTING BARBICAN https://review.openstack.org/353262 | 05:12 |
| *** shardy has joined #tripleo | 05:17 | |
| *** pgadiya has joined #tripleo | 05:24 | |
| *** bana_k has joined #tripleo | 05:27 | |
| *** akuznetsov has joined #tripleo | 05:28 | |
| *** florianf has joined #tripleo | 05:40 | |
| Jokke_ | morning folks | 05:44 |
| *** skramaja has quit IRC | 05:45 | |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: Create entries for overcloud VIPs in /etc/hosts https://review.openstack.org/357765 | 05:45 |
| *** liverpooler has joined #tripleo | 05:50 | |
| *** skramaja has joined #tripleo | 05:50 | |
| *** dmacpher has quit IRC | 05:51 | |
| jaosorior | shardy: there was an error in heat that was causing some errors that look like: UPDATE_FAILED NotFound_Remote: resources[0]: Software config with id aff00d7a-88ea-4775-a247-77b69614e401 not found | 05:53 |
| jaosorior | was taht fixed? | 05:54 |
| *** aufi has joined #tripleo | 06:01 | |
| *** saneax-_-|AFK is now known as saneax | 06:06 | |
| *** leanderthal|afk is now known as leanderthal | 06:07 | |
| openstackgerrit | Babu Shanmugam proposed openstack/puppet-tripleo: Add ML2/OVN support for neutron profile https://review.openstack.org/314875 | 06:09 |
| shardy | jaosorior: Hi, yeah I saw the same thing on update, I believe the problem has been fixed in heat master, but not yet promoted to current-tripleo | 06:10 |
| shardy | jaosorior: https://review.openstack.org/#/c/360831/ was the final part of the fix I believe | 06:11 |
| *** akuznetsov has quit IRC | 06:14 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/os-collect-config: Updated from global requirements https://review.openstack.org/350905 | 06:16 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-tripleoclient: Updated from global requirements https://review.openstack.org/361875 | 06:17 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/tripleo-common: Updated from global requirements https://review.openstack.org/361501 | 06:18 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/tripleo-validations: Updated from global requirements https://review.openstack.org/361877 | 06:18 |
| Jokke_ | jaosorior, shardy: any chance to get +A for https://review.openstack.org/#/c/354047/ couple of +2s already and finally all tests green? | 06:25 |
| *** fragatina has joined #tripleo | 06:25 | |
| *** fragatina has quit IRC | 06:26 | |
| *** fragatina has joined #tripleo | 06:26 | |
| Jokke_ | and would really like to get this reviewed as well, last link on the chain https://review.openstack.org/#/c/358525/ | 06:27 |
| shardy | Jokke_: approved the first one | 06:28 |
| shardy | the pattern with defaulting the parameters to hiera for a different module is weird, but I see it's been discussed earlier in the series | 06:28 |
| Jokke_ | cheers ... yeah we had bit of back and forth how this should be done | 06:29 |
| shardy | https://review.openstack.org/#/c/358525/3/puppet/services/manila-base.yaml | 06:30 |
| shardy | why are we adding all backends to manila-base.yaml? | 06:30 |
| marios | morning o/ | 06:31 |
| shardy | I commented on another patch adding a netapp backend, we'll end up with one manila-base that contains all the backends | 06:31 |
| shardy | marios: aha, good morning, was just about to ping you re this manila stuff :) | 06:31 |
| Jokke_ | shardy: I followed the example of the previous work and what marios did in https://review.openstack.org/#/c/354019/5 ... so from my part that was for consistency | 06:32 |
| marios | shardy: i had a quick look at the comments thanks for looking. yeah so it is possible for multiple backends to be enabled at the same time. it is kinda weird, you're right. for example, having a dedicated 'ManilaGenericEnabled' and also 'ManilaNetapEnabled' | 06:32 |
| Jokke_ | morning marios | 06:32 |
| marios | shardy: so i thought it would somehow keep it simpler by having all the config in the base class, enabled only by that param from the environment file, | 06:33 |
| marios | Jokke_: o/ | 06:33 |
| shardy | marios: Yeah, but it's not that consistent with the new composable services interface | 06:33 |
| shardy | I'm wondering if there's a way to rework it so it's more like how we enable backends for neutron | 06:33 |
| marios | shardy: so it is just a matter of moving the config to a dedicated file. right. so one thing i wanted to do but didn't get to it before i left | 06:34 |
| marios | shardy: was to tidy up puppet-tripleo seide, since you may have seen, we are setting hiera from tht as well as re-resetting it on puppet-tripleo side unnecessarily | 06:34 |
| shardy | marios: given that we now have two backends trying to add to -base, do you think it's reasonable to just say we've got to break out the backends now? | 06:35 |
| marios | shardy: so yeah i will have another pass probably first a tidy up to include ^^^ and what you're saying about moving the generic stuff to dedicated service file | 06:35 |
| shardy | Yeah I noticed that | 06:35 |
| shardy | Ok, I'll +1 the patches and hopefully we can get those cleanups in | 06:35 |
| shardy | in general alarm bells start going off every time I see an EnableFoo parameter nowadays ;) | 06:35 |
| Jokke_ | :) | 06:36 |
| marios | shardy: so sure man, i will move/tidy up today at some point. problem is time, my involvement here is as you know 'assist' purely because I have other stuff that people will call me up on if i don't do :) | 06:36 |
| Jokke_ | I'm happy to attack it now if someone would give me a pointer how to get it done (would really like to see this work reaching Newton) | 06:36 |
| marios | shardy: yeah, this was the first questino i asked, about the enable foo | 06:36 |
| marios | shardy: there is some discussino on the tht | 06:36 |
| Jokke_ | fwiw I kind of liked the approach marios took. Looked way cleaner and easier to follow ;) | 06:37 |
| marios | shardy: i just copied what was in the original reviews at first, but clarified that it is indeed the case that we can have >1 backend enabled simultaneously. we could ofcourse make that a list param 'EnabledBackends' | 06:37 |
| *** mcornea has joined #tripleo | 06:37 | |
| *** bana_k has quit IRC | 06:37 | |
| marios | shardy: but i think that would add complexity to handle the list. we *do* have a dedicated env file for each backend, and the inclusion of either of those will deploy that backend | 06:38 |
| shardy | marios: Yeah, there is a heat feature to help with that, ramishra finished it, let me see if it landed now | 06:39 |
| shardy | marios: basically instead of defining the entire ControllerServices list, it's possible to define a merge strategy so your environment file just appends to the list | 06:39 |
| shardy | which will make this much easier/cleaner | 06:40 |
| shardy | https://review.openstack.org/#/q/status:open+project:openstack/heat+branch:master+topic:bp/environment-merging | 06:40 |
| shardy | two patches remaining to merge :( | 06:40 |
| shardy | I'll chase for reviews | 06:40 |
| marios | shardy: ack .. interesting thanks noted | 06:42 |
| shardy | marios: if we have to, I suppose we can land the monolithic -base stuff, provided folks are OK with pretty quickly refactoring to split out the backends | 06:43 |
| marios | shardy: sure, won't be for a couple hours/after lunch i have a look there so let's see if there are any other thoughts/opinions and i'll put up a revision and we can discuss there? | 06:44 |
| *** rasca has joined #tripleo | 06:44 | |
| shardy | marios: yup, sounds good - it's a public holiday here so I'll be inconsistently available, but will be around later to discuss further if needed | 06:44 |
| shardy | thanks! | 06:44 |
| marios | ack thanks shardy enjoy | 06:45 |
| *** rcernin has joined #tripleo | 06:47 | |
| *** jubapa has joined #tripleo | 06:47 | |
| Jokke_ | shardy & marios: as said I have the cycles to attack that if needed (I just need some pointers how to approach it). I'm fine for either one, hitting it (quite literally) now or doing/helping out the refactoring after my PTO (I'll be back at 12th of Sept) | 06:49 |
| Jokke_ | shardy: enjoy the bank holiday. Thanks for jumping in even during it. | 06:50 |
| marios | Jokke_: thanks feel free to do so i will ping you later when i am going to look /and/or you update it before then thanks for any help | 06:51 |
| marios | Jokke_: wrt pointers, maybe more specific question/unknown ? i guess the one question now is whether we keep the config in base and iterate, or we split. the split is simple enough ... look at neutron as shardy says | 06:52 |
| marios | Jokke_: 'simple enough' ... * t&c apply I mean should be but i can't see why it wouldn't be, need to move into dedicated files for each backend | 06:53 |
| marios | Jokke_: so either way, it would be very useful if you would pose that question on the review for other reviewers as a start ? | 06:53 |
| Jokke_ | marios: ok, I'll have a look into how the neutron is done and if it is simple enough to understand I propose new PS to that ceph driver first so we can see how it works | 06:53 |
| marios | Jokke_: ceph driver? | 06:54 |
| Jokke_ | marios: https://review.openstack.org/#/c/354047/7 | 06:54 |
| shardy | there's a ceph and netapp manila backend/driver, both modifying manila-base | 06:54 |
| marios | Jokke_: shardy ah thanks I missed that part of the discussion/those reviews | 06:55 |
| shardy | can you have multiple manila backends? | 06:55 |
| marios | shardy: yes | 06:55 |
| marios | shardy: simultaneously apparently | 06:55 |
| Jokke_ | shardy: yes and that's the direction we have been working this towards | 06:55 |
| marios | shardy: see qestion in the early comments at https://review.openstack.org/#/c/354019/ | 06:56 |
| marios | shardy: and discussino with tbarron | 06:56 |
| shardy | Ok, thanks, will check the earlier discussion | 06:56 |
| * Jokke_ has no idea how one is supposed to schedule those, but that's Manila's problem, not ours. The request was to provide integration for these drivers and wishing that multiple drivers could be enabled simultaneously so apparently they can handle it | 06:56 | |
| *** dsariel has joined #tripleo | 06:56 | |
| *** akuznetsov has joined #tripleo | 06:57 | |
| *** jprovazn has joined #tripleo | 06:59 | |
| *** pcaruana has joined #tripleo | 07:00 | |
| jaosorior | easy commit updating help text if anyone feel like merging code :D https://review.openstack.org/#/c/361020/ | 07:01 |
| *** tesseract- has joined #tripleo | 07:01 | |
| *** akuznetsov has quit IRC | 07:02 | |
| jaosorior | shardy: aww, still getting that error even with the code you passed me :( | 07:04 |
| shardy | jaosorior: did you pull that and all of the related patches? | 07:05 |
| jaosorior | ah | 07:05 |
| shardy | I think there were at least two patches related to that bug | 07:05 |
| jaosorior | not the related patches, no | 07:05 |
| jaosorior | only that | 07:05 |
| shardy | I'd check the bug to make sure you got them all or pull latest master heat | 07:05 |
| *** panda has joined #tripleo | 07:06 | |
| jaosorior | shardy: I think I rather just get the latest heat | 07:06 |
| jaosorior | shardy: is there a way to fetch it from the repos? | 07:07 |
| *** anshul has joined #tripleo | 07:08 | |
| *** ohamada has joined #tripleo | 07:09 | |
| shardy | jaosorior: I just build a delorean package normally, then update the installed version on the undercloud | 07:09 |
| shardy | jaosorior: Or, you can add it to includpkgs in the delorean-current yum.repos.d file | 07:09 |
| shardy | includepkgs | 07:09 |
| jaosorior | shardy: alright, thanks dude | 07:10 |
| *** nyechiel has joined #tripleo | 07:11 | |
| *** jpena|off is now known as jpena | 07:13 | |
| *** tremble has joined #tripleo | 07:14 | |
| jaosorior | anybody up for merging this commit by shardy? https://review.openstack.org/#/c/361029/ :D | 07:15 |
| *** ifarkas_afk is now known as ifarkas | 07:16 | |
| shardy | there are two more patches in that series too which are passing CI and ready for review: | 07:17 |
| shardy | https://review.openstack.org/#/c/361578 | 07:17 |
| shardy | https://review.openstack.org/#/c/348974 | 07:17 |
| shardy | reviews appreciated! :) | 07:17 |
| shadower | shardy: the first one lgtm. May I? | 07:18 |
| shardy | shadower: sure, thanks! | 07:19 |
| *** atom_ has joined #tripleo | 07:19 | |
| marios | shardy: ack i have https://review.openstack.org/#/c/348974 in my list for a revisit, but looks like i will have to stare at the tht filters/yaql for a while but will get to it. also, if you're not going to use your day off, can i have it please? | 07:19 |
| shardy | lol :) | 07:20 |
| * shardy is planning to use some of his day off shortly :D | 07:20 | |
| shadower | haha | 07:20 |
| shardy | marios: Yeah, some of the nested intrinsic functions are a little hairy, I'm planning a blog post to hopefully demystify it | 07:21 |
| shadower | shardy: btw the mistral validation series is passing CI, has a +2 and several +1s and is good to go imho: https://review.openstack.org/#/c/353898/ | 07:21 |
| shardy | when you break it down it's not that hard, but combined I agree it's slightly brain teasing | 07:21 |
| marios | shardy: cool thanks. yeah i saw a much earlier versino of that where you had a couple of filters but i think you have changed it now so make it much more complicated | 07:21 |
| marios | simpler! | 07:21 |
| marios | simpler :) | 07:21 |
| openstackgerrit | Merged openstack/puppet-tripleo: Add Manila CephFS backend to manila class https://review.openstack.org/354047 | 07:22 |
| shardy | shadower: thanks, I'll check it out | 07:22 |
| shadower | thanks! | 07:22 |
| shardy | we need to figure out how to stop the ovb jobs timing out all the time ref the POST_FAILURE's :( | 07:23 |
| shardy | gtg for a bit, bbl | 07:26 |
| *** shardy has quit IRC | 07:26 | |
| Jokke_ | and Manila folks would be really happy if this one-liner merged, has +2 from shardy already and finally passed all the tests so should be ~freebie https://review.openstack.org/#/c/355394/ | 07:27 |
| *** r-mibu has quit IRC | 07:27 | |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: Create entries for overcloud VIPs in /etc/hosts https://review.openstack.org/357765 | 07:27 |
| *** fzdarsky has joined #tripleo | 07:29 | |
| *** akuznetsov has joined #tripleo | 07:30 | |
| *** karthiks_afk is now known as karthikc | 07:31 | |
| *** karthikc is now known as karthiks | 07:31 | |
| *** jpich has joined #tripleo | 07:37 | |
| Jokke_ | thanks marios :) | 07:37 |
| *** hewbrocca-afk is now known as hewbrocca | 07:38 | |
| *** ioggstream has joined #tripleo | 07:39 | |
| *** zoli_gone-proxy is now known as zoliXXL | 07:39 | |
| marios | np Jokke_ :) | 07:40 |
| hewbrocca | Morning folks | 07:40 |
| zoliXXL | good morning | 07:41 |
| jaosorior | jistr: ping | 07:43 |
| hewbrocca | marios: welcome back! | 07:45 |
| marios | hewbrocca: o/ | 07:47 |
| marios | thanks :) | 07:47 |
| hewbrocca | gfidente was trying to make more bugs while you were away | 07:48 |
| hewbrocca | but I told him to stop it | 07:48 |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/tripleo-ci: Increase timeout for ironic power status report https://review.openstack.org/361915 | 07:48 |
| *** zoliXXL is now known as zoli|brb | 07:48 | |
| jaosorior | sshnaidm: should you be using single quote there? https://review.openstack.org/#/c/361915/1/scripts/tripleo.sh | 07:49 |
| marios | thanks hewbrocca. but you know, i think we have to accept that it is futile. the gfidente has been released. there is no earthly power to restrain the gfidente. such is the nature of, the gfidente. :( | 07:49 |
| hewbrocca | ROFL | 07:50 |
| sshnaidm | jaosorior, not sure, but singles are more safe, thanks | 07:50 |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/tripleo-ci: Increase timeout for ironic power status report https://review.openstack.org/361915 | 07:51 |
| jaosorior | sshnaidm: also, I suggest you do that here instead https://github.com/openstack-infra/tripleo-ci/blob/master/scripts/deploy.sh#L18 | 07:54 |
| jaosorior | sshnaidm: that way it will be persisted on the undercloud and will even work if we get an update gate | 07:54 |
| *** iogg has joined #tripleo | 07:54 | |
| *** ioggstream has quit IRC | 07:55 | |
| sshnaidm | jaosorior, yeah, it seems better, what is path to this parameter though? | 07:55 |
| jaosorior | sshnaidm: I'm figuring that out. | 07:56 |
| jaosorior | sshnaidm: is post_deploy_get_power_state_retries in the DEFAULT group? | 07:58 |
| sshnaidm | jaosorior, no, it's in [agent[ | 07:58 |
| jaosorior | ok | 07:59 |
| jaosorior | you can use this | 07:59 |
| *** tvignaud has quit IRC | 07:59 | |
| jaosorior | https://github.com/openstack/puppet-ironic/blob/master/manifests/config.pp | 07:59 |
| *** oshvartz has joined #tripleo | 07:59 | |
| jaosorior | sshnaidm: http://paste.openstack.org/show/564508/ | 07:59 |
