| *** diogogmt has quit IRC | 00:02 | |
| *** gfidente has quit IRC | 00:12 | |
| *** tfukushima has joined #kolla | 00:13 | |
| *** dayou has quit IRC | 00:13 | |
| *** devlaps has quit IRC | 00:15 | |
| *** pbourke_ has quit IRC | 00:18 | |
| *** pbourke_ has joined #kolla | 00:18 | |
| *** tfukushima has quit IRC | 00:20 | |
| *** achanda_ has quit IRC | 00:31 | |
| *** diogogmt has joined #kolla | 00:32 | |
| *** achanda has joined #kolla | 00:33 | |
| *** achanda has quit IRC | 00:38 | |
| *** diogogmt has quit IRC | 00:41 | |
| *** diogogmt has joined #kolla | 00:42 | |
| *** jruano has quit IRC | 00:51 | |
| *** jruano has joined #kolla | 00:58 | |
| *** dims_ has quit IRC | 01:01 | |
| *** erkules has joined #kolla | 01:06 | |
| *** jtriley has quit IRC | 01:08 | |
| *** erkules_ has quit IRC | 01:09 | |
| *** jruano has quit IRC | 01:15 | |
| *** achanda has joined #kolla | 01:22 | |
| *** kjelly has joined #kolla | 01:25 | |
| *** dims_ has joined #kolla | 01:26 | |
| *** rhallisey has quit IRC | 01:31 | |
| *** diogogmt has quit IRC | 01:32 | |
| *** diogogmt has joined #kolla | 01:33 | |
| *** tfukushima has joined #kolla | 01:36 | |
| *** tfukushima has quit IRC | 01:37 | |
| *** achanda has quit IRC | 01:41 | |
| *** jtriley has joined #kolla | 01:44 | |
| *** weiyu has joined #kolla | 01:45 | |
| *** dayou has joined #kolla | 01:50 | |
| *** tfukushima has joined #kolla | 01:50 | |
| *** jruano has joined #kolla | 01:51 | |
| openstackgerrit | Qiu Yu proposed openstack/kolla: Expose HAProxy stats for management purpose https://review.openstack.org/224380 | 01:55 |
|---|---|---|
| *** jerryz has quit IRC | 02:01 | |
| *** unicell has quit IRC | 02:11 | |
| *** jtriley has quit IRC | 02:28 | |
| *** jtriley has joined #kolla | 02:29 | |
| *** jtriley has quit IRC | 02:32 | |
| *** achanda has joined #kolla | 02:41 | |
| *** achanda has quit IRC | 02:45 | |
| *** achanda has joined #kolla | 03:03 | |
| *** loth has quit IRC | 03:08 | |
| *** diogogmt has quit IRC | 03:10 | |
| *** coolsvap is now known as coolsvap|holiday | 03:17 | |
| *** tfukushima has quit IRC | 03:32 | |
| *** tfukushima has joined #kolla | 03:32 | |
| *** achanda has quit IRC | 03:35 | |
| *** achanda has joined #kolla | 03:35 | |
| *** sdake_ has joined #kolla | 03:51 | |
| *** sdake has quit IRC | 03:54 | |
| *** unicell has joined #kolla | 04:03 | |
| *** unicell has quit IRC | 04:03 | |
| *** unicell has joined #kolla | 04:03 | |
| *** jruano has quit IRC | 04:20 | |
| *** unicell1 has joined #kolla | 04:32 | |
| *** unicell has quit IRC | 04:33 | |
| *** weiyu has quit IRC | 04:36 | |
| *** dims_ has quit IRC | 04:38 | |
| *** achanda has quit IRC | 04:39 | |
| *** tfukushima has quit IRC | 04:42 | |
| *** dims_ has joined #kolla | 04:43 | |
| *** dims_ has quit IRC | 04:44 | |
| *** sdake has joined #kolla | 04:46 | |
| *** sdake_ has quit IRC | 04:49 | |
| *** unicell1 has quit IRC | 04:51 | |
| *** unicell has joined #kolla | 04:51 | |
| *** unicell has quit IRC | 04:58 | |
| *** achanda has joined #kolla | 05:20 | |
| *** unicell has joined #kolla | 05:28 | |
| SamYaple | morning | 05:34 |
| *** tfukushima has joined #kolla | 05:36 | |
| *** weiyu_ has joined #kolla | 05:38 | |
| *** weiyu_ has quit IRC | 05:42 | |
| *** dims_ has joined #kolla | 05:45 | |
| *** gfidente has joined #kolla | 05:51 | |
| *** dims_ has quit IRC | 05:53 | |
| *** jtriley has joined #kolla | 05:57 | |
| *** jtriley has quit IRC | 06:03 | |
| openstackgerrit | Martin André proposed openstack/kolla: Implement configuration file for kolla-build https://review.openstack.org/224442 | 06:09 |
| *** weiyu_ has joined #kolla | 06:19 | |
| *** tfukushima has quit IRC | 06:26 | |