| jaosorior | sshnaidm: but the best thing to do is to actually add post_deploy_get_power_state_retries to puppet-ironic | 08:00 |
| sshnaidm | jaosorior, cool, so I can change any value in configurations, right? | 08:00 |
| jaosorior | sshnaidm: yeah | 08:00 |
| jaosorior | and actually I wrote it wrong | 08:00 |
| jaosorior | this | 08:01 |
| jaosorior | http://paste.openstack.org/show/564509/ | 08:01 |
| jaosorior | sshnaidm: ^^ | 08:01 |
| jaosorior | sshnaidm: but yeah, the best thing to do is to add that parameter to puppet-ironic. That way if someone adds that parameter there, they won't break our CI | 08:01 |
| *** r-mibu has joined #tripleo | 08:03 | |
| *** mpsairam has quit IRC | 08:03 | |
| d0ugal | Morning! | 08:03 |
| sshnaidm | jaosorior, I'm not sure I can do it, puppet modules are still in my tolearn list | 08:03 |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/tripleo-ci: Increase timeout for ironic power status report https://review.openstack.org/361915 | 08:05 |
| jaosorior | sshnaidm: alright, lets see if we can manage with this for now. I'll try to poke people into how to add taht to pupept-ironic | 08:05 |
| sshnaidm | jaosorior, do you have any example maybe? | 08:06 |
| *** links has quit IRC | 08:06 | |
| jaosorior | sshnaidm: so, for instance, here I added some parameters to puppet-nova https://review.openstack.org/#/c/359604/ | 08:07 |
| *** dsariel has quit IRC | 08:09 | |
| sshnaidm | jaosorior, ok, maybe next time :) | 08:09 |
| *** dsariel has joined #tripleo | 08:09 | |
| *** derekh has joined #tripleo | 08:12 | |
| *** vijaykc4 has joined #tripleo | 08:12 | |
| *** tvignaud has joined #tripleo | 08:13 | |
| *** greghaynes has quit IRC | 08:13 | |
| *** lucas-afk is now known as lucasagomes | 08:21 | |
| *** coolsvap is now known as coolsvap_ | 08:21 | |
| *** pblaho has joined #tripleo | 08:21 | |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: [WIP] Set parameters https://review.openstack.org/360391 | 08:26 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Add 'openstack overcloud plan create' command https://review.openstack.org/359784 | 08:26 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Print the execution ID when starting Mistral workflows https://review.openstack.org/340711 | 08:26 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: [WIP] Add `openstack overcloud deploy --plan` https://review.openstack.org/360305 | 08:26 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Update the 'overcloud deploy --templates' command to use Mistral https://review.openstack.org/342261 | 08:26 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Migrate to using osc-lib https://review.openstack.org/335460 | 08:28 |
| jistr | jaosorior: pong, hi | 08:28 |
| openstackgerrit | Christian Schwede proposed openstack/tripleo-heat-templates: Enable Swift account_quotas middleware by default https://review.openstack.org/344755 | 08:29 |
| jaosorior | jistr: hey dude, was re-reading your comment here https://review.openstack.org/#/c/359038/3/manifests/profile/base/haproxy.pp | 08:29 |
| openstackgerrit | Dougal Matthews proposed openstack/tripleo-common: Fix an autoclass reference and add missing pages to the toctree https://review.openstack.org/342747 | 08:30 |
| jaosorior | jistr: I could add the balancermember reload. But do you think I should go for the refresh on the concat? I think that might be a bit of overkill and quite focused on an implementation detail, no? | 08:30 |
| openstackgerrit | Merged openstack/tripleo-ui: Fix multiple default exports in single module https://review.openstack.org/360641 | 08:31 |
| openstackgerrit | Dougal Matthews proposed openstack/tripleo-common: Better namespace the list of TripleO actions https://review.openstack.org/357682 | 08:32 |
| * d0ugal is rebasing all the things | 08:32 | |
| jistr | jaosorior: yea true that it would rely on puppet-haproxy implementation. I'm fine with whichever you prefer. | 08:33 |
| jaosorior | jistr: alright, just wanted to know your opinion | 08:33 |
| openstackgerrit | Jiri Tomasek proposed openstack/tripleo-ui: Refactor DeploymentPlan Component https://review.openstack.org/358814 | 08:33 |
| openstackgerrit | Jiri Tomasek proposed openstack/tripleo-ui: Mistral Actions error handling https://review.openstack.org/359764 | 08:33 |
| openstackgerrit | Jiri Tomasek proposed openstack/tripleo-ui: Parameters listing and updating https://review.openstack.org/351554 | 08:33 |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/puppet-tripleo: Reload HAProxy on refresh https://review.openstack.org/359038 | 08:35 |
| jaosorior | jistr: done ^^ | 08:35 |
| jistr | +2'd, thanks | 08:36 |
| jaosorior | jistr: thanks dude | 08:37 |
| *** zoli|brb is now known as zoli | 08:37 | |
| *** zoli is now known as zoliXXL | 08:37 | |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/tripleo-ci: Disable sahara tests in tempest https://review.openstack.org/361936 | 08:38 |
| *** pgadiya has quit IRC | 08:38 | |
| *** greghaynes has joined #tripleo | 08:40 | |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Create hiera service_enabled keys for enabled services https://review.openstack.org/361029 | 08:40 |
| hewbrocca | Folks, shardy was looking for reviews on these: https://review.openstack.org/#/c/361029/ | 08:40 |
| hewbrocca | https://review.openstack.org/#/c/361578 | 08:40 |
| hewbrocca | https://review.openstack.org/#/c/348974 | 08:40 |
| hewbrocca | Passing CI, ready to look at | 08:41 |
| hewbrocca | he's out today so I am advocating on his behalf | 08:41 |
| *** jbadiapa has joined #tripleo | 08:41 | |
| *** pgadiya has joined #tripleo | 08:41 | |
| openstackgerrit | Merged openstack/tripleo-puppet-elements: Install OpenStack Manila UI https://review.openstack.org/355394 | 08:41 |
| shadower | hewbrocca: #1 is merged, #2 waiting for the gate | 08:42 |
| hewbrocca | Excellent | 08:44 |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: Add flag to assert that puppet manages the keystone endpoints https://review.openstack.org/353307 | 08:45 |
| shadower | hewbrocca: is he really out today? /me has talked to him couple hours ago | 08:45 |
| *** abregman has joined #tripleo | 08:47 | |
| hewbrocca | shadower: well, it's a bank holiday in the UK today | 08:47 |
| hewbrocca | so I think he's hiding | 08:47 |
| hewbrocca | Now these two appear to be gfidente fault: | 08:48 |
| hewbrocca | https://review.openstack.org/#/c/355067 | 08:48 |
| hewbrocca | https://review.openstack.org/#/c/355068/ | 08:48 |
| hewbrocca | 08:48 | |
| *** akrivoka has joined #tripleo | 08:48 | |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Rename opendaylight service to opendaylight_api https://review.openstack.org/361578 | 08:48 |
| openstackgerrit | Babu Shanmugam proposed openstack/puppet-tripleo: Add ML2/OVN support for neutron profile https://review.openstack.org/314875 | 08:49 |
| shadower | Does this need a rechec? https://review.openstack.org/#/c/353898/ | 08:50 |
| shadower | Jenkins says +1 and the "Jenkins check" runs pass but the "Jenkins" ones are POST_FAILURE | 08:51 |
| jaosorior | shadower: I think it does. both ha and nonha got that :/ | 08:53 |
| shadower | will do, thanks jaosorior | 08:53 |
| *** gfidente has joined #tripleo | 08:53 | |
| *** gfidente has quit IRC | 08:54 | |
| *** gfidente has joined #tripleo | 08:54 | |
| jaosorior | shadower: even though ha passed the pingtest, it timed out shortly after. So that's not a big deal. BUT, in that run, nonha timed out in the introspection. So yeah, I think recheck is necessary | 08:54 |
| openstackgerrit | Jiri Tomasek proposed openstack/tripleo-ui: Migrate Deploy action to Mistral https://review.openstack.org/357125 | 08:54 |
| shadower | jaosorior: right. I was a bit confused because of the +1 and the fact that the two checks are separate from the rest | 08:54 |
| jaosorior | shadower: yeah. Jenkins doesn't take into account the ovb ha and nonha for voting | 08:55 |
| shadower | right | 08:55 |
| *** electrofelix has joined #tripleo | 09:00 | |
| *** vijaykc4 has quit IRC | 09:07 | |
| jistr | btw folks i'm a bit confused -- i saw some mentions of feature freeze today, but i think the official OpenStack one is Sept 1? When do we aim for feature freeze? | 09:08 |
| hewbrocca | My understanding was that the official freeze is August 29 -- but I have no basis for that other than somebody said it at some point | 09:08 |
| jaosorior | jistr: that's a great question. Cause it seems I need to apply for an FFE O_o | 09:08 |
| hewbrocca | jaosorior: alternatively, you could just land all your patches :D | 09:09 |
| jaosorior | hewbrocca: interesting proposition O_O | 09:10 |
| Jokke_ | I have heard everyone talking about today, official date 'though by the release schedule is 1st as that's the Thu of FF week where all dates defaults unless otherwise specified and with FF there is no such specification | 09:11 |
| jistr | ^^ that | 09:12 |
| jpich | That's my experience as well, I think folks are just hoping to merge things earlier in the week so the branch can be cut on Thursday without having to wait for the gate queue to empty | 09:13 |
| *** sshnaidm is now known as sshnaidm|afk | 09:13 | |
| Jokke_ | jpich: the stable branches should not be cut for 2 weeks still (that's aligned with -RC1 not with -3) | 09:14 |
| jistr | btw i found this http://lists.openstack.org/pipermail/openstack-dev/2016-August/102303.html | 09:14 |
| jpich | Jokke_: I guess I meant apply the tag then? | 09:14 |
| Jokke_ | jpich: that works :P | 09:14 |
| jpich | :P | 09:15 |
| gfidente | Jokke_, commented on the manila submission | 09:15 |
| *** oshvartz has quit IRC | 09:15 | |
| gfidente | I think that's an issue we haven't solved in neutron because neutron plugins are exclusive | 09:15 |
| Jokke_ | thanks gfidente! | 09:16 |
| gfidente | I wanted to add a reference to the services _enabled which shardy did though | 09:17 |
| gfidente | cause I think it will be useful | 09:17 |
| gfidente | can't find it | 09:17 |
| *** limao has quit IRC | 09:17 | |
| gfidente | it's basically a mechanism to push in hiera a list of _enabled profiles | 09:17 |
| Jokke_ | gfidente: I was looking that neutron implementation for past couple of hours and really couldn't figure out how that would work. Thanzks for confirming thta it wouldn't | 09:17 |
| jaosorior | gfidente: it merged | 09:18 |
| gfidente | yeah https://review.openstack.org/#/c/361029/ | 09:18 |
| gfidente | jaosorior tks :) | 09:18 |
| jaosorior | gfidente: here it is https://review.openstack.org/#/c/361029/ | 09:18 |
| jaosorior | ah, I was too slow :P | 09:19 |
| gfidente | jaosorior morning :) | 09:19 |
| *** akuznetsov has quit IRC | 09:19 | |
| Jokke_ | yeah, I'd like to understand what we want to do there and would be happy to work towards it. But something shardy referred this morning apparently is still missing couple of patches before it can be consumed here | 09:19 |
| jaosorior | gfidente: morning indeed | 09:19 |
| jaosorior | Jokke_: what's the issue? | 09:19 |
| Jokke_ | 08:39 < shardy> marios: basically instead of defining the entire ControllerServices list, it's possible to define a merge strategy so your environment file just appends to the list | 09:20 |
| Jokke_ | 08:40 < shardy> which will make this much easier/cleaner | 09:20 |
| Jokke_ | 08:40 < shardy> https://review.openstack.org/#/q/status:open+project:openstack/heat+branch:master+topic:bp/environment-merging | 09:20 |
| *** limao has joined #tripleo | 09:21 | |
| gfidente | I am going to use that for ceph/osd on computes if I can figure how :) | 09:21 |
| gfidente | just do add more bugs | 09:21 |
| jaosorior | Jokke_: is the manila backend exclusive? Or can you have mutliple backends there? | 09:23 |
| Jokke_ | so by the looks of it the two missing pieces merged into heat this morning | 09:23 |
| Jokke_ | jaosorior: there can be multiple simultanious ones | 09:23 |
| jaosorior | I see | 09:23 |
| gfidente | jaosorior, same problem is with cinder | 09:23 |
| gfidente | I was thinking if we shouldn't address this in the puppet module | 09:23 |
| gfidente | and build there the list of enabled backends from the enabled drivers | 09:24 |
| jaosorior | gfidente, Jokke_ cause the way I'm dealing with a pluggable option in HAProxy is to add another substack https://review.openstack.org/#/c/356430/8/puppet/services/haproxy.yaml | 09:24 |
| jaosorior | gfidente, Jokke_ but then again, I don't have the requirement of multiple backends there | 09:24 |
| Jokke_ | jaosorior: yeah ... and how would you do the same dynamically? | 09:25 |
| jaosorior | the word dynamically is a bit overused... what do you mean? | 09:26 |
| gfidente | ramishra, you around? | 09:27 |
| Jokke_ | jaosorior: so how do you do that based on one or multiple environment files listed on the call? | 09:27 |
| hewbrocca | gfidente: you're looking at https://review.openstack.org/#/c/355067 | 09:28 |
| hewbrocca | https://review.openstack.org/#/c/355068/ right? | 09:28 |
| hewbrocca | shardy explicitly mentioned those need to merge for comp roles | 09:28 |
| *** oshvartz has joined #tripleo | 09:28 | |
| hewbrocca | jpich: shardy also asked one other thing, if you have time: | 09:29 |
| hewbrocca | Then I can convert this tripleoclient hack to a mistral action and add it | 09:29 |
| hewbrocca | to the deployment workflow: | 09:29 |
| hewbrocca | https://review.openstack.org/#/c/336625/ | 09:29 |
| hewbrocca | 09:29 | |
| hewbrocca | We could perhaps ask someone (rbrady or jpich?) to help create the mistral | 09:29 |
| hewbrocca | action should they have time to do it, in theory it should be a pretty | 09:29 |
| hewbrocca | simple cut/paste from my patch. If not then I'll do it tomorrow. | 09:29 |
| hewbrocca | 09:29 | |
| Jokke_ | Also while I do understand shardy's comment of composability, I kind of disagree the need of backends of composable roles being composable. At least with current capabilities this seems like cleaner approach | 09:29 |
| ramishra | gfidente: hi | 09:30 |
| jaosorior | Jokke_: so on my path it's just a matter of adding the environment file, and it will replace that heat stack from None to the one I'm giving there. But yeah, like I said, it only works with one backend. | 09:30 |
| Jokke_ | jaosorior: yeah | 09:30 |
| *** pgadiya has quit IRC | 09:30 | |
| jaosorior | Jokke_: I guess you could do some yaql trickery. But it would overcomplicate things. I'm fine with what you posted. | 09:31 |
| gfidente | ramishra, can you comment on https://review.openstack.org/#/c/338113/ to see how we can make it merge a single line change into ComputeServicesList ? | 09:32 |
| ramishra | gfidente: sure | 09:33 |
| gfidente | ramishra, thanks :) | 09:34 |
| ramishra | gfidente: heatclient patch for merge strategies is not merged yet though. | 09:34 |
| gfidente | ramishra, is it expected to land soon? also read as, can we count on it be available? | 09:35 |
| ramishra | yep, n3 | 09:35 |
| *** flaper87 has quit IRC | 09:35 | |
| openstackgerrit | Merged openstack/tripleo-heat-templates: M/N upgrade set scheduler_host_manager right. https://review.openstack.org/358005 | 09:36 |
| ramishra | panned for n3, just someone has to approve it. But, I don't know if tripleo would need a new heatclient release or not. | 09:36 |
| jistr | ready for landing and awesome https://review.openstack.org/#/c/348974 (another spot where we got rid of misusing Heat->Hiera bug... one day we'll be able to fix it!!) | 09:36 |
| jpich | hewbrocca: Thanks! dbecker also pointed me to this. I'm having a look but I haven't really worked on deployments yet and I'm afraid a lot of it still looks like black magic to me. I'll see how far I can get, if anywhere, before rbrady gets up but I can't say I'm super confident about it | 09:36 |
| jpich | Sorry! | 09:36 |
| *** tosky has joined #tripleo | 09:37 | |
| gfidente | jistr, just w+1 | 09:38 |
| jistr | w00t | 09:38 |
| *** social has quit IRC | 09:42 | |
| *** social has joined #tripleo | 09:43 | |
| marios | jistr: gfidente i was reviewing that, but in any case would not have +2 yet ;) | 09:43 |
| marios | jistr: gfidente i mean https://review.openstack.org/#/c/348974/6 | 09:44 |
| *** dtantsur|afk is now known as dtantsur | 09:44 | |
| marios | jistr: gfidente also o/ :) | 09:45 |
| *** social has quit IRC | 09:46 | |
| gfidente | marios, mongo name has been replaced by mongodb in puppet-tripleo | 09:46 |
| *** xuao has joined #tripleo | 09:46 | |
| *** xuao has quit IRC | 09:47 | |
| *** xuao has joined #tripleo | 09:47 | |
| jistr | marios, gfidente: yea in fact in the original code removed by the patch ^^ it had mongo_node_ips as heat param but it generated mongodb_node_ips hiera key | 09:48 |
| jistr | so all should be well? | 09:48 |
| *** vijaykc4 has joined #tripleo | 09:49 | |
| *** vijaykc4 has quit IRC | 09:51 | |
| *** oshvartz has quit IRC | 09:51 | |
| dtantsur | morning folks! how much time do we have left before the feature freeze? | 09:52 |
| marios | jistr: gfidente ack it was the discrepancy between the hiera key and heat param in all-nodes-config that caught my eye so thanks, seems ok then | 09:52 |
| *** flaper87 has joined #tripleo | 09:53 | |
| *** flaper87 has quit IRC | 09:54 | |
| *** flaper87 has joined #tripleo | 09:54 | |
| *** vijaykc4 has joined #tripleo | 09:54 | |
| *** limao has quit IRC | 09:55 | |
| shadower | d0ugal: a docs job is now failing because this: https://review.openstack.org/#/c/341572/ expects "input" in every workflow | 09:55 |
| shadower | d0ugal: but e.g. this doesn't take any so the key is not there: https://review.openstack.org/#/c/353898/5/workbooks/validations.yaml | 09:55 |
| shadower | d0ugal: is that a bug in the sphinx extension (just checking, happy to write) | 09:55 |
| *** zigo_ is now known as zigo | 09:56 | |
| d0ugal | shadower: dang, yeah, it is. I can update it if you like. Should be super easy. | 09:56 |
| d0ugal | shadower: just give me 1 min | 09:56 |
| shadower | ok | 09:56 |
| *** iogg has quit IRC | 09:58 | |
| *** iogg has joined #tripleo | 09:59 | |
| hewbrocca | dtantsur: it's a ticking time bomb | 09:59 |
| openstackgerrit | Derek Higgins proposed openstack/tripleo-heat-templates: Remove outdated upgrade steps https://review.openstack.org/361276 | 09:59 |
| dtantsur | hewbrocca, yeah, I get it. I need to plan my ironic work accordingly. We've landed all required bits, but there quite a few optional ones. | 10:00 |
| hewbrocca | folks ^^^ merge early, merge often | 10:00 |
| *** vijaykc4 has quit IRC | 10:02 | |
| openstackgerrit | Dougal Matthews proposed openstack/tripleo-common: Don't expect input to be in every workflow in doc generation https://review.openstack.org/361984 | 10:03 |
| d0ugal | shadower: ^ | 10:03 |
| *** oshvartz has joined #tripleo | 10:03 | |
| shadower | d0ugal: nice one, thanks! | 10:04 |
| d0ugal | shadower: FWIW, we have avoided using workflows when they just call one action - users are expected to call the action directly | 10:04 |
| d0ugal | shadower: (I am not a fan of this, I prefer your way, but I was the only one) | 10:04 |
| *** vijaykc4 has joined #tripleo | 10:05 | |
| shadower | d0ugal: hm, good to know | 10:05 |
| *** vijaykc4 has quit IRC | 10:06 | |
| gfidente | jistr, so as I understand it hiera() will return string anyway | 10:06 |
| gfidente | but any2array should do? | 10:06 |
| jistr | gfidente: btw as shardy wrote about the interpolation here https://review.openstack.org/#/c/361108/2/manifests/profile/base/ceph.pp -- do you know if we need that somewhere? I don't think it can produce anything else than a string, regardless what the input is. | 10:06 |
| *** coolsvap_ is now known as coolsvap | 10:06 | |
| gfidente | nice was writing about same thing in same moment :) | 10:06 |
| jistr | hehe yea :)) | 10:07 |
| d0ugal | shadower: but maybe if we can sneak that in then I have an excuse to make the others match it ;) | 10:07 |
| jpich | d0ugal: 0_0 | 10:08 |
| shadower | haha | 10:08 |
| shadower | d0ugal: well as soon as it passes CI, it's yours to +A :-) | 10:08 |
| jaosorior | derekh: I actually don't see the pki directory in the dump we get of the overcloud nodes. What's up with that? | 10:09 |
| derekh | jaosorior: looking | 10:10 |
| openstackgerrit | Dmitry Tantsur proposed openstack/instack-undercloud: Handling defaults for pxe_bootfile_name and pxe_config_template moved to puppet https://review.openstack.org/361165 | 10:10 |
| d0ugal | shadower, jpich - hah, well, at this point I don't want to suggest changing it as I assume the UI is already using that approach | 10:10 |
| dtantsur | lucasagomes, mind taking a look please ^^^ | 10:10 |
| shadower | yea | 10:10 |
| * lucasagomes looks | 10:11 | |
| jpich | d0ugal: tbh if you create the one-action-workflows so they're available it probably works too, that won't prevent anyone from calling the actions directly | 10:11 |
| lucasagomes | dtantsur, btw you will need to rebase that I believe | 10:11 |
| lucasagomes | due https://review.openstack.org/#/c/350627/ | 10:11 |
| d0ugal | jpich: Indeed, and then the workflows could be the documented interface - if you want to dive behind that, go for it. | 10:12 |
| dtantsur | lucasagomes, just rebased | 10:12 |
| lucasagomes | dtantsur, a-ha, right lgtm | 10:12 |
| d0ugal | Anyone got a moment to review an if statement? https://review.openstack.org/#/c/361984/ :) | 10:12 |
| jpich | d0ugal: Hmmm I'd discuss a bit more widely before implying that only workflow calls are "supported"/"approved" and the actions are unstable | 10:13 |
| d0ugal | jpich: they are already unstable because we don't version them like we do with workflows ... ;) | 10:13 |
| *** zoliXXL is now known as zoli|lunch | 10:14 | |
| derekh | jaosorior: ahh I see it now http://git.openstack.org/cgit/openstack-infra/tripleo-ci/tree/scripts/common_vars.bash#n22 | 10:14 |
| derekh | jaosorior: --exclude=etc/pki | 10:14 |
| derekh | jaosorior: so the patch is be used I guess? | 10:14 |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack-infra/tripleo-ci: Inject CA certificate via service profile https://review.openstack.org/357791 | 10:14 |
| jaosorior | derekh: no. I just figured out a problem | 10:14 |
| jaosorior | derekh: thanks for checking it out | 10:15 |
| derekh | jaosorior: no prob, looks like I came to the correct conclusion for the wrong reason ... | 10:15 |
| openstackgerrit | Dmitry Tantsur proposed openstack/instack-undercloud: Stop using deprecated ironic::drivers::deploy https://review.openstack.org/361291 | 10:16 |
| *** pgadiya has joined #tripleo | 10:16 | |
| jaosorior | derekh: it was a good review either way :D | 10:17 |
| jaosorior | dtantsur: hey man, currently nothing under the "agent" section in the ironic configuration can be set through puppet-ironic, right?\ | 10:18 |
| dtantsur | jaosorior, I'm afraid you're right | 10:18 |
| jaosorior | dtantsur: I see. So currently we don't tweak any of that in tripleo? | 10:18 |
| dtantsur | I guess so | 10:19 |
| dtantsur | what exactly do you have in mind? | 10:19 |
| openstackgerrit | Lukas Bezdicka proposed openstack/instack-undercloud: Use ironic::inspector::ipxe class to setup iPXE https://review.openstack.org/345980 | 10:19 |
| openstackgerrit | Lukas Bezdicka proposed openstack/instack-undercloud: Update puppet dependancies before running puppet https://review.openstack.org/328264 | 10:19 |
| *** paramite has joined #tripleo | 10:20 | |
| openstackgerrit | Lukas Bezdicka proposed openstack/instack-undercloud: Use ironic::pxe class to setup PXE https://review.openstack.org/345980 | 10:20 |
| openstackgerrit | Lukas Bezdicka proposed openstack/instack-undercloud: Update puppet dependancies before running puppet https://review.openstack.org/328264 | 10:20 |
| jaosorior | dtantsur: I was just wondering how necessary is it to add those to puppet-ironic. This commit https://review.openstack.org/#/c/361915/3/scripts/deploy.sh sets post_deploy_get_power_state_retries which we needed to tweak (apparently) for the undercloud. And I was thinking if we should leave it like that or actually get a proper parameter in the puppet manifests for it | 10:20 |
| dtantsur | jaosorior, ugh, this is ugly. please do expose it in puppet-ironic | 10:21 |
| *** vijaykc4 has joined #tripleo | 10:21 | |
| dtantsur | otherwise we'll break tripleo CI if we start exposing it | 10:21 |
| jaosorior | dtantsur: you know someone that might have time to add this to puppet-ironic? This just came to my mind when reviewing that code, but unfortunately am quite caught with other things/ | 10:22 |
| dtantsur | jaosorior, or in this case I'd even consider increasing the ironic default | 10:22 |
| dtantsur | jaosorior, I can do it in 15-20 minutes :) | 10:23 |
| dtantsur | but if someone wants to increase the upstream default, they should bring some data to show the problem | 10:23 |
| Jokke_ | gfidente, jaosorior: added comment to https://review.openstack.org/#/c/358525/ Would appreciate your insight if I got it right | 10:27 |
| shadower | Folks, basically a one-liner that's fixing some of the broken docs jobs now: https://review.openstack.org/#/c/361984/ | 10:27 |
| openstackgerrit | Florian Fuchs proposed openstack/tripleo-ui: Port plan-edit file listing to use Swift https://review.openstack.org/361561 | 10:28 |
| jaosorior | shadower: merged | 10:28 |
| jaosorior | I mean, approved | 10:28 |
| jaosorior | shadower: can you check this out? https://review.openstack.org/#/c/361020/ also quite quick review | 10:29 |
| shadower | jaosorior: thanks and will do | 10:30 |
| jaosorior | shadower: do you have time to review a t-h-t patch? | 10:35 |
| *** vijaykc4 has quit IRC | 10:36 | |
| shadower | jaosorior: I do, but I'm still not that familiar with the new changes | 10:36 |
| openstackgerrit | Sagi Shnaidman proposed openstack/tripleo-quickstart: Devmode: rename delorean current so any existing repo is overwritten https://review.openstack.org/335953 | 10:36 |
| openstackgerrit | Sagi Shnaidman proposed openstack/tripleo-quickstart: POC: new devmode https://review.openstack.org/362000 | 10:36 |
| *** vijaykc4 has joined #tripleo | 10:39 | |
| *** fultonj has joined #tripleo | 10:39 | |
| jaosorior | shadower: https://review.openstack.org/#/c/359645/ | 10:40 |
| shadower | looking | 10:41 |
| jaosorior | shadower: wait up | 10:41 |
| *** akuznetsov has joined #tripleo | 10:42 | |
| shadower | k | 10:42 |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: Conditionally expose haproxy endpoints https://review.openstack.org/359645 | 10:42 |
| jaosorior | shadower: this version ^^ | 10:43 |
| shadower | all right | 10:43 |
| *** zoli|lunch is now known as zoli | 10:44 | |
| *** zoli is now known as zoliXXL | 10:44 | |
| jaosorior | shadower: those *_enabled values come from this commit https://review.openstack.org/#/c/361029/ | 10:45 |
| jaosorior | shadower: and are available in all-nodes-config | 10:45 |
| shadower | yeah, thanks | 10:45 |
| *** oshvartz has quit IRC | 10:45 | |
| openstackgerrit | Tom Barron proposed openstack/puppet-tripleo: Add manila-netapp backend to manila class https://review.openstack.org/354014 | 10:50 |
| jistr | gfidente: oooh apparently Hiera 2.0 has an alias function, which doesn't convert everything to string https://tickets.puppetlabs.com/browse/HI-183 | 10:50 |
| *** social has joined #tripleo | 10:50 | |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/tripleo-heat-templates: Populate HAProxy endpoint ports from EndpointMap https://review.openstack.org/355366 | 10:50 |
| gfidente | jistr, right I think we have 1.3 though | 10:51 |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Create composable mapping between enabled services and role ips https://review.openstack.org/348974 | 10:51 |
| gfidente | we also wanted to upgrade at some point to be able to do nested hashes lookups | 10:51 |
| gfidente | but couldn't manage to | 10:51 |
| gfidente | as existing manifests had issues | 10:51 |
| jistr | yea... seems like this would be a very helpful upgrade | 10:51 |
| jistr | gfidente: do you know if it was oonly our manifests or is this a general openstack-puppet problem? | 10:52 |
| gfidente | so I think I can take out the brackets | 10:52 |
| gfidente | and proceed with split | 10:52 |
| gfidente | it's the only way I can think of at least | 10:52 |
| jistr | gfidente: can we remove this altogether, since inerpolating an array value isn't any good https://review.openstack.org/#/c/355067/4/puppet/services/ceph-mon.yaml | 10:54 |
| jistr | gfidente: and do a ceph_mon_node_ips lookup from the .pp file itself? | 10:54 |
| jistr | gfidente: until we switch to hiera 2 | 10:55 |
| gfidente | jistr, yes for ceph | 10:55 |
| gfidente | jistr, but we'll have same problem with the other modules probably? | 10:55 |
| gfidente | but I am good doing it with ceph profile for now | 10:56 |
| gfidente | and deal with the others on a case by case | 10:57 |
| gfidente | wdyt? | 10:57 |
| jistr | gfidente: yea we might have to do the aliasing in puppet-tripleo for the time being for other modules too? Or do you see another option? E.g. i don't see even the CSV output approach solving this, because that would work only for modules which actually accept CSV and not arrays, which is a weird thing to do and AFAIK only puppet-ceph does that. | 10:57 |
| *** oshvartz has joined #tripleo | 10:57 | |
| jaosorior | gfidente: hey dude, so regarding this https://review.openstack.org/#/c/360062/ | 10:57 |
| jistr | gfidente: yea +1 for case by case. As i see it we don't have a choice. (?) | 10:58 |
| jaosorior | it depends on this patch, right? https://review.openstack.org/#/c/361362/6 | 10:58 |
| gfidente | jistr+1 :) | 10:58 |
| gfidente | jaosorior, should be the other way around | 10:58 |
| jaosorior | gfidente: you might want to rebase https://review.openstack.org/#/c/360062/14 to stop depending on that one. Since the other patch seems to have issues. | 10:58 |
| gfidente | jaosorior, ah I see what you mean yeah | 10:59 |
| gfidente | okay | 10:59 |
| gfidente | thanks | 10:59 |
| marios | Jokke_: o/ hey am getting ready to look at the manila-netap a bit, fyi cos you said you might do that too | 10:59 |
| gfidente | jistr, doing ceph_mon_ips lookup for now then | 10:59 |
| marios | Jokke_: tbarron just updated the puppet-tripleo side for the merge conflict | 10:59 |
| marios | thanks tbarron and sorry for the non-reply to email, i was away last week | 11:00 |
| tbarron | marios: thanks for all *your* help, i did hear you were out. | 11:00 |
| Jokke_ | marios, tbarron: cool, havin my lunch atm. Still trying to figure out how we want to do that or if the current approach is ok | 11:01 |
| marios | tbarron: heh np didn't really do much ... we had a chat with shardy this morning and i think he *may* be ameanable to landing with all in 'base' and revisit, but i will have a go at a split out and see if it is ok that way too, shouldn't be too much work | 11:02 |
| marios | tbarron: either way Jokke_ we need to decide on a common way forward for both https://review.openstack.org/#/c/358525/3 and https://review.openstack.org/#/c/354019/ | 11:02 |
| *** sshnaidm|afk is now known as sshnaidm | 11:03 | |
| tbarron | marios: yeah, we need a common approach | 11:03 |
| *** panda is now known as panda|lunch | 11:03 | |
| Jokke_ | marios: agreed! | 11:04 |
| tbarron | marios: Jokke_ need to work towards being able to apply multiple backends, non-exclusively, for both manila and cinder | 11:04 |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Set the default gnocchi driver to swift https://review.openstack.org/354385 | 11:04 |
| tbarron | marios: Jokke_ AND (maybe harder ??) multiple instances of the same backend, e.g. cephfsnative1 (to cluster ceph1) and cephfsnative2 (to cluster2) | 11:05 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Move snmp settings into composable services https://review.openstack.org/360062 | 11:05 |
| gfidente | jaosorior, ^^ | 11:05 |
| tbarron | marios: Jokke_ but right now minimum viable solution for N is get netapp deployed as it's the only backend that's not tech preview for Newton | 11:06 |
| jaosorior | gfidente: awesome! I'll marge it when it passes CI | 11:06 |
| tbarron | Jokke_: of courxe I want to see cephfs native land too, have been doing a lot of work on the non OOO part of that myself, etc. | 11:06 |
| Jokke_ | tbarron: cephfs native is close ... THT change left, the two others have been merged | 11:07 |
| tbarron | Jokke_: yeah, I saw. You are ahead of netapp :) | 11:08 |
| *** rodrigods has quit IRC | 11:13 | |
| *** rodrigods has joined #tripleo | 11:13 | |
| *** vijaykc4 has quit IRC | 11:13 | |
| *** vijaykc4 has joined #tripleo | 11:14 | |
| sshnaidm | cores, please take a look and merge if possible: https://review.openstack.org/#/c/355641/ https://review.openstack.org/#/c/352916/ passed ci | 11:17 |
| jaosorior | sshnaidm: done | 11:18 |
| sshnaidm | jaosorior, thanks! | 11:18 |
| openstackgerrit | Giulio Fidente proposed openstack/puppet-tripleo: Adds brackets to mon_host, removing old _v6 param https://review.openstack.org/361108 | 11:19 |
| gfidente | jistr, ^^ can you double check? | 11:19 |
| gfidente | I want to update tht part too | 11:19 |
| *** ramishra has quit IRC | 11:21 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Move ceph cluster config out of allNodesConfig https://review.openstack.org/355067 | 11:22 |
| jistr | gfidente: https://review.openstack.org/#/c/361108/3 found 2 issues otherwise lgtm | 11:22 |
| gfidente | jistr, yeah thanks | 11:22 |
| *** ramishra has joined #tripleo | 11:22 | |
| openstackgerrit | Giulio Fidente proposed openstack/puppet-tripleo: Adds brackets to mon_host, removing old _v6 param https://review.openstack.org/361108 | 11:23 |
| *** thrash|wknd is now known as thrash | 11:23 | |