| openstackgerrit | Martin André proposed openstack/kolla: Update documentation for Vagrant environment https://review.openstack.org/223420 | 06:31 |
| openstackgerrit | Martin André proposed openstack/kolla: Set namespace to point to local registry https://review.openstack.org/223419 | 06:31 |
| *** tfukushima has joined #kolla | 06:32 | |
| *** banix has joined #kolla | 06:34 | |
| sdake | hey SamYaple you about? | 06:55 |
| sdake | nice work on ceph dude ! | 06:55 |
| openstackgerrit | Merged openstack/kolla: Expose HAProxy stats for management purpose https://review.openstack.org/224380 | 06:58 |
| SamYaple | sdake: i am here! | 07:00 |
| SamYaple | that set_configs.py broke all the things | 07:00 |
| SamYaple | i wish it had been tested | 07:00 |
| SamYaple | i shouldnt assume people test | 07:00 |
| openstackgerrit | Martin André proposed openstack/kolla: Move build.ini to etc/kolla/source-location.conf https://review.openstack.org/224459 | 07:02 |
| *** tfukushima has quit IRC | 07:03 | |
| *** tfukushima has joined #kolla | 07:03 | |
| *** tfukushima has quit IRC | 07:05 | |
| *** tfukushima has joined #kolla | 07:06 | |
| *** weiyu_ has quit IRC | 07:18 | |
| *** jmccarthy has joined #kolla | 07:19 | |
| *** weiyu has joined #kolla | 07:20 | |
| *** shardy has joined #kolla | 07:25 | |
| *** banix has quit IRC | 07:26 | |
| sdake | samyaple yo | 07:26 |
| sdake | shame set_configs broke shit | 07:27 |
| SamYaple | eh it ws going to get an immediate refactor anyway | 07:27 |
| sdake | Ryan assured me he tested it,but I was pretty suspicious of that after i saw -DFOREGROUDN missing | 07:27 |
| SamYaple | thats what im doing now | 07:27 |
| sdake | i'm not sure how keystone even worked for him | 07:27 |
| SamYaple | it didnt | 07:27 |
| SamYaple | that wasnt tested | 07:27 |
| SamYaple | but the permissions broke for galera | 07:27 |
| sdake | ya thats why I -1 voted on that patch | 07:27 |
| openstackgerrit | Martin André proposed openstack/kolla: Implement configuration file for kolla-build https://review.openstack.org/224442 | 07:28 |
| sdake | well mitaka - we are going full monty on functional testing | 07:28 |
| sdake | and i'm going to try to get infrastructure cats to mirror the repos so we can gate on all the things | 07:28 |
| sdake | and vote the gates | 07:29 |
| sdake | anyway I'm headed to bed | 07:29 |
| sdake | night :) | 07:29 |
| SamYaple | mirror what repos? | 07:29 |
| sdake | whichever ones we need to make the gates work reliably | 07:30 |
| SamYaple | thats alot of stuff and the entire ubuntu repo.... | 07:30 |
| sdake | ya, may not be feasible, going to try before giving up tho :) | 07:30 |
| SamYaple | yup | 07:30 |
| sdake | if we don't get the mirros in infra, we will never have a voting gate | 07:30 |
| sdake | which imo is not acceptable for a mature project | 07:31 |
| sdake | although atm, our project isn't mature, so its a-ok for the moment | 07:31 |
| SamYaple | you mean you dont want to run 'recheck' 3-4 times to get something to merge :) | 07:31 |
| sdake | or recheck for 2 days while rdo upstream repos are broke | 07:31 |
| SamYaple | ubuntu test is pretty solid | 07:31 |
| sdake | its not hte recheck i fear, its the 2 day "our xfs filesystem is busted" rdo mess | 07:31 |
| sdake | ya i'e never seen an ubuntu gate fail | 07:32 |
| SamYaple | but we really need some early abort type stuff so we aren't gating ubuntu when only centos changed | 07:32 |
| sdake | unless something was legitimately wrong with teh patch | 07:32 |
| sdake | i was thinking of making the ubuntu gate voting, but I didn't want to disrupt the liberty rc1/rc2/rc3 releases with that | 07:33 |
| sdake | what we really need is a deployment gate | 07:33 |