| jistr | gfidente: on the other one i just wonder about the removal of CephIPv6 param -- if the puppet-tripleo place where we removed it is the only place where we used it then probably it's ok, but we need to remove it from the net iso v6 file too | 11:26 |
| gfidente | jistr, commented there we're using it in ceph-mon.yaml | 11:26 |
| jistr | ah ha! ok +2 | 11:27 |
| gfidente | though I think both changes should pass CI | 11:27 |
| *** rbowen has joined #tripleo | 11:32 | |
| EmilienM | hello | 11:35 |
| *** maeca1 has joined #tripleo | 11:37 | |
| openstackgerrit | Merged openstack/tripleo-validations: Updated from global requirements https://review.openstack.org/361877 | 11:39 |
| *** pradk has joined #tripleo | 11:42 | |
| hewbrocca | EmilienM: good morning! | 11:42 |
| openstackgerrit | Brad P. Crochet proposed openstack/tripleo-heat-templates: Composable Zaqar services https://review.openstack.org/331682 | 11:42 |
| *** panda|lunch is now known as panda | 11:43 | |
| *** Guest85979 is now known as beagles | 11:44 | |
| openstackgerrit | Merged openstack/tripleo-common: Don't expect input to be in every workflow in doc generation https://review.openstack.org/361984 | 11:44 |
| *** asbishop has joined #tripleo | 11:44 | |
| *** pradk has quit IRC | 11:46 | |
| karthiks | EmilienM, can you please have a look at https://review.openstack.org/#/c/361350/ | 11:47 |
| EmilienM | karthiks: I will | 11:47 |
| karthiks | Thanks EmilienM . | 11:48 |
| openstackgerrit | Brad P. Crochet proposed openstack/tripleo-heat-templates: Composable Mistral services https://review.openstack.org/323436 | 11:48 |
| d0ugal | shadower, mandre: Can you rebase 353898? The docs change has landed. | 11:50 |
| *** shardy has joined #tripleo | 11:50 | |
| d0ugal | docs fix* | 11:50 |
| *** jeckersb_gone is now known as jeckersb | 11:50 | |
| mandre | d0ugal: on it | 11:50 |
| *** asbishop has quit IRC | 11:50 | |
| openstackgerrit | Jiri Stransky proposed openstack/puppet-tripleo: Hanlde galera_node_names being an array https://review.openstack.org/362046 | 11:51 |
| jistr | gfidente: ^^ here's for galera node names | 11:51 |
| *** shardy_ has joined #tripleo | 11:52 | |
| *** shardy_ has quit IRC | 11:52 | |
| gfidente | jistr, nice | 11:53 |
| openstackgerrit | Jiri Stransky proposed openstack/puppet-tripleo: Handle galera_node_names being an array https://review.openstack.org/362046 | 11:54 |
| *** jayg|g0n3 is now known as jayg | 11:54 | |
| gfidente | cause it's safe and will enable https://review.openstack.org/#/c/355068/ | 11:54 |
| jistr | gfidente: yea at least for the galera_node_names part | 11:55 |
| gfidente | so for controller_node_names we should actually add something in ServiceNetMap | 11:55 |
| *** lucasagomes is now known as lucas-hungry | 11:55 | |
| EmilienM | jaosorior: can you look https://review.openstack.org/#/c/361164/ please? | 11:56 |
| jistr | gfidente: wondering what do we need controller_node_names for. Should we rather use service specific naming like pacemaker_node_names and haproxy_node_names etc? | 11:56 |
| gfidente | yeah I see those two and midonet plugin using it | 11:56 |
| gfidente | though list of controller names seems fair to have to me | 11:57 |
| openstackgerrit | Martin André proposed openstack/tripleo-common: Allow running validation against different plans https://review.openstack.org/318194 | 11:57 |
| openstackgerrit | Martin André proposed openstack/tripleo-common: Workflows to run validations https://review.openstack.org/353899 | 11:57 |
| openstackgerrit | Martin André proposed openstack/tripleo-common: Workflows to load validations https://review.openstack.org/353898 | 11:57 |
| openstackgerrit | Martin André proposed openstack/tripleo-common: Disable retry files for ansible validations https://review.openstack.org/329039 | 11:57 |
| *** limao has joined #tripleo | 11:58 | |
| *** jpena is now known as jpena|lunch | 11:58 | |
| gfidente | jistr, we can probably do that per-role | 11:58 |
| jistr | gfidente: maybe yea, but probably not for the use cases we have right now? e.g. the midonet plugin should be using something neutron-specific? | 11:58 |
| gfidente | neutron_node_names | 11:59 |
| gfidente | right | 11:59 |
| openstackgerrit | Jiri Stransky proposed openstack/tripleo-heat-templates: Generate composable service node_names lists https://review.openstack.org/355068 | 12:00 |
| *** dprince has joined #tripleo | 12:00 | |
| gfidente | haproxy is the one consuming controller_node_ips too | 12:00 |
| *** jaosorior has quit IRC | 12:02 | |
| *** jaosorior has joined #tripleo | 12:03 | |
| *** limao_ has joined #tripleo | 12:03 | |
| mandre | d0ugal: quick question... | 12:03 |
| mandre | d0ugal: there is a bit of manual setup required after deploying the undercloud to boostrap the ansible validations framework (namely copy the SSH key to the overcloud nodes) | 12:04 |
| mandre | it's all bundled in a mistral workflow | 12:04 |
| mandre | do we want to make it a tripleoclient command? | 12:04 |
| d0ugal | mandre: I guess we might want to run that workflow at the end of the undercloud install? | 12:05 |
| d0ugal | mandre: at the same time we make the default plan | 12:05 |
| shadower | I like that | 12:05 |
| EmilienM | gfidente, jaosorior: I think I found an issue in CI, it seems like puppet modules are not deployed from source but from delorean | 12:05 |
| EmilienM | and since we haven't had promotion for 5 days, our CI is not deploying latest bits | 12:06 |
| *** limao has quit IRC | 12:06 | |
| EmilienM | sshnaidm: ^ | 12:06 |
| d0ugal | mandre, shadower - akrivoka recently done this for the default plan creation, so she probably knows exactly where to do it :) | 12:06 |
| hewbrocca | EmilienM: blearrgh | 12:07 |
| mandre | d0ugal: can we do it conditionally? i.e. when we enable validations | 12:07 |
| sshnaidm | EmilienM, afaik puppet modules should be installed from packages, not from sources now | 12:07 |
| d0ugal | mandre: Good question, I assume so but I've not looked into the process enough | 12:07 |
| *** rlandy has joined #tripleo | 12:07 | |
| mandre | d0ugal: ok. I'll dig around | 12:08 |
| sshnaidm | EmilienM, but puppet modules are installed from current rdo trunk, not from promoted current-tripleo | 12:08 |
| EmilienM | sshnaidm: that's a problem if our promotion system doesn't happen | 12:08 |
| EmilienM | sshnaidm: are you sure? | 12:08 |
| sshnaidm | EmilienM, yep: https://github.com/openstack-infra/tripleo-ci/blob/master/scripts/tripleo.sh#L263 | 12:08 |
| sshnaidm | EmilienM, puppet-* is installed from http://trunk.rdoproject.org/centos7/current/delorean.repo | 12:09 |
| EmilienM | ok, let me check why we have old packages in our CI | 12:09 |
| *** pradk has joined #tripleo | 12:09 | |
| EmilienM | sshnaidm: probably a false alarm :) | 12:10 |
| sshnaidm | EmilienM, should be installed those: https://trunk.rdoproject.org/centos7/current/ | 12:11 |
| sshnaidm | EmilienM, do you have logs? | 12:11 |
| akrivoka | mandre: https://github.com/openstack/instack-undercloud/blob/master/instack_undercloud/undercloud.py#L1013 | 12:12 |
| *** mburned_out is now known as mburned | 12:12 | |
| EmilienM | sshnaidm: it was a false alarm, nevermind. | 12:12 |
| openstackgerrit | Emilien Macchi proposed openstack/instack-undercloud: configure heat with new authtoken parameters https://review.openstack.org/361409 | 12:13 |
| sshnaidm | EmilienM, ok | 12:13 |
| openstackgerrit | Marios Andreou proposed openstack/puppet-tripleo: Add manila-netapp backend to manila class and tidy up generic https://review.openstack.org/354014 | 12:14 |
| openstackgerrit | Marios Andreou proposed openstack/puppet-tripleo: Add manila-netapp backend to manila class and tidy up generic https://review.openstack.org/354014 | 12:15 |
| *** rhallisey has joined #tripleo | 12:19 | |
| jaosorior | EmilienM, gfidente, marios: keystone endpoints via puppet is green :D https://review.openstack.org/#/c/359680/ and so is the commit that disables post-config from tripleoclient https://review.openstack.org/#/c/353307/ | 12:23 |
| gfidente | jaosorior, great | 12:23 |
| hewbrocca | jaosorior: woooooooooooooo | 12:23 |
| hewbrocca | os-cloud-config.die.die.die! | 12:24 |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Setup CORS settings for tripleo-ui https://review.openstack.org/360593 | 12:24 |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Introduce 'enable_ui' option https://review.openstack.org/344140 | 12:24 |
| shardy | gfidente: thanks for updating https://review.openstack.org/#/c/355067/ | 12:24 |
| shardy | +2 on moving *node_ips lookup into puppet-tripleo | 12:25 |
| shardy | I did the same for rabbitmq in https://review.openstack.org/#/c/361730/ | 12:25 |
| shardy | so we can remove the yaml aliases in allNodesConfig | 12:25 |
| gfidente | shardy, we worked on that with jistr and were looking at the two submissions above | 12:26 |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Setup CORS settings for tripleo-ui https://review.openstack.org/360593 | 12:26 |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Introduce 'enable_ui' option https://review.openstack.org/344140 | 12:26 |
| gfidente | shardy, looks like either we add pacemaker and haproxy into servicenetmap or keep _controller_names,_ips in there for now | 12:27 |
| shardy | gfidente: I'd say we should put the services into ServiceNetMap | 12:27 |
| gfidente | shardy, ok | 12:27 |
| *** chem has joined #tripleo | 12:27 | |
| jaosorior | shardy: is there a way you can think of of fetching on which network a service is listening on? (on puppet) | 12:27 |
| shardy | jaosorior: Yes, we set the bind_ip by doing a substitution of the IP for the network assigned to the service in ServiceNetMap | 12:28 |
| * shardy finds the patch that does it | 12:28 | |
| EmilienM | jaosorior: reviewing it too today | 12:29 |
| shardy | https://github.com/openstack/tripleo-heat-templates/commit/b31d80f74c0b34ddfe5365900738819ba94305ee#diff-8c0eec772f51e7572337db4fc7b3c286R92 | 12:29 |
| shardy | jaosorior: ^^ like that | 12:29 |
| jaosorior | shardy: right; in the internal TLS case, the service's internal endpoint will point in most of the cases to the internal API VIP, but in other cases it might point to storage and such. | 12:30 |
| mandre | thanks akrivoka, very helpful | 12:30 |
| jaosorior | shardy: so, we need a certificate per VIP endpoint (or at least the used ones) | 12:30 |
| jaosorior | shardy: so I need a way to tell haproxy which certificate to use in the internal endpoint depending on the service | 12:31 |
| EmilienM | jaosorior: +A, excellent work here | 12:31 |
| jaosorior | EmilienM: thanks :D | 12:31 |
| * hewbrocca does happy dance | 12:32 | |
| * jaosorior puts celebration beer on the fridge | 12:32 | |
| * EmilienM still at breakfast | 12:32 | |
| jaosorior | EmilienM: breakfast beer? | 12:33 |
| jaosorior | :P | 12:33 |
| EmilienM | you're asking a french guy if he takes beer at breakfast. Sigh. | 12:33 |
| hewbrocca | Breakfast wine, whole other story | 12:34 |
| jaosorior | s/beer/wine | 12:34 |
| EmilienM | hewbrocca knows french better | 12:34 |
| *** pkovar has joined #tripleo | 12:36 | |
| sshnaidm | jaosorior, should it be so then? "ironic::drivers::agent:post_deploy_get_power_state_retries: 15" | 12:36 |
| sshnaidm | jaosorior, all (double)colons are in place I hope | 12:36 |
| jaosorior | shardy: sounds about right | 12:36 |
| jaosorior | sshnaidm: | 12:36 |
| sshnaidm | jaosorior, great, then will update the patch | 12:37 |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/tripleo-ci: Increase timeout for ironic power status report https://review.openstack.org/361915 | 12:38 |
| numans | EmilienM, gfidente can you please add this into your review queue - https://review.openstack.org/#/c/314875/, Babu has submitted another patchset addressing gfidente's comments and resolving merge conflict. Thanks | 12:38 |
| jaosorior | sshnaidm: it looks about right, but you forgot the depends-on | 12:39 |
| sshnaidm | jaosorior, yeah, I'm looking how to do it | 12:39 |
| *** milan has joined #tripleo | 12:40 | |
| jaosorior | sshnaidm: Add this to the commit messsage -> Depends-On: If2c6987d0f38aecabf41c09505c4f0dca1116034 | 12:40 |
| sshnaidm | jaosorior, thanks | 12:40 |
| openstackgerrit | Merged openstack/tripleo-common: Exclude more unneeded files from default plan https://review.openstack.org/355472 | 12:40 |
| openstackgerrit | Sagi Shnaidman proposed openstack-infra/tripleo-ci: Increase timeout for ironic power status report https://review.openstack.org/361915 | 12:41 |
| gfidente | numans, I see it defaults to eth0 but if the encap ip should be from one of the provisioned networks | 12:42 |
| gfidente | you can collect that from heat | 12:42 |
| gfidente | numans, should it be the network where tenants networks go? | 12:43 |
| gfidente | numans, or rather the network where the neutron api is? | 12:43 |
| shadower | mandre: question about this validation: https://github.com/rthallisey/clapper/blob/ansible1.9/ansible-tests/validations/compute_node_connectivity.yaml | 12:43 |
| gfidente | or none of the two? | 12:43 |
| *** vijaykc4 has quit IRC | 12:43 | |
| shadower | mandre: are the values (e.g. controller_ctlplane_ip_address) coming from anywhere? I don't see how they could | 12:43 |
| mandre | shadower: you have to pass the values as ansible environment variables, which we don't do | 12:45 |
| gfidente | shardy, so shall I add haproxy/pacemaker in servicenetmap in https://review.openstack.org/#/c/355068 ? | 12:45 |
| shadower | mandre: ah, right. Might be something we could figure out from Heat or Mistral later. So by default this just runs the gateway ping, correct? | 12:45 |
| mandre | shadower: I think it was our first ansible validation, so it was a bit experimental | 12:45 |
| shadower | yeah | 12:45 |
| *** egafford has quit IRC | 12:46 | |
| mandre | shadower: yup, we should be able to get what we need from the environment | 12:46 |
| jistr | gfidente: just replied there | 12:46 |
| *** pradk has quit IRC | 12:46 | |
| karthiks | gfidente, jaosorior shardy for review https://review.openstack.org/#/c/329415/ has two +2s and now CI is all green now. PTAL | 12:46 |
| jistr | gfidente: i don't think we need them in servicenetmap essentially? if we are only concerned about the hostnames | 12:47 |
| shardy | gfidente: Yes, please do, thanks! | 12:47 |
| jaosorior | karthiks: done | 12:47 |
| shardy | then we can adjust the profiles to match | 12:47 |
| mandre | shadower: also, do note there is currently no way to pass ansible variables via our mistral workflow | 12:47 |
| karthiks | Thanks jaosorior | 12:47 |
| mandre | shardy: it shouldn't be too difficult to add, though | 12:48 |
| jaosorior | karthiks: what does PTAL stand for? | 12:48 |
| mandre | shadower: ^^ | 12:48 |
| sshnaidm | derekh, hi | 12:48 |
| shadower | mandre: yeah fair enough. I think that's fine atm | 12:48 |
| shardy | gfidente: if it's easier, we can just not remove controller_node_names | 12:48 |
| gfidente | shardy, jistr we need to find agreement :) | 12:48 |
| mandre | sorry shardy, please ignore | 12:48 |
| *** jdob has joined #tripleo | 12:48 | |
| hewbrocca | Please Take A Look! | 12:48 |
| shardy | I think I actually removed that by mistake :( | 12:48 |
| hewbrocca | jaosorior: ^^^ :D | 12:48 |
| karthiks | jaosorior, yes hewbrocca | 12:48 |
| jaosorior | hewbrocca: I see. Thanks | 12:48 |
| gfidente | shardy, jistr ok then maybe let's do it in follow up | 12:48 |
| gfidente | jistr, will you update it to keep _names? | 12:49 |
| shardy | gfidente: Yeah, let's add controller_node_ips back for now, and potentially add haproxy/pacemaker to ServiceNetMap later | 12:49 |
| jistr | gfidente: sorry maybe i just don't understand all the context... https://review.openstack.org/#/c/355068 is about hostnames only, no? ServiceNetMap doesn't play any role in that does it? | 12:49 |
| shardy | I only meant to remove the *service* names, not the role names | 12:49 |
| *** Goneri has joined #tripleo | 12:50 | |
| jistr | but yea in general +1 for keeping controller_names | 12:50 |
| jistr | gfidente: yea i'll update it | 12:50 |
| shardy | jistr: you're right, we only filter *node_ips based on ServiceNetMap | 12:50 |
| openstackgerrit | Florian Fuchs proposed openstack/tripleo-ui: Port plan-edit file listing to use Swift https://review.openstack.org/361561 | 12:51 |
| *** vijaykc4 has joined #tripleo | 12:51 | |
| gfidente | yeah I raised that cause haproxy was using node_ips too | 12:51 |
| jaosorior | and what was the issue on adding haproxy and pacemaker to the servicenetmap? | 12:51 |
| gfidente | guys I'm going for lunch, back soon | 12:51 |
| jistr | i just noticed i'll need to update the puppet-tripleo galera_node_names patch too as it now will be mysql_node_names | 12:51 |
| *** vijaykc4 has quit IRC | 12:51 | |
| shardy | jaosorior: just that we probably don't need to do it right now if we fix my patch | 12:52 |
| shardy | jistr, gfidente: FYI I had planned to put per-role *_names in https://review.openstack.org/#/c/348973 for all roles | 12:52 |
| shardy | but I've since started to prefer using per-service data wherever possible, as it'll work much better when folks come to move services between roles | 12:53 |
| shardy | jistr: are you OK to update https://review.openstack.org/#/c/355068 or shall I do it? | 12:55 |
| *** lucas-hungry is now known as lucasagomes | 12:56 | |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Setup CORS settings for tripleo-ui https://review.openstack.org/360593 | 12:57 |
| jistr | shardy: i'll update, just wanted to do if after i update the galera part too | 12:58 |
| jistr | and i wanted to test that... | 12:58 |
| jistr | but given the time constraints, i may just update them both right now | 12:58 |
| shardy | jistr: ack, sounds good, thanks! | 12:58 |
| jistr | and test them subsequently | 12:58 |
| *** tzumainn has joined #tripleo | 12:58 | |
| openstackgerrit | Jiri Stransky proposed openstack/puppet-tripleo: Handle galera_node_names being an array https://review.openstack.org/362046 | 12:59 |
| *** anshul has quit IRC | 12:59 | |
| shardy | jistr: Ok, sounds good | 13:00 |
| shardy | I just noticed another issue with https://review.openstack.org/#/c/355068 so I may have to push another revision too | 13:00 |
| *** trozet has joined #tripleo | 13:02 | |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Setup CORS settings for tripleo-ui https://review.openstack.org/360593 | 13:02 |
| numans | gfidente, the encap ip should be the host ip. | 13:02 |
| *** pcaruana has quit IRC | 13:02 | |
| jistr | shardy, gfidente: i'll also remove this for the same reason as with the IPs, wdyt? https://review.openstack.org/#/c/355068/2/puppet/services/ceph-mon.yaml | 13:03 |
| jistr | it gave me a rebase conflict so that's why i noticed it | 13:03 |
| *** social has quit IRC | 13:03 | |
| shardy | jistr: +1 | 13:03 |
| *** jpena|lunch is now known as jpena | 13:04 | |
| *** social has joined #tripleo | 13:04 | |
| openstackgerrit | Martin Mágr proposed openstack/tripleo-heat-templates: Availability monitoring agents support https://review.openstack.org/254788 | 13:05 |
| numans | gfidente, this ip is internal ip of the host - https://github.com/openstack/networking-ovn/blob/master/devstack/lib/networking-ovn#L352 | 13:05 |
| *** Goneri has quit IRC | 13:06 | |
| EmilienM | shardy: taking over 361730 to fix lint | 13:06 |
| shardy | EmilienM: ack, thanks! | 13:06 |
| openstackgerrit | Emilien Macchi proposed openstack/puppet-tripleo: Make service profiles default to rabbitmq_node_ips https://review.openstack.org/361730 | 13:08 |
| openstackgerrit | Jiri Stransky proposed openstack/tripleo-heat-templates: Generate composable service node_names lists https://review.openstack.org/355068 | 13:10 |
| *** vijaykc4 has joined #tripleo | 13:10 | |
| jistr | shardy, gfidente: ^ updated & rebased on top of the latest patchset of the patch underneath | 13:11 |
| shardy | jistr: thanks! | 13:11 |
| *** anbu has joined #tripleo | 13:11 | |
| anbu | gfidente, Hi Giulio. | 13:12 |
| anbu | Thank you for the review. | 13:12 |
| shadower | mandre: the controller/gateway ping test is in the tripleo-heat-templates anyway so I think we can just drop it | 13:13 |
| mandre | shadower: right, it's not super useful right now | 13:14 |
| *** rbrady has joined #tripleo | 13:14 | |
| *** bkopilov has quit IRC | 13:14 | |
| *** shardy has quit IRC | 13:17 | |
| *** pcaruana has joined #tripleo | 13:17 | |
| openstackgerrit | Babu Shanmugam proposed openstack/puppet-tripleo: Add ML2/OVN support for neutron profile https://review.openstack.org/314875 | 13:18 |
| *** pgadiya has quit IRC | 13:19 | |
| *** vijaykc4 has quit IRC | 13:22 | |
| *** pkovar has quit IRC | 13:22 | |
| *** abregman_ has joined #tripleo | 13:27 | |
| *** abregman has quit IRC | 13:30 | |
| *** myoung|afk is now known as myoung | 13:32 | |
| *** nyechiel has quit IRC | 13:34 | |
| *** dmacpher has joined #tripleo | 13:38 | |
| *** chem has quit IRC | 13:40 | |
| *** chem has joined #tripleo | 13:40 | |
| jaosorior | jistr: ping | 13:40 |
| jistr | jaosorior: pong, heya | 13:40 |
| jaosorior | jistr: hey dude, any idea why would the HA job fail with "pcs command not found" here? https://review.openstack.org/#/c/359038/4/manifests/profile/base/haproxy.pp | 13:41 |
| jistr | jaosorior: ah yea the exec should pass `path =>` | 13:41 |
| jistr | jaosorior: actually it does but i think pcs is in sbin maybe? | 13:42 |
| jistr | jaosorior: you might try with: path => ['/usr/bin', '/usr/sbin'], | 13:43 |
| EmilienM | or give the path to pcs | 13:43 |
| jistr | yea | 13:43 |
| EmilienM | like /usr/sbin/pcs | 13:43 |
| *** dtrainor has quit IRC | 13:43 | |
| *** dtrainor has joined #tripleo | 13:43 | |
| jaosorior | EmilienM, jistr thanks. Will do | 13:43 |
| openstackgerrit | Marios Andreou proposed openstack/tripleo-heat-templates: Add NetApp Manila driver integration and tidy up generic https://review.openstack.org/354019 | 13:46 |
| *** skramaja_ has joined #tripleo | 13:46 | |
| *** skramaja has quit IRC | 13:46 | |
| *** skramaja_ is now known as skramaja | 13:46 | |
| *** ayoung has joined #tripleo | 13:47 | |
| *** jubapa has quit IRC | 13:48 | |
| dtantsur | dprince, morning! re ironic patches, mind merging both into one final for me to test locally? please see a few comments there | 13:51 |
| *** pkovar has joined #tripleo | 13:51 | |
| openstackgerrit | Karthik S proposed openstack/puppet-tripleo: Configure the numvfs for SRIOV interfaces https://review.openstack.org/361350 | 13:52 |
| openstackgerrit | Merged openstack/tripleo-quickstart: Add ci-script to upload images in OVB https://review.openstack.org/355476 | 13:52 |
| openstackgerrit | Emilien Macchi proposed openstack-infra/tripleo-ci: Implement scenario001 CI job https://review.openstack.org/360039 | 13:52 |
| EmilienM | bnemec: I addressed your comments on https://review.openstack.org/#/c/360039/ | 13:53 |
| dprince | dtantsur: ack. They are slightly different but I don't mind merging them I guess | 13:54 |
| gfidente | jistr, so to make https://review.openstack.org/#/c/355068/6/puppet/all-nodes-config.yaml pass we need to patch p-t to set mon names from array as well | 13:54 |
| *** xuao has quit IRC | 13:54 | |
| gfidente | jistr, do you think I can make it in https://review.openstack.org/#/c/361108/ so we have less changes around? | 13:54 |
| jaosorior | EmilienM: do the puppet execs get ran with the root user? or which? | 13:55 |
| *** pcaruana has quit IRC | 13:55 | |
| jistr | gfidente: yea +1 on that | 13:56 |
| gfidente | ok | 13:56 |
| EmilienM | jaosorior: sudo afik | 13:56 |
| *** zoliXXL is now known as zoli|brb | 13:56 | |
| *** egafford has joined #tripleo | 13:57 | |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/puppet-tripleo: Reload HAProxy on refresh https://review.openstack.org/359038 | 13:59 |
| jistr | jaosorior: +2 ^^ | 14:00 |
| jaosorior | jistr: thanks | 14:00 |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Fix Ironic service bind IPs https://review.openstack.org/361459 | 14:00 |
| dprince | dtantsur: ^^ | 14:00 |
| dtantsur | thanks! | 14:01 |
| *** pradk has joined #tripleo | 14:02 | |
| jaosorior | gfidente, dprince: where does the replacement take place again? I remember it happens but not how exactly | 14:02 |
| jaosorior | the IP replacement | 14:02 |
| dprince | dtantsur: NP, with regards to the my_ip change note that I for me (and likely other network isolation users) the MySql traffic runs on the internal_api network which is a vlan (tagged traffic). This wouldn't lend itself well to PXE booting other instances so I think we want a separate setting that can default to 'ctlplane' instead here | 14:03 |
| gfidente | jaosorior https://github.com/openstack/tripleo-heat-templates/blob/master/puppet/controller.yaml#L470 | 14:03 |
| dtantsur | dprince, yeah, this is totally fair | 14:04 |
| dtantsur | I still have to learn a lot about our networking :( | 14:04 |
| d0ugal | thrash: ping | 14:04 |
| *** marcusvrn_ has quit IRC | 14:04 | |
| thrash | d0ugal: pong | 14:05 |
| dprince | dtantsur: it would be cool to have another isolated network on different NICS for overcloud tenant baremetal provisioning. We would actually support that with os-net-config but it would require custom network templates | 14:05 |
| *** mhenkel has joined #tripleo | 14:05 | |
| jaosorior | gfidente: I see. So with that, if I try to pass what network is a service on, it will get replaced, right? | 14:05 |
| dtantsur | this is what I would recommend people to do honestly | 14:05 |
| dprince | dtantsur: but as an out of the box I think re-using the ctlplane for this would make testing, etc. easiest so long as the overcloud Ironic doesn't conflict with the undercloud Ironic. | 14:05 |
| *** jprovazn has quit IRC | 14:06 | |
| gfidente | jaosorior, if a key value is a network name, it gets replaced with the local ip on that network | 14:06 |
| dtantsur | dprince, the "conflict" bit is worrying. it has high chances of clashing with ironic-inspector. but for now, yes, we can delay it | 14:06 |
| jaosorior | gfidente: well, crap | 14:06 |
| dtantsur | dprince, while you're on it, is it possible you skim through https://review.openstack.org/354016 in case I write something stupid there from network pov? | 14:06 |
| jaosorior | gfidente: I guess I'll have to figure out something else then. | 14:07 |
| dprince | dtantsur: ack, anything running a wildcard DNS would be concerning on a shared network | 14:07 |
| *** lmiccini has quit IRC | 14:07 | |
| d0ugal | thrash: I am adding --plan back! https://review.openstack.org/#/c/360305/17/tripleoclient/v1/overcloud_deploy.py | 14:07 |
| dprince | dtantsur: DHCP rather | 14:07 |
| *** akuznetsov has quit IRC | 14:07 | |
| thrash | d0ugal: wow | 14:07 |
| thrash | lol | 14:07 |
| gfidente | jaosorior, sorry I am not sure what you're dealing with but purpose there was to set the local ip address of any given network | 14:07 |
| d0ugal | thrash: I want that to never be used with another arg, but I don't want to put everything in a mutually exclusive group. Any clever tricks? | 14:07 |
| jaosorior | gfidente: it makes sense. | 14:07 |
| thrash | d0ugal: I've struggled with that one for other things. And no, I haven't come up with anything yet. | 14:08 |
| *** adarazs is now known as adarazs_afk | 14:08 | |
| thrash | d0ugal: other than *maybe* using a subparser. | 14:08 |
| thrash | d0ugal: but that gets tricky too... | 14:08 |
| *** pcaruana has joined #tripleo | 14:09 | |
| jaosorior | gfidente: the thing is that, for TLS, we will need a certificate per-VIP. And, some services don't listen on the internalapi network. So for these cases I need to figure out which network these services are listening on on HAProxy and use an appropriate certificate | 14:09 |
| d0ugal | thrash: damn. I did think about checking sys.argv... but that might be a bit gross. | 14:09 |
| jaosorior | gfidente: for that I was thinking of using the network name | 14:09 |
| thrash | d0ugal: let me think about that one. It's a bit different than what I was trying to accomplish. | 14:09 |
| d0ugal | thrash: k, thanks! | 14:09 |
| jaosorior | gfidente: so I wanted to pass the servicenetmap via hieradata | 14:09 |
| thrash | d0ugal: and you probably don't want to use return from take_action. | 14:10 |
| thrash | d0ugal: that will always return 0, which is not ideal for scripting | 14:10 |
| thrash | d0ugal: unless _start_mistral_deploy throws exceptions | 14:10 |
| *** abregman_ is now known as abregman_|mtg | 14:11 | |
| d0ugal | thrash: oh, damn. How does it know I am returning? (given no return, returns None anyway?) | 14:11 |
| openstackgerrit | Karthik S proposed openstack/puppet-tripleo: Configure the numvfs for SRIOV interfaces https://review.openstack.org/361350 | 14:12 |
| d0ugal | thrash: I am tempted to just go with a new command to make things easier. `openstack overcloud plan deploy` | 14:12 |
| thrash | d0ugal: if _start_mistral_deploy throws an exception, that should be sufficient. I'm not saying the return is bad, just don't have that as your *only* way of returning | 14:12 |
| *** iogg is now known as ioggstream | 14:12 | |
| d0ugal | thrash: gotcha, that makes sense. | 14:12 |
| thrash | d0ugal: I would not be *completely* opposed to a new command. It's the same problem I have with tripleo-build-images vs. openstack overcloud image build | 14:13 |
| *** Goneri has joined #tripleo | 14:14 | |
| *** lmiccini has joined #tripleo | 14:14 | |
| d0ugal | thrash: we already have commands to create/delete/list plans. So it actually might even make more sense now that I think about it. | 14:15 |
| *** skramaja has quit IRC | 14:15 | |
| *** skramaja has joined #tripleo | 14:15 | |
| *** zoli|brb is now known as zoli | 14:16 | |
| *** zoli is now known as zoliXXL | 14:16 | |
| openstackgerrit | Giulio Fidente proposed openstack/puppet-tripleo: Defaults to ceph_mon_node_ips and ceph_mon_node_names https://review.openstack.org/361108 | 14:17 |
| gfidente | jaosorior, and it gets replaced | 14:17 |
| jaosorior | gfidente: yeah, that's the issue I have | 14:17 |
| sshnaidm | dprince, hi, could we reconsider caching images again? | 14:18 |
| openstackgerrit | Dmitry Tantsur proposed openstack/tripleo-heat-templates: Ironic: enabled iPXE by default and make it configurable https://review.openstack.org/362148 | 14:18 |
| dprince | sshnaidm: um, are we still having job timeouts? which jobs? | 14:19 |
| jaosorior | gfidente: Do you know if map_replace does a deep search of the values? | 14:19 |
| openstackgerrit | Merged openstack/puppet-tripleo: Configure keystone endpoints in service profile https://review.openstack.org/359680 | 14:20 |
| openstackgerrit | Merged openstack/tripleo-common: Add Mistral action for boot configuration https://review.openstack.org/354174 | 14:20 |
| openstackgerrit | Merged openstack/tripleo-common: Add Mistral action & workflow for root device configuration https://review.openstack.org/355947 | 14:20 |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Configure the pci_passthrough_whitelist via THT https://review.openstack.org/329415 | 14:20 |
| *** zoliXXL is now known as zoli|dentist | 14:21 | |
| sshnaidm | dprince, it's much much better, but still timeouts both for ha and nonha, you can see here: http://status-tripleoci.rhcloud.com | 14:21 |
| openstackgerrit | Dmitry Tantsur proposed openstack/tripleo-heat-templates: Ironic: enabled iPXE by default and make it configurable https://review.openstack.org/362148 | 14:21 |
| gfidente | jaosorior, it should be happening here https://github.com/openstack/heat/blob/master/heat/engine/hot/functions.py#L568 | 14:22 |
| gfidente | and I doesn't look like it does deep search | 14:22 |
| sshnaidm | dprince, 3 for nonha and 7-8 for ha from Fri, not sure if it's critical though | 14:22 |
| *** dsariel has quit IRC | 14:23 | |
| *** oshvartz has quit IRC | 14:23 | |
| social | dtantsur: https://review.openstack.org/#/c/345980/ so it'll work fine without ipxe element, I can add the include but I prefer to do so after the other review is in :) | 14:24 |
| lucasagomes | derekh, in the tripleo-ci we don't have any function like iniset from devstack to edit .conf files do we ? (/me was looking but couldn't find it) | 14:24 |
| dprince | sshnaidm: is that mostly the periodic jobs? if so would it be safer to just disable the extra stack delete the periodic jobs perform? | 14:24 |
| sshnaidm | dprince, no, it's gate jobs | 14:25 |
| *** jaosorior is now known as jaosorior_away | 14:25 | |
| dtantsur | social, sorry, I don't quite get it.. how does it work without including the class? | 14:25 |
| sshnaidm | dprince, periodical must build images | 14:25 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Add hyperconverged-ceph environment to include CephOSD on computes https://review.openstack.org/338113 | 14:25 |
| social | dtantsur: the class is in ironic::inspector because of backwards compatibility | 14:25 |
| social | dtantsur: the warning with note that after newton cycle it'll be removed + releasenotes ,) | 14:25 |
| dtantsur | social, oh true | 14:26 |
| openstackgerrit | Karthik S proposed openstack/tripleo-heat-templates: Customize the reservation of memory for host processes https://review.openstack.org/327705 | 14:26 |
| sshnaidm | dprince, you are welcome to comment here about periodic: https://review.openstack.org/#/c/361429/ | 14:26 |
| gfidente | rhallisey, shall we pull in openstack-selinux explicitly? | 14:27 |
| gfidente | or is it installed as dep? | 14:27 |
| dprince | sshnaidm: could we enable caching next Friday evening? Pending no other outages? | 14:28 |
| sshnaidm | dprince, sure | 14:28 |
| derekh | lucasagomes: no don't think so | 14:28 |
| *** saneax is now known as saneax-_-|AFK | 14:29 | |