| SamYaple | nah im fine with leaving it alone for now | 07:33 |
| sdake | aio would do :) | 07:33 |
| sdake | even nonvoting would be helpful | 07:33 |
| SamYaple | yea but that gate would _also_ have to build | 07:33 |
| SamYaple | the current gates should deploy | 07:33 |
| sdake | yup | 07:33 |
| sdake | yup that could work although the cost to us is low to have one build and one build+deploy | 07:34 |
| SamYaple | we dont need differnt built/deploy gates if deploy also builds | 07:34 |
| sdake | our AIO deploy gate will be flakey for awhile, while the build gate won't be | 07:34 |
| SamYaple | i guess but what id rather see is a build+push to infra docker registry then another gate trigger off that | 07:34 |
| sdake | since teh build gtes are pretty solid now unless upstream repos are busted | 07:34 |
| sdake | i am pretty sure we can't do that :) | 07:35 |
| sdake | anyway 1am, and I'v ebeen burning the candle from both ends for over a year | 07:35 |
| sdake | need some R&R | 07:35 |
| sdake | so ttyl | 07:35 |
| openstackgerrit | Martin André proposed openstack/kolla: Implement configuration file for kolla-build https://review.openstack.org/224442 | 07:36 |
| *** erkules has quit IRC | 07:38 | |
| *** erkules has joined #kolla | 07:38 | |
| *** achanda has quit IRC | 07:38 | |
| *** dwalsh has joined #kolla | 07:41 | |
| *** shardy_ has joined #kolla | 07:47 | |
| *** shardy has quit IRC | 07:47 | |
| *** achanda has joined #kolla | 07:47 | |
| *** achanda has quit IRC | 07:50 | |
| *** dims_ has joined #kolla | 07:51 | |
| *** shardy_ has quit IRC | 07:51 | |
| openstackgerrit | Martin André proposed openstack/kolla: Update documentation for Vagrant environment https://review.openstack.org/223420 | 07:52 |
| openstackgerrit | Martin André proposed openstack/kolla: Set namespace to point to local registry https://review.openstack.org/223419 | 07:52 |
| *** shardy has joined #kolla | 07:52 | |
| SamYaple | night sdake | 07:52 |
| openstackgerrit | Martin André proposed openstack/kolla: Move build.ini to etc/kolla/source-location.conf https://review.openstack.org/224459 | 07:53 |
| *** dims_ has quit IRC | 07:57 | |
| -openstackstatus- NOTICE: Gate is currently stuck, failing grenade upgrade tests due the release of oslo.utils 1.4.1 for Juno. | 08:00 | |
| *** inc0 has joined #kolla | 08:08 | |
| inc0 | good morning | 08:08 |
| SamYaple | morning inc0 | 08:09 |
| *** akwasnie has joined #kolla | 08:10 | |
| inc0 | SamYaple, so how do we make optional bindmounting? | 08:11 |
| inc0 | I thought we need to redefine task and use "when" statement | 08:11 |
| SamYaple | nope much simplier | 08:12 |
| SamYaple | im just waiting on your patch to merge to do it project wide | 08:12 |
| inc0 | it won't merge without it | 08:12 |
| SamYaple | ignore sdake's comment on that | 08:12 |
| SamYaple | it will | 08:12 |
| inc0 | lol | 08:12 |
| inc0 | just make patchset dependant on mine | 08:12 |
| inc0 | and sdake will be happy | 08:12 |
| inc0 | I'll be happier person as well tbh;) | 08:13 |
| SamYaple | your patch is still waiting on a few services implemented to show it logs correctly, right? | 08:14 |
| SamYaple | like nova with multiple files or something | 08:14 |
| inc0 | well, I'll make new commit with keystone | 08:14 |
| SamYaple | do glance if you dont mind | 08:14 |
| *** dayou has quit IRC | 08:14 | |
| SamYaple | that has registry adn api | 08:14 |
| inc0 | and split services per patchset | 08:14 |
| inc0 | glance it is then | 08:14 |
| SamYaple | plus keystone uses apache which isnt a good meausre of openstack config file changes | 08:15 |
| inc0 | so this patchset won't really change | 08:15 |
| SamYaple | i know | 08:15 |
| inc0 | feel free to make dependency on it | 08:15 |