| dprince | sshnaidm: so essentially we are scheduling the upgrade to occur then (if others agree) so long as no other outages block us | 14:29 |
| lucasagomes | derekh, ack, I'll add one to fix that MTU thingy with neutron and ironic | 14:29 |
| *** jtomasek|afk has quit IRC | 14:30 | |
| sshnaidm | dprince, although I don't understand the value of building images in each gate job, we lived with cached images pretty well before. | 14:30 |
| derekh | lucasagomes: ok, | 14:30 |
| shadower | jaosorior_away: what does "recheck timed out" do? | 14:30 |
| dprince | sshnaidm: I like image caching very much | 14:30 |
| derekh | sshnaidm: just saw your mail setting new keys now | 14:31 |
| sshnaidm | derekh, yeah.. :( | 14:31 |
| dprince | sshnaidm: just want to flip the switch to test it at a time that doesn't impact the work week, landing code, etc. | 14:31 |
| sshnaidm | dprince, sure | 14:32 |
| sshnaidm | dprince, I can keep eye on it even in Sun | 14:32 |
| openstackgerrit | Saravanan KR proposed openstack/tripleo-heat-templates: Added DPDK environment file and nic configs samples https://review.openstack.org/361238 | 14:33 |
| dprince | jaosorior_away: good catch on https://review.openstack.org/#/c/361362/6/puppet/services/horizon.yaml | 14:34 |
| dprince | jaosorior_away: fixing... | 14:34 |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Upgrade ceph-mon https://review.openstack.org/357130 | 14:34 |
| *** rhallisey has quit IRC | 14:35 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Upgrade ceph-osd https://review.openstack.org/357312 | 14:35 |
| thrash | d0ugal: I'd go with that. | 14:36 |
| thrash | d0ugal: perhaps add --plan to the existing command, and deprecate. | 14:37 |
| hewbrocca | dprince: I think scheduling it for the weekend is sensible | 14:38 |
| hewbrocca | dprince: pabelanger tells me they're going to restart nodepool.o.o later this week, with some shade changes that will hopefully fix our FIP problem | 14:39 |
| hewbrocca | but we'll see | 14:39 |
| dprince | hewbrocca: ideally we'd queue up this so we can verify them one at a time | 14:40 |
| hewbrocca | Yes | 14:40 |
| *** rhallisey has joined #tripleo | 14:41 | |
| openstackgerrit | Lars Kellogg-Stedman proposed openstack/tripleo-heat-templates: Add fluentd client service https://review.openstack.org/353506 | 14:42 |
| openstackgerrit | Lucas Alvares Gomes proposed openstack-infra/tripleo-ci: Do not reuse dnsmasq-ironic.conf for forcing MTU 1400 https://review.openstack.org/362162 | 14:43 |
| lucasagomes | derekh, ^ not tested, waiting on jenkins | 14:43 |
| gfidente | rhallisey, I saw those selinux rules in the .spec as well | 14:46 |
| gfidente | rhallisey, I don't see it installed so I am going to add it to tpe | 14:47 |
| rhallisey | ok | 14:47 |
| rhallisey | it may add ~1 min or two to deployment, but if you're running in enforcing you'll need it | 14:48 |
| gfidente | I think we can use a depends-on and see if we pass as well | 14:49 |
| *** rbowen has quit IRC | 14:49 | |
| paramite | dprince, EmilienM, larsks: Hi guys, can we join a BJ and talk about https://review.openstack.org/#/c/254788 and https://review.openstack.org/#/c/353506/ ? Because those two patches are doing basically the same thing and one received +2 from Dan and the other -1 from Emilien. It would be great if we could agree if those patches are OK or not and eventually merge them today. | 14:50 |
| hewbrocca | heh | 14:50 |
| *** jtomasek has joined #tripleo | 14:53 | |
| dprince | paramite: your patch needs a rebase. | 14:54 |
| paramite | doh | 14:54 |
| dprince | paramite: I think I'd be happy for both of these to land | 14:54 |
| dprince | EmilienM: ^^ | 14:54 |
| paramite | dprince, ok, thanks | 14:54 |
| dprince | paramite: agree the approaches are similar. I +2'd fluent because it was passing CI at the time | 14:56 |
| openstackgerrit | Giulio Fidente proposed openstack/python-tripleoclient: Do not use selinux-permissive for the CentOS image https://review.openstack.org/360097 | 14:57 |
| *** vijaykc4 has joined #tripleo | 14:57 | |
| larsks | dprince: so, this is more of an architecture question, rather than a "does it pass ci question". Both approaches are the same, so we're sort of getting conflicting feedback on our chosen approach... | 14:57 |
| *** chem has quit IRC | 14:58 | |
| larsks | dprince: but yeah, so many rebases! :) | 14:58 |
| dprince | larsks: there are only a couple of ways to do this I think and much of the decision is where to manage the extra Fluent, and Sensu parameters I think | 14:58 |
| social | hmm lately successfull deploy of tripleo master means everything is broken it just passed the deployment :( | 14:58 |
| *** abregman_|mtg has quit IRC | 14:59 | |
| larsks | dprince: Exactly. I'm not sure I see a good alternative for aggregating parameters like we're doing. I'm hoping we can convince EmilienM that this is an okay solution... | 14:59 |
| dprince | larsks: like we spoke last week so long as we treat these parameters as pseudo internal (i.e. no descriptions on them, used for advanced settings) I thik we are ookay | 14:59 |
| dprince | OoOkay even | 15:00 |
| EmilienM | dprince: will look after mtg | 15:00 |
| larsks | EmilienM: thanks! | 15:00 |
| openstackgerrit | Martin Mágr proposed openstack/tripleo-heat-templates: Availability monitoring agents support https://review.openstack.org/254788 | 15:00 |
| social | hiera(): Cannot find datafile /etc/puppet/hieradata/extraconfig.json but still CREATE_COMPLETE :/ | 15:00 |
| *** rbowen has joined #tripleo | 15:01 | |
| *** adarazs_afk is now known as adarazs | 15:02 | |
| *** rhallisey has quit IRC | 15:03 | |
| *** rhallisey has joined #tripleo | 15:04 | |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Mv Nova, Neutron, Horizon out of controller.yaml https://review.openstack.org/361362 | 15:07 |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Mv pacemaker and firewall out of controller https://review.openstack.org/361398 | 15:07 |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Move snmp settings into composable services https://review.openstack.org/360062 | 15:07 |
| *** jaosorior_away is now known as jaosorior | 15:10 | |
| jaosorior | shadower: recheck does the regular recheck, timed out was just the reason I did the recheck | 15:10 |
| shadower | jaosorior: ah cool. Thanks | 15:10 |
| *** dprince has quit IRC | 15:11 | |
| *** zoli|dentist is now known as zoli | 15:11 | |
| *** zoli is now known as zoliXXL | 15:11 | |
| *** abregman_|mtg has joined #tripleo | 15:12 | |
| *** beagles is now known as beagles_brb | 15:15 | |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Add `overcloud parameters set` to set Heat params is a plan https://review.openstack.org/360391 | 15:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Print the execution ID when starting Mistral workflows https://review.openstack.org/340711 | 15:16 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Add `openstack overcloud plan deploy` https://review.openstack.org/360305 | 15:16 |
| openstackgerrit | Merged openstack/python-tripleoclient: Add 'openstack overcloud node configure' command https://review.openstack.org/344875 | 15:18 |
| openstackgerrit | Merged openstack/python-tripleoclient: Use Mistral for baremetal boot configuration https://review.openstack.org/358012 | 15:18 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Add `overcloud parameters set` to set Heat params in a plan https://review.openstack.org/360391 | 15:18 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Print the execution ID when starting Mistral workflows https://review.openstack.org/340711 | 15:18 |
| openstackgerrit | Dougal Matthews proposed openstack/python-tripleoclient: Add `openstack overcloud plan deploy` https://review.openstack.org/360305 | 15:18 |
| jaosorior | ayoung: hey man, so how did the mellon stuff go? Did you get a chance to get that working? | 15:20 |
| ayoung | jaosorior, never could get the module to load | 15:20 |
| jaosorior | ayoung: I saw there was some package-related stuff missing from puppet. Is that needed in the upstream apache modules? | 15:21 |
| openstackgerrit | Merged openstack/instack-undercloud: Fix outdated help text from undercloud VIPs https://review.openstack.org/361020 | 15:21 |
| openstackgerrit | Merged openstack/puppet-tripleo: Removing WARNING: line has more than 140 characters in puppet-tripleo profiles https://review.openstack.org/354054 | 15:21 |
| ayoung | jaosorior, I tried to keep the patch in sync with my understanding/what I was trying | 15:21 |
| ayoung | I think that it is necessary, but not yet sufficient | 15:21 |
| ayoung | jaosorior, that coupled with the Jenkis failures makes me somewhat puzzled | 15:22 |
| hewbrocca | Anybody working on doing a release of oooq this week | 15:22 |
| *** nyechiel has joined #tripleo | 15:23 | |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Introduce 'enable_validations' option https://review.openstack.org/322893 | 15:23 |
| openstackgerrit | Martin André proposed openstack/instack-undercloud: Deploy SSH keys to overcloud in post config https://review.openstack.org/362194 | 15:23 |
| jaosorior | ayoung: the errors in the gate seem unrelated | 15:23 |
| ayoung | jaosorior, ah, those are gone now, too | 15:24 |
| ayoung | that was older reviews...so, yeah, we are down to getting the Package to install | 15:24 |
| ayoung | jaosorior, any suggestion? Does it need to be in the Puppet module? Is that something I can test? | 15:25 |
| jaosorior | ayoung: first lets get it to work. Then we get it into puppetlabs-apache | 15:25 |
| openstackgerrit | Juan Antonio Osorio Robles proposed openstack/puppet-tripleo: Optionally include mod auth mellon for apache https://review.openstack.org/359942 | 15:27 |
| jaosorior | ayoung: do you have a log of the errors you were seeing? | 15:28 |
| jaosorior | ayoung: I got the package to install in my undercloud testing manually the stuff I just put up. But I'm not sure if that's all that's needed | 15:28 |
| ayoung | jaosorior, this is overcloud deploy | 15:29 |
| jaosorior | ayoung: yeah, just wanted to see if it actually installs. (gotta test it somehow cause right now I'm doing another deployment testing HAProxy TLS) | 15:30 |
| ayoung | jaosorior, I am not concerned about Federation for the undercloud, only the overcloud right now | 15:30 |
| jaosorior | ayoung: you're using the beaker machine, right? | 15:31 |
| ayoung | jaosorior, nope | 15:31 |
| ayoung | I am suing the Dell on my desktop right next to me | 15:31 |
| jaosorior | ayoung: is it accessible somehow? | 15:31 |
| jaosorior | ok | 15:32 |
| ayoung | jaosorior, so, yeah, but it is a PITA due to OOOqs using Keys and proxies. Let me put your key in the right places... | 15:32 |
| *** thrash is now known as thrash|biab | 15:33 | |
| *** nyechiel has quit IRC | 15:34 | |
| ayoung | jaosorior, .....and here is where I find my redeploy from Friday night failed. | 15:36 |
| jaosorior | ayoung: actually, mod_auth_mellon SHOULD be installed by the include we did. we shouldn't need that package statement | 15:36 |
| ayoung | jaosorior, one moment while I bring it back up. | 15:36 |
| jaosorior | ok | 15:36 |
| *** yamahata has joined #tripleo | 15:37 | |
| ayoung | jaosorior, and, by "moment" I mean I am doing an undercloud deploy...\ | 15:37 |
| *** xuao has joined #tripleo | 15:38 | |
| jaosorior | crap | 15:40 |
| *** tremble has quit IRC | 15:40 | |
| jaosorior | ayoung: just ran on my undercloud (like I mentioned earlier, to test that the puppet manifest is doing what it should) and it seems it removed my keystone configuration | 15:40 |
| openstackgerrit | Giulio Fidente proposed openstack/puppet-tripleo: Defaults to ceph_mon_node_ips and ceph_mon_node_names https://review.openstack.org/361108 | 15:40 |
| jaosorior | ayoung: http://paste.openstack.org/show/564595/ | 15:41 |
| jaosorior | EmilienM: Any idea why including ::apache::mod::auth_mellon would remove my keystone configuration? http://paste.openstack.org/show/564595/ | 15:41 |
| *** tesseract- has quit IRC | 15:42 | |
| *** pcaruana has quit IRC | 15:43 | |
| EmilienM | jaosorior: I guess you didn't include other keystone classes | 15:43 |
| jaosorior | EmilienM: didn't know it would purge every other configuration there | 15:44 |
| jaosorior | well crap | 15:44 |
| *** rcernin has quit IRC | 15:44 | |
| EmilienM | jaosorior: puppet manages apache config | 15:44 |
| *** leanderthal is now known as leanderthal|afk | 15:45 | |
| jaosorior | EmilienM: fair enough. It was just unexpected to me that it would purge all the rest if it's not in the catalog | 15:45 |
| *** chem has joined #tripleo | 15:47 | |
| *** yamahata has quit IRC | 15:48 | |
| *** vijaykc4 has quit IRC | 15:48 | |
| *** chem has quit IRC | 15:48 | |
| *** chem has joined #tripleo | 15:49 | |
| *** chem has quit IRC | 15:49 | |
| *** chem has joined #tripleo | 15:49 | |
| *** chem has quit IRC | 15:50 | |
| jaosorior | EmilienM: have you seen something like this? | 15:51 |
| jaosorior | Error: /Stage[main]/Main/Augeas[lvm.conf]: Could not evaluate: Saving failed, see debug | 15:51 |
| *** limao_ has quit IRC | 15:51 | |
| EmilienM | no | 15:51 |
| EmilienM | but it looks related to the undercloud thing with lvm | 15:51 |
| jaosorior | EmilienM: which thing? | 15:51 |
| jaosorior | Just hit that while trying to re-deploy my undercloud | 15:52 |
| *** chem has joined #tripleo | 15:54 | |
| trozet | dtantsur: hi | 15:57 |
| *** zoliXXL is now known as zoli|gone | 15:58 | |
| *** tobias_fiberdata has quit IRC | 15:58 | |
| EmilienM | jaosorior: https://github.com/openstack/instack-undercloud/commit/4c51d6c159dcdc977ad035e6ad9d0322d59ae9f0 | 15:59 |
| openstackgerrit | Merged openstack-infra/tripleo-ci: Print services statuses of nodes in CI https://review.openstack.org/352916 | 15:59 |
| *** zoli|gone is now known as zoli_gone-proxy | 15:59 | |
| *** yamahata has joined #tripleo | 15:59 | |
| *** limao has joined #tripleo | 16:00 | |
| openstackgerrit | Merged openstack/tripleo-heat-templates: Add flag to assert that puppet manages the keystone endpoints https://review.openstack.org/353307 | 16:00 |
| *** chem has quit IRC | 16:00 | |
| dtantsur | trozet, hi | 16:00 |
| *** thrash|biab is now known as thrash | 16:01 | |
| *** david-lyle_ is now known as david-lyle | 16:01 | |
| trozet | dtantsur: I'm seeing in an OOO deployment to baremetal, the node fails to ipxe boot, and says "undionly". It seems to get an IP, then says configuring net0 unidonly and times out | 16:01 |
| *** chem has joined #tripleo | 16:02 | |
| trozet | dtantsur: i see multiple bugs filed on this from last year: https://bugs.launchpad.net/ironic/+bug/1454332, https://bugs.launchpad.net/ironic/+bug/1415017 | 16:02 |
| openstack | Launchpad bug 1454332 in ironic (Ubuntu) "Fail to chainload iPXE (undionly.kpxe) " [Undecided,Confirmed] | 16:02 |
| openstack | Launchpad bug 1415017 in Ironic "failed deploy undionly.kpxe then can't get ipxe script boot.ipxe" [Undecided,Fix released] | 16:02 |
| trozet | dtantsur: these are old Kontron servers, any ideas? | 16:02 |
| dtantsur | trozet, my first guess is old iPXE ROM on servers | 16:02 |
| dtantsur | lucasagomes, any ideas here ^^^ | 16:03 |
| trozet | dtantsur: i have a screen shot of the boot let me get it for you | 16:03 |
| * lucasagomes looks | 16:03 | |
| *** numans has quit IRC | 16:04 | |
| jaosorior | hewbrocca: time to open my celebration beer | 16:04 |
| jaosorior | hewbrocca: overcloud deploy should now be skipping the post-config :D | 16:05 |
| hewbrocca | WOOOOOOOOOO | 16:05 |
| hewbrocca | os-cloud-config.die.die.die.die.die.die | 16:05 |
| gfidente | jaosorior++ | 16:05 |
| gfidente | :) | 16:05 |
| jaosorior | guess we can't delete o-c-c from being called yet, but it's it's progress :D | 16:05 |
| lucasagomes | trozet, hmm strange so it keeps looping and fetching undionly.kpxe over and over ? | 16:06 |
| hewbrocca | so happy | 16:06 |
| jaosorior | life is good :D | 16:06 |
| trozet | lucasagomes, dtantsur: https://drive.google.com/file/d/0B-jVuJkFCkUPc05jTzNXVkNwNzQ/view?usp=sharing | 16:06 |
| lucasagomes | trozet, that looks fine actually, so it loaded iPXE and now it's configuring the nic | 16:07 |
| gfidente | jaosorior, I think we can because pki init was already supefluous | 16:07 |
| lucasagomes | trozet, what happens next ? | 16:07 |
| trozet | lucasagomes: right hten it just sits here and times out | 16:07 |
| trozet | lucasagomes: i can see on the undercloud it says sent tftpboot to the IP | 16:08 |
| dtantsur | trozet, the iPXE version looks correct | 16:08 |
| jaosorior | gfidente: thought we had to support n-1 t-h-t ? | 16:08 |
| lucasagomes | trozet, probably it's failing to get an IP from the DHCP server, can you check with tcpdump and see if iPXE is sending DHCPREQUEST packets ? | 16:08 |
| *** fzdarsky is now known as fzdarsky|afk | 16:09 | |
| trozet | lucasagomes: yeah I need to get vpn access to this pod so it will take me a little bit | 16:09 |
| lucasagomes | trozet, $ sudo tcpdump -i any port 69 should do it | 16:09 |
| trozet | lucasagomes: but you suspect hte pxe boot is fine, and its some network problem? | 16:09 |
| trozet | lucasagomes: or some configuration problem on undercloud | 16:09 |
| dtantsur | I would take -i port 67 or port 68 or port 69 | 16:09 |
| jaosorior | EmilienM: any idea where the logs go to from the calls that puppet-augeas do? | 16:10 |
| trozet | dtantsur, lucasagomes: ok will do that hten report back with what I find | 16:10 |
| gfidente | jaosorior, right sorry, forgot n-1 :) | 16:10 |
| *** ifarkas is now known as ifarkas_afk | 16:11 | |
| lucasagomes | trozet, right, it could be network problem but it's a bit unlikely since the PXE worked before (it chainloaded to iPXE) | 16:11 |
| dtantsur | folks, have we had a promotion recently? | 16:11 |
| lucasagomes | trozet, it could be a problem wiht the ipxe image itself, I also would recommend testing with the ipxe image from ipxe.org to see if the problem for that nic was fixed already | 16:12 |
| lucasagomes | trozet, you can download it from here http://boot.ipxe.org/ (the latest version already compiled from master) | 16:12 |
| *** beagles_brb is now known as beagles | 16:12 | |
| lucasagomes | trozet, see this too: http://ipxe.org/err/040ee1 | 16:12 |
| lucasagomes | dtantsur, ? | 16:13 |
| openstackgerrit | Merged openstack-infra/tripleo-ci: Extract /var/log from all nodes https://review.openstack.org/361164 | 16:13 |
| * lucasagomes might be missing the promotion context | 16:13 | |
| dtantsur | lucasagomes, this is unrelated to ironic question, please ignore :) | 16:13 |
| trozet | lucasagomes: ok thanks | 16:13 |
| *** abregman_|mtg has quit IRC | 16:17 | |
| *** ohamada has quit IRC | 16:19 | |
| *** atom_ has quit IRC | 16:20 | |
| *** dprince has joined #tripleo | 16:21 | |
| *** bana_k has joined #tripleo | 16:23 | |
| *** jpich has quit IRC | 16:25 | |
| hewbrocca | Folks, I gotta head out | 16:25 |
| hewbrocca | Keep those patches coming | 16:25 |
| *** limao has quit IRC | 16:26 | |
| dtantsur | hmmm, is it a known problem on master: | 16:27 |
| dtantsur | Invalid input [name=tripleo.baremetal.v1.register_or_update, class=DirectWorkflowSpec, unexpected=[u'instance_boot_option']] | 16:27 |
| dtantsur | some mismatch between the client and workflows? | 16:28 |
| dtantsur | dprince, have you tried https://review.openstack.org/#/c/361459/ locally anyhow? My new local environment does not work... | 16:28 |
| *** yamahata has quit IRC | 16:28 | |
| *** abregman has joined #tripleo | 16:30 | |
| *** jprovazn has joined #tripleo | 16:30 | |
| dprince | dtantsur: haven't tried it | 16:31 |
| dprince | dtantsur: this is why I separated the patches FWIW | 16:31 |
| dtantsur | sigh... ok, I'll continue repairing my env, will report back | 16:31 |
| dprince | dtantsur: but the local IP configuration was definately wrong in the first patch | 16:31 |
| dtantsur | dprince, it was wrong but working :) so now I want to make sure we switch to a correct and working, not to a correct and broken | 16:32 |
| *** abregman has quit IRC | 16:34 | |
| paramite | dprince, EmilienM: seems like both https://review.openstack.org/#/c/254788/ https://review.openstack.org/#/c/353506/ are green, can you guys check it before there will be another merge conflic? | 16:34 |
| florianf | r-KoyS8& | 16:35 |
| dprince | dtantsur: could it be Ironic API needs to be on the ctplane too since IPA will try to contact back on it? | 16:36 |
| *** bana_k has quit IRC | 16:36 | |
| dtantsur | dprince, yeah, this is correct. good catch | 16:36 |
| dtantsur | but that goes to haproxy configuration, right? | 16:37 |
| dprince | dtantsur: we should just be able to change the value in ServiceNetMap I think | 16:37 |
| dtantsur | not sure I understand it :( | 16:38 |
| dtantsur | hmm, or maybe I do. sorry, pretty late here, my brain does not work too well | 16:39 |
| *** Guest35918 is now known as mgagne | 16:39 | |
| dtantsur | dprince, are you planning to update your patch with it? | 16:39 |
| *** mgagne has quit IRC | 16:39 | |
| *** mgagne has joined #tripleo | 16:39 | |
| dprince | dtantsur: commented on the patch, I will update if you agree | 16:39 |
| dtantsur | dprince, yeah, I agree. thanks! | 16:40 |
| openstackgerrit | Dan Prince proposed openstack/tripleo-heat-templates: Fix Ironic service bind IPs https://review.openstack.org/361459 | 16:40 |
| dprince | dtantsur: okay, lets see how this plays | 16:40 |
| dprince | dtantsur: thanks for the feedback on this | 16:40 |
| dtantsur | thank you! my understanding of OoO networking leaves a lot to be desired still | 16:41 |
| dtantsur | I'll test it as soon as I figure out broken workflows | 16:41 |
| *** tobias_fiberdata has joined #tripleo | 16:41 | |
| *** hewbrocca is now known as hewbrocca-afk | 16:43 | |
| *** lucasagomes is now known as lucas-afk | 16:45 | |
| *** aufi has quit IRC | 16:45 | |
| *** jcoufal has joined #tripleo | 16:46 | |
| *** jbadiapa has quit IRC | 16:48 | |
| *** ioggstream has quit IRC | 16:49 | |
| openstackgerrit | Giulio Fidente proposed openstack/tripleo-heat-templates: Remove hard-coded per-service rabbitmq hieradata https://review.openstack.org/361731 | 16:49 |
| *** jtomasek has quit IRC | 16:50 | |
| bnemec | jaosorior: So we want https://review.openstack.org/#/c/290571 now? | 16:51 |
| jaosorior | bnemec: I believe we do | 16:52 |
| jaosorior | bnemec: that is not getting ran if you use master :D | 16:52 |
| EmilienM | paramite: done | 16:52 |
| *** Arkady_Kanevsky has joined #tripleo | 16:52 | |
| EmilienM | paramite: with some comments though | 16:52 |
| bnemec | jaosorior: Okay, I'll update that then. If we have puppet init for Newton then we probably want the deprecation warning for Newton too. | 16:53 |
| paramite | EmilienM, well LoggigSOurces and LoggingGroups is also on the similar base as MonitoringSubscriptions or ServiceNames | 16:54 |
| Arkady_Kanevsky | giulio, per you request bringing swift/ceph chat here | 16:54 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/python-tripleoclient: Updated from global requirements https://review.openstack.org/361875 | 16:54 |
| Arkady_Kanevsky | question is do we want to support an option that we have single proxy on controllers that will be swift one, and does that config automatically use local ceph RGW on controller node | 16:55 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/tripleo-common: Updated from global requirements https://review.openstack.org/361501 | 16:55 |
| gfidente | tty tomorrow guys | 16:55 |
| gfidente | bye all | 16:56 |
| EmilienM | gfidente: o/ | 16:56 |
| jaosorior | gfidente: have a good one | 16:56 |
| EmilienM | jaosorior: not sure about logs | 16:56 |
| Arkady_Kanevsky | Alternative is to use current 2 proxies: one for swift via HA proxy, and one for Ceph. Not sure if Ceph will be udner ha-proxy or not | 16:56 |
| EmilienM | jaosorior: iscsi? | 16:56 |
| jaosorior | EmilienM: not sure either, and didn't figure it out | 16:56 |
| jaosorior | EmilienM: for now, I'm getting a strange augeas error related to lvm.conf | 16:56 |
| jaosorior | with the latest undercloud | 16:57 |
| EmilienM | we might need to revert it | 16:57 |
| jaosorior | EmilienM: what merged related to lvm.conf? | 16:57 |
| tdasilva | Arkady_Kanevsky: what's the use case for having both a swift proxy and RGW node? | 16:57 |
| jaosorior | EmilienM: it's failing on update, FWIW | 16:58 |
| EmilienM | dang | 16:58 |
| EmilienM | we really need the undercloud upgrade job I did | 16:58 |
| Arkady_Kanevsky | swift front (API & horizon) and ceph providing object stoarge via RGW | 16:58 |
| jaosorior | EmilienM: might be this then https://review.openstack.org/#/c/360790/1/elements/puppet-stack-config/puppet-stack-config.pp | 16:58 |
| openstackgerrit | Emilien Macchi proposed openstack-infra/tripleo-ci: Implement undercloud upgrade job - Mitaka -> Newton https://review.openstack.org/346995 | 16:58 |
| EmilienM | let's see | 16:58 |
| Arkady_Kanevsky | my appologies to jumpinng in the middle of other discusoisn | 16:59 |
| *** gfidente has quit IRC | 16:59 | |
| jaosorior | I gotta go, but will probably get back to this tomorrow | 16:59 |
| EmilienM | jaosorior: thx I'll look too this aft | 16:59 |
| jaosorior | Have a good rest of the day folks! | 16:59 |
| *** lucas-afk is now known as lucasagomes | 17:00 | |
| *** jaosorior has quit IRC | 17:00 | |
| dprince | bnemec: you might review this. I need this setting for tenant vlan networks https://review.openstack.org/#/c/361539/ | 17:00 |
| *** vijaykc4 has joined #tripleo | 17:00 | |
| tdasilva | Arkady_Kanevsky: sorry, I must be missing some context. I'm not sure what you mean by "swift front (API & horizon)" If ceph is going to provide object storage via RGW, what's the need for a swift proxy behind HAProxy and vice versa? | 17:02 |
| dprince | bnemec: specifically my external net requires it when using network isolation | 17:02 |
| *** chem has quit IRC | 17:03 | |
| vijaykc4 | dprince, we are waiting for our patches to get past the OVB CI and get our workflows... considering today's the last day how does it work? | 17:04 |
| openstackgerrit | John Browning proposed openstack/tripleo-validations: Added check for keystone token_flush cronjob. https://review.openstack.org/355557 | 17:05 |
| dprince | vijaykc4: after today you'd apply for feature freeze exceptions | 17:05 |
| openstackgerrit | Martin Mágr proposed openstack/tripleo-heat-templates: [WIP] Heat stack for monitoring/logging node https://review.openstack.org/349690 | 17:06 |
| *** yamahata has joined #tripleo | 17:07 | |
| bnemec | dprince: Bleh, MTU | 17:07 |
| bnemec | +2 on the patch | 17:08 |
| *** jpena is now known as jpena|off | 17:09 | |
| *** tobias_fiberdata has quit IRC | 17:09 | |
| *** jtomasek has joined #tripleo | 17:10 | |
| *** tobias_fiberdata has joined #tripleo | 17:11 | |
| *** paramite has quit IRC | 17:11 | |
| vijaykc4 | dprince, thanks. | 17:12 |
| larsks | dprince: EmilienM: well, not mmagr as a +2 and I have a -1, so we're at about the same place. EmilienM, re: your comment, I can't put those configs in the individual service definition because they need the aggregate values which are only available from the ServiceChain. I could probably move those up into overcloud.yaml instead; does that make sense? | 17:12 |
| Arkady_Kanevsky | <tdasilva>: as a user I want to use swift APIs. But I donot want to deplpoy swift. Instead I want to use new or existing Ceph cluster. and configure swift to use Ceph for it. | 17:14 |
| Arkady_Kanevsky | This works now. But currently there are 2 proxys. One for swift as part of swift confug. And another part of ceph RGW config | 17:14 |
| trown | derekh: did the mirror server move? or does tripleo not publish any images any more? been out for a few weeks :) | 17:14 |
| dsneddon | dprince, This change should be done in conjunction with removing the NeutronTenantMtu and instead using the NeutronGlobalPhysnetMtu to calculate things like path_mtu. If you'd prefer, I can do that in a follow up patch to https://review.openstack.org/#/c/361539 | 17:17 |
| dprince | dsneddon: NeutronTenantMTU was removed almost 2 months ago | 17:18 |
| dprince | dsneddon: I just pushed this because I needed it to get my local dev environment working | 17:19 |
| dsneddon | dprince, Oops, my git repo is not on master branch. | 17:19 |
| dsneddon | dprince, thanks | 17:19 |
| dprince | dsneddon: apparently nobody is testing network isolation with an external network on a Vlan besides me :) | 17:19 |
| dprince | dsneddon: I mentioned this in my commit message, but there is a potential upgrade impact here | 17:20 |
| dsneddon | dprince, I've been having a tough time testing master because I don't have enough RAM to handle Newton virt envs, so I'm behind on my testng. | 17:20 |
| dprince | dsneddon: Neutron as I understand it is supposed to leave the MTU's of previously existing networks alone, but users deploying from scratch may hit issues due to this config change | 17:20 |
| dprince | dsneddon: so in hindsight perhaps we should have treated the removal of NeutronTenantMTU a bit more carefullly in TripleO... | 17:21 |
| dsneddon | dprince, Hmm, I'm surprised that you would need to set GlobalPhysnetMtu lower than physical. I was under the impression that Neutron would do the right thing and subtract the overhead from the physical MTU for VXLAN. | 17:22 |
| dsneddon | dprince, But I'll take a look at that and see if there is someplace we need to take that into account, like we did with forcing the MTU through dnsmasq based on the NeutronTenantMtu. | 17:22 |
| beagles | dsneddon, dprince, if neutron isn't subtracting the tunnel overhead there is a bug somewhere | 17:23 |
| dprince | dsneddon, beagles: neutron only claims to auto magically set the MTU for tunnelled networks like GRE and VXLAN I think | 17:24 |
| beagles | dsneddon, dprince, unless the MTU of the actual underlying physical network is less than 1500, you shouldn't have to set anything at all | 17:24 |
| beagles | dprince, yeah.. you are talking about VLAN right? | 17:24 |
| dprince | dsneddon, beagles: for VLAN, flat networks the *global* setting gets used instead | 17:24 |
| beagles | dprince, yup that's right | 17:24 |
| dsneddon | beagles, And if the physical network is less than 1500, you should still only have to set the global physical mtu equal to the real MTU. | 17:24 |
| beagles | dsneddon, correct | 17:24 |
| EmilienM | larsks: I'm ok with a first iteratio | 17:25 |
| EmilienM | larsks: just keep in mind the files won't exist anymore | 17:25 |
| EmilienM | +2 on both | 17:25 |
| larsks | EmilienM: what do you think about my proposal for moving those settings into overcloud.yaml? Or are you thinking of a different solution? | 17:26 |
| EmilienM | larsks: it would be great indeed | 17:28 |
| openstackgerrit | Harry Rybacki proposed openstack/tripleo-quickstart: Revert "Move unique socket generation into quickstart.sh" https://review.openstack.org/362292 | 17:30 |
| *** bana_k has joined #tripleo | 17:31 | |
| *** trown is now known as trown|lunch | 17:32 | |
| *** yamahata__ has quit IRC | 17:33 | |
| beagles | dsneddon: in your patch, when you say "previous" and "has" are you referring to before this patch or before newton? | 17:33 |
| beagles | dsneddon, the global physnet mtu patch, I'm referring to | 17:33 |
| *** maeca1 has quit IRC | 17:34 | |
| *** electrofelix has quit IRC | 17:37 | |
| beagles | dprince, is your "infrastructure"'s network MTU less than 1500? | 17:38 |
| dprince | beagles: no | 17:38 |
| dprince | beagles: just normal I think | 17:39 |
| *** pkovar has quit IRC | 17:39 | |
| * beagles sees the three letter MTU and goes snaky these days ;) | 17:41 | |
| dprince | beagles: http://paste.openstack.org/show/564631/ | 17:42 |
| dprince | beagles: once I apply that setting I set my global_physnet_mtu: 1400 | 17:43 |
| dprince | beagles: and it has the effect of making my vlan network show up like that which is correct | 17:43 |
| *** tosky has quit IRC | 17:43 | |
| dprince | beagles: without it I'm able to ping instances, but I can't ssh to them. Classic MTU issue | 17:44 |
| *** fragatina has quit IRC | 17:44 | |
| beagles | dprince, is this a fresh system or post upgrade? | 17:44 |
| dprince | beagles: totally fresh | 17:44 |
| karthiks | EmilienM, For https://review.openstack.org/#/c/361350/, we've used recursion in puppet for iterating a list. Do you see any issues here ? | 17:44 |
| dprince | beagles: this setting is required for me for all new deployments from master | 17:44 |
| dprince | beagles: to make it work like it used to... i.e. having an external network on a vlan (segment 10) | 17:45 |
| *** tobias_fiberdata has quit IRC | 17:45 | |
| dtantsur | folks, is it time already to request FFEs? I see people doing that... | 17:46 |
| beagles | dprince, is this an ovb thing? | 17:46 |
| dprince | beagles: I don't use OVB locally :) | 17:46 |
| dprince | beagles: these are real machines. | 17:46 |
| dprince | beagles: its a *real* thing | 17:46 |
| EmilienM | karthiks: where? | 17:46 |
| karthiks | EmilienM, https://review.openstack.org/#/c/361350/9/manifests/host/sriov/numvfs_persistence.pp | 17:47 |
| dprince | beagles: real machines, Cisco switch, real vlans. Same setup that worked for Mitaka.... | 17:47 |
| EmilienM | karthiks: this should work | 17:47 |
| karthiks | Thanks EmilienM . Just to check if the approach is ok :) | 17:48 |
| beagles | dprince, this strikes me as odd | 17:49 |
| dprince | beagles: perhaps I'm missing something? Is my setup odd you mean? Or the fact that this need to be set? | 17:49 |
| beagles | dprince, that this needs to be set | 17:50 |
| dprince | beagles: well, if you read the Neutron release notes I think it is intended behavior | 17:50 |
| skramaja | dtantsur: we are already at the end of 29th accorrding to IST :) | 17:51 |
| dprince | beagles: and yes, I am impling there is a hidden bug in that Neutron has refactored the way MTU's are handled. They have upgrades code in place... but this will still confuse people deploying with the new code | 17:52 |
| dtantsur | skramaja, FF is not on 29th, it's on the week of 29th, hence my question | 17:53 |
| *** xuao has quit IRC | 17:54 | |
| *** rajinir has joined #tripleo | 17:55 | |
| karthiks | dtantsur, we thought FF is on 29th. isn't it dprince ? | 17:57 |
| dprince | karthiks, dtantsur: see the Plans re newton-3 release and feature freeze exceptions' | 17:59 |
| dprince | email to the list from Steve | 17:59 |
| dtantsur | dprince, it does not have dates, that's the problem :) | 17:59 |
| dprince | I think the plan is to tag the release tomorrow | 17:59 |
| dprince | dtantsur: ^^ | 17:59 |
| dtantsur | (unless I've missed them) | 17:59 |
| dtantsur | ah cool. so after tomorrow I start begging for FFE for ironic work, right? :) | 17:59 |
| dprince | dtantsur: I guess :) | 17:59 |
| dtantsur | cool | 18:00 |
| rook | Anyone have a osp+ceph cluster up... curious if you could try something :/ | 18:01 |
| *** isq has joined #tripleo | 18:01 | |
| rook | newton cloud. | 18:01 |
| dprince | rook: talk to me | 18:01 |
| rook | dprince hey man, if you have a instance on your overcloud... if you issue nova image-create <guest you want to copy> <name to put into glance> ; does it work? | 18:02 |
| rook | dprince: I am getting a traceback on the computes : https://gist.github.com/jtaleric/4e0367d9f78e822deefdc541ee7ec28b | 18:02 |
| dprince | rook: ack, trying | 18:02 |
| rook | however I am not finding much out there... | 18:02 |
| rook | w/o ceph, this operation works great | 18:02 |
| *** fragatina has joined #tripleo | 18:02 | |
| *** dtantsur is now known as dtantsur|afk | 18:02 | |
| dprince | rook: worked for me, the image went 'ACTIVE' at least | 18:03 |
| dprince | rook: looking at the logs... | 18:03 |
| rook | wtf | 18:03 |
| rook | dprince interesting. | 18:03 |
| dprince | rook: did your image go active? | 18:04 |
| rook | dprince negative.. the guest went down, and never came back up.. no image was created. | 18:04 |
| rook | dprince was this image raw? | 18:04 |
| rook | maybe that is the hiccup | 18:04 |
| dprince | rook: could be | 18:04 |
| dprince | rook: I was using RBD, ceph backed instance | 18:04 |
| rook | i import into glance a qcow image (stupid) instead of raw... | 18:04 |
| rook | yeah, me too | 18:04 |
| beagles | dprince, sorry, was off down a rabbit hole. Are you setting an MTU of 1400 somewhere, like in an undercloud.conf or something? | 18:05 |
| dprince | beagles: no, just this new setting | 18:05 |
| beagles | dprince, oh.. it wasn't 1400 before? | 18:05 |
| rook | dprince: let me try setting the image to raw and try again... I am also seeing a MTU issue with Newton :/ | 18:05 |
| dprince | rook: then my patch might fix you! | 18:06 |
| beagles | dprince, wati.. backing up | 18:06 |
| beagles | dprince, the paste you sent before, was that 1400 before or after your patch? | 18:06 |
| dprince | beagles: correct, before it wasn't since we revoved NeutronTenantMTU | 18:06 |
| dprince | beagles: after | 18:06 |
| beagles | dprince, ah okay | 18:06 |
| dprince | beagles: if I didn't apply my patch then ext-net would ping but ssh fails | 18:07 |
| rook | dprince: word... can we please figure out a way to get MTU right :P | 18:07 |
| rook | yeah, ssh fails... but so does large user-data payloads from nova | 18:07 |
| dprince | rook: https://review.openstack.org/#/c/361539/ | 18:07 |
| dprince | rook: set that to 1400 and it WFM | 18:07 |
| ayoung | do we havea doc or something explaining how mistral is going to work in with Heat for openstack overcloud deploy? | 18:09 |
| dprince | ayoung: https://specs.openstack.org/openstack/tripleo-specs/specs/mitaka/tripleo-mistral-deployment-library.html | 18:10 |
| ayoung | dprince, thanks | 18:10 |
| *** trown|lunch is now known as trown | 18:11 | |
| *** jcoufal has quit IRC | 18:11 | |
| ayoung | dprince, and Mistral is going to be part of the Newton release, right? | 18:11 |
| dprince | ayoung: it has to be, critical feature | 18:12 |
| dprince | ayoung: Mistral itself was in the undercloud for Mitaka. Just new... | 18:13 |
| dprince | ayoung: we are only now starting to use it | 18:13 |
| ayoung | dprince, and it is only going to be used to kick off the Heat workflow from the undercloud, right? It will not e something that we can use during an overcloud deployment to do complex workflows? | 18:15 |
| dprince | ayoung: it is already used for several things. It is a generic in-cloud workflow service | 18:16 |
| dprince | ayoung: I used it to implement 'openstack overcloud execute' for example | 18:16 |
| *** mcornea has quit IRC | 18:17 | |
| ayoung | dprince, so when we deploy an overcloud, we need to do some IPA operations that really should not be executed on the controller or compute nodes. THey would have to happen in the middle of an overcloud deploy. So heat would call Mistral. That is not part of what we can do now, right? But, I could have a bunch of stuff done in mistral prior to calling heat for, say an overcloud deployment? | 18:18 |
| dprince | ayoung: right, we haven't gotten into that business yet. But we can. | 18:18 |
| dprince | ayoung: I would like to do a similar thing for Swift rings for example | 18:19 |
| *** isq has quit IRC | 18:19 | |
| dprince | ayoung: we want to add to our composable services with "composable workflows" interface I think | 18:19 |
| dprince | ayoung: so that any service can tie into it and do some custom workflow'y thinks | 18:19 |
| dprince | ayoung: haven't built that yet... I think it is more of an O thing | 18:20 |
| *** isq has joined #tripleo | 18:20 | |
| ayoung | dprince, OK, I'll go do my homework. The thing I like about working through Mistral is we can set it up with a Kerberos principal, set up access controls , and do all the security hardening properly to do things like create new hosts and services, grant new hosts elevated degrees of permissions if required and all that. It really should not be done on the controller, but right now that is our only option. | 18:21 |
| openstackgerrit | Sarath Kumar Sankaran Kutty proposed openstack/os-net-config: Cleanups and new unit test for IVS https://review.openstack.org/346161 | 18:23 |
| *** maeca1 has joined #tripleo | 18:24 | |
| *** jcoufal has joined #tripleo | 18:25 | |
| *** isq has quit IRC | 18:26 | |
| *** chem has joined #tripleo | 18:28 | |
| *** lucasagomes is now known as lucas-dinner | 18:33 | |
| *** isq has joined #tripleo | 18:34 | |
| *** jcoufal has quit IRC | 18:36 | |
| *** jcoufal has joined #tripleo | 18:37 | |
| openstackgerrit | Merged openstack/tripleo-common: Workflows to load validations https://review.openstack.org/353898 | 18:38 |
| *** liverpooler has quit IRC | 18:47 | |
| *** jbadiapa has joined #tripleo | 18:50 | |
| *** jcoufal has quit IRC | 18:51 | |
| *** jcoufal has joined #tripleo | 18:52 | |
| *** oshvartz has joined #tripleo | 18:59 | |
| EmilienM | dprince: any chance to review https://review.openstack.org/#/c/360039/ ? slagle had +2 on it before I rebased | 19:00 |
| dprince | EmilienM: looking | 19:00 |
| EmilienM | dprince: it's non voting jobs FYI | 19:01 |
| openstackgerrit | Emilien Macchi proposed openstack/tripleo-heat-templates: configure Heat with new authtoken parameters https://review.openstack.org/361412 | 19:04 |
| *** jcoufal has quit IRC | 19:06 | |
| *** jcoufal has joined #tripleo | 19:07 | |
| *** fzdarsky|afk has quit IRC | 19:14 | |
| *** mhenkel has quit IRC | 19:14 | |
| openstackgerrit | Jeff Peeler proposed openstack/tripleo-common: Add List Roles Action https://review.openstack.org/330283 | 19:16 |
| *** rasca has quit IRC | 19:21 | |
| *** rbowen has quit IRC | 19:24 | |
| openstackgerrit | Michele Baldessari proposed openstack/puppet-pacemaker: WIP DO NOT MERGE for location_rule constraints https://review.openstack.org/362335 | 19:25 |
| *** rbowen has joined #tripleo | 19:25 | |
| *** radeks has joined #tripleo | 19:25 | |
| rook | so dprince looked at this MTU thing with beagles some.. | 19:31 |
| rook | I am not convinced using a 1400 mtu is the right thing.. | 19:31 |
| rook | I was able to leave my MTUs at 1450 across the env-- except the L3, i set that to 1446 and everything still worked. | 19:32 |
| rook | *accounting for a 4byte vlan* which I am not sure why that isn't being stripped | 19:32 |
| rook | dsneddon ^^^^ | 19:32 |
| dprince | rook: what you set it to is up to you | 19:33 |
| dprince | rook: the point is making it configurable | 19:33 |
| rook | dprince: ok - as long as the default isn't taking us back in time to a 1350 MTU | 19:33 |
| dprince | rook: I just used 1400 because that was the default value of our old NeutronTenantMtu setting in t-h-t | 19:33 |
| rook | roger. | 19:33 |
| rook | dprince just making sure. | 19:33 |
| dprince | rook: so that setting worked for you then? | 19:33 |
| rook | there is a configuration issue somewhere -- i believe we should be stripping tags before it hits the tunnel | 19:34 |
| dprince | rook: or you did it manually some other way? | 19:34 |
| dsneddon | dprince, rook: Yeah, 1400 should never be necessary. You only need 50 bytes in the worst case. | 19:34 |
| rook | dprince manually tinkering | 19:34 |
| rook | dsneddon yeah man, 50 for the vxlan oh. | 19:34 |
| rook | overhead, i am getting bad about shorting words. | 19:34 |
| rook | dsneddon: however, i think the vlan tag is making it into the vxlan tunnel, which bumps things another 4bytes. | 19:35 |
| rook | this is a hunch. /me waits for beagles to backme up or say I am crazy | 19:35 |
| dprince | rook: all my patch does is make use of an existing Neutron flag. I think it is safe to expose it in t-h-t | 19:35 |
| dsneddon | rook, What VXLAN tunnel? I thought dprince was having trouble on his external network, which is a VLAN network. | 19:35 |
| dprince | dsneddon: yes, my ext-net is a vlan network | 19:36 |
| dprince | dsneddon: which gets used for the floating IPs for my instances | 19:36 |
| rook | dsneddon ah, same problem different envs. | 19:36 |
| rook | the instances use a vxlan tunnel? | 19:36 |
| rook | or not | 19:36 |
| rook | mine was, ext-net (vlan) tenant-net(vxlan) | 19:36 |
| *** skramaja has quit IRC | 19:38 | |
| *** fragatina has quit IRC | 19:39 | |
| dprince | rook: exactly, that is what I'm doing too. ext-net is the vlan network | 19:39 |
| *** fragatina has joined #tripleo | 19:39 | |
| vijaykc4 | EmilienM, dprince the builds are done for these patches: https://review.openstack.org/#/c/361350/ https://review.openstack.org/#/c/361430/... need your attention please. | 19:41 |
| *** pradk has quit IRC | 19:46 | |
| EmilienM | vijaykc4: ack | 19:46 |
| *** vijaykc4 has quit IRC | 19:47 | |
| openstackgerrit | Merged openstack/tripleo-docs: Deploy and Scale Swift https://review.openstack.org/293311 | 19:48 |
| colonwq | EmilienM, For the scenario004 patch, I am working on getting RGW to work as an endpoint with SSL. | 19:49 |
| EmilienM | colonwq: ok | 19:50 |
| dprince | rbrady: hi, left some feedback on https://review.openstack.org/#/c/342261/ | 19:54 |
| dprince | rbrady: hit an odd template error | 19:54 |
| dprince | rbrady: something munged my network template or something... | 19:54 |
| rbrady | dprince: did you happen to add a template that did not contain json or yaml? | 19:56 |
| *** sshnaidm is now known as sshnaidm|afk | 19:57 | |
| openstackgerrit | Christian Schwede proposed openstack/tripleo-docs: Update advanced Swift deployment doc https://review.openstack.org/360353 | 19:57 |
| openstackgerrit | Christian Schwede proposed openstack/tripleo-docs: Include Swift ring building example https://review.openstack.org/361032 | 19:57 |
| *** pradk has joined #tripleo | 19:59 | |
| dprince | rbrady: nothing out of the ordinary | 20:02 |
| dprince | rbrady: I'm using the same templates from t-h-t master | 20:02 |
| dprince | rbrady: I use network isolation, perhaps that is the difference here? | 20:03 |
| *** derekh has quit IRC | 20:03 | |
| *** radeks has quit IRC | 20:04 | |
| *** maeca1 has quit IRC | 20:04 | |
| larsks | does anyone have a moment to chat about "split stack" deployments as we have previously discussed w/r/t the opstools work? | 20:09 |
| rbrady | dprince: maybe. can you tell if this error is being thrown by template_utils.process_multiple_environments_and_files or is it from heat.stacks.create ? | 20:10 |
| rbrady | dprince: and what is in custom.yaml? | 20:10 |
| *** vijaykc4 has joined #tripleo | 20:12 | |
| openstackgerrit | Michele Baldessari proposed openstack/puppet-tripleo: WIP DO NOT MERGE Initial Composable HA https://review.openstack.org/362389 | 20:13 |
| *** trown is now known as trown|outtypewww | 20:16 | |
| vijaykc4 | EmilienM, dprince : thanks... hope we can close these patches tomorrow and the CI errors will be resolved. Good night | 20:19 |
| *** vijaykc4 has quit IRC | 20:22 | |
| *** milan has quit IRC | 20:26 | |
| dprince | rbrady: custom.yaml is my own YAML file | 20:32 |
| dprince | rbrady: it is how I inject parameters in convieniently | 20:32 |
| beagles | rook: not crazy.. there is something weird if bumping your physical device's MTU to 9000 but nothing else works, or that dropping the effective MTU of the tenant network at the router by 4 bytes does the trick.. there is something else really weird | 20:38 |
| openstackgerrit | Ben Nemec proposed openstack/python-tripleoclient: Re-enable keystone init deprecation message https://review.openstack.org/290571 | 20:39 |
| *** jbadiapa has quit IRC | 20:42 | |
| *** akrivoka has quit IRC | 20:45 | |
| openstackgerrit | Florian Fuchs proposed openstack/tripleo-ui: Port plan-edit file listing to use Swift https://review.openstack.org/361561 | 20:47 |
| *** dprince has quit IRC | 20:49 | |
| *** florianf has quit IRC | 20:50 | |
| openstackgerrit | Athlan-Guyot sofer proposed openstack/tripleo-heat-templates: M/M upgrade sahara-api fails to restart. https://review.openstack.org/358022 | 20:50 |
| *** jayg is now known as jayg|g0n3 | 20:54 | |
| openstackgerrit | wes hayutin proposed openstack/tripleo-quickstart: pin the undercloud-post and overcloud roles https://review.openstack.org/356669 | 21:04 |
| openstackgerrit | Mathieu Mitchell proposed openstack/diskimage-builder: Generate and use upper-constraints for ironic-agent https://review.openstack.org/360547 | 21:05 |
| *** Goneri has quit IRC | 21:09 | |
| *** ioggstream has joined #tripleo | 21:18 | |
| *** jprovazn has quit IRC | 21:22 | |
| *** jcoufal has quit IRC | 21:24 | |
| *** dmacpher has quit IRC | 21:24 | |
| *** Arkady_Kanevsky has quit IRC | 21:24 | |
| *** jcoufal has joined #tripleo | 21:24 | |
| *** dmacpher has joined #tripleo | 21:36 | |
| *** lucas-dinner has quit IRC | 21:56 | |
| openstackgerrit | Ryan Brady proposed openstack/tripleo-common: WIP wire in jinja templating for custom roles https://review.openstack.org/362465 | 21:56 |
| *** wfoster has quit IRC | 21:56 | |
| *** sai has quit IRC | 21:58 | |
| *** rook has quit IRC | 21:58 | |
| *** wfoster has joined #tripleo | 21:59 | |
| *** sai has joined #tripleo | 21:59 | |
| *** lucasagomes has joined #tripleo | 21:59 | |
| *** rook has joined #tripleo | 22:00 | |
| *** rook is now known as Guest64782 | 22:00 | |
| *** rhallisey has quit IRC | 22:01 | |
| *** ayoung has quit IRC | 22:07 | |
| *** assassin has quit IRC | 22:12 | |
| *** ioggstream has quit IRC | 22:33 | |
| *** fragatin_ has joined #tripleo | 22:35 | |
| *** yamahata has quit IRC | 22:36 | |
| *** fragatina has quit IRC | 22:38 | |
| *** rlandy has quit IRC | 22:43 | |
| *** Goneri has joined #tripleo | 22:43 | |
| *** yamahata has joined #tripleo | 22:55 | |
| openstackgerrit | Emilien Macchi proposed openstack/tripleo-heat-templates: configure Heat with new authtoken parameters https://review.openstack.org/361412 | 23:04 |
| *** rajinir has quit IRC | 23:05 | |
| *** mpsairam has joined #tripleo | 23:06 | |
| *** egafford has quit IRC | 23:12 | |
| *** jmiu has quit IRC | 23:12 | |
| *** jmiu has joined #tripleo | 23:13 | |
| *** pradk has quit IRC | 23:17 | |
| *** ayoung has joined #tripleo | 23:20 | |
| *** saneax-_-|AFK is now known as saneax | 23:20 | |
| openstackgerrit | Merged openstack/puppet-tripleo: Configure the numvfs for SRIOV interfaces https://review.openstack.org/361350 | 23:29 |
| *** myoung is now known as myoung|afk | 23:31 | |
| *** fultonj has quit IRC | 23:34 | |
| *** Guest64782 is now known as rook | 23:34 | |
| openstackgerrit | Merged openstack-infra/tripleo-ci: Implement scenario001 CI job https://review.openstack.org/360039 | 23:35 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!