| SamYaple | i will, got alot of stuff right now | 08:15 |
| SamYaple | ceph is looking good | 08:15 |
| SamYaple | but im sidetracked by this new set_configs.py thing | 08:15 |
| SamYaple | it broke galera :/ | 08:15 |
| inc0 | then just tell me what's your idea for optional bindmount | 08:15 |
| SamYaple | so im refactoring now | 08:15 |
| inc0 | I'll do it | 08:16 |
| SamYaple | dude i need to do it projectwide, its fine | 08:16 |
| SamYaple | please focus on the logging | 08:16 |
| inc0 | well, *we* need to | 08:16 |
| SamYaple | thats teh big ? here | 08:16 |
| inc0 | we as community | 08:16 |
| SamYaple | i understand, its handled. dont worry. please address my concerns as they are far more important | 08:16 |
| inc0 | ok, going to do glance then | 08:17 |
| inc0 | just sdake really wants optional bindmount and I don't want to bypass his -1 really | 08:17 |
| SamYaple | right, but we arent starting out with the hardcoded bindmount | 08:18 |
| SamYaple | start out with teh data container | 08:18 |
| SamYaple | he doesnt want the optional bindmount, he wants to make sure its not _required_ | 08:18 |
| inc0 | rsyslog container will be sort of data container | 08:19 |
| inc0 | remove volume - it's done | 08:19 |
| inc0 | so optional bindmounting is all we need | 08:19 |
| SamYaple | no it still needs the data container treatment we do with everything else | 08:19 |
| SamYaple | nova_data, glance_data, mariadd_date | 08:19 |
| SamYaple | log_data | 08:19 |
| inc0 | well, I talked with sdake about it and he was ok with this serving as both | 08:20 |
| inc0 | let's not be overzealotus about it please | 08:20 |
| SamYaple | so the logs will get wiped on container upgrade | 08:20 |
| SamYaple | trust me he isn't ok with it | 08:20 |
| inc0 | (I hope I've used this fancy word correctly) | 08:20 |
| SamYaple | he just doesnt know the situtation | 08:20 |
| SamYaple | youll need log_data | 08:20 |
| inc0 | are you planning to upgrade rsyslog?;) | 08:21 |
| SamYaple | it doesnt matter, an upgrade is a new container | 08:21 |
| SamYaple | rsyslog will get rebuilt like the rest | 08:21 |
| inc0 | same as if we upgrade log_data | 08:21 |
| openstackgerrit | Martin André proposed openstack/kolla: Implement configuration file for kolla-build https://review.openstack.org/224442 | 08:21 |
| SamYaple | i dont really want to go over the architeture again for why we do different data containers | 08:22 |
| inc0 | so you want data container in this patch? that's no problem for me | 08:22 |
| openstackgerrit | Martin André proposed openstack/kolla: Move build.ini to etc/kolla/source-location.conf https://review.openstack.org/224459 | 08:23 |
| inc0 | just getting mixed messages here | 08:23 |
| SamYaple | yea itll need to be there so we can completely remove all of the bind mount stuff (/dev/log exluded) | 08:23 |
| SamYaple | im sorry sdake is confusing everyone. but this is the implementation that the rest of the project has | 08:23 |
| openstackgerrit | Martin André proposed openstack/kolla: Update documentation for Vagrant environment https://review.openstack.org/223420 | 08:23 |
| openstackgerrit | Martin André proposed openstack/kolla: Set namespace to point to local registry https://review.openstack.org/223419 | 08:23 |
| *** dayou has joined #kolla | 08:31 | |
| *** weiyu has quit IRC | 08:37 | |
| *** dwalsh has quit IRC | 08:38 | |
| *** weiyu_ has joined #kolla | 08:58 | |
| *** dayou has quit IRC | 09:05 | |
| *** athomas has joined #kolla | 09:17 | |
| *** chandankumar has joined #kolla | 09:18 | |
| *** dims_ has joined #kolla | 09:59 | |
| *** cfarquhar has quit IRC | 10:06 | |
| *** cfarquhar has joined #kolla | 10:08 | |
| *** cfarquhar has quit IRC | 10:08 | |
| *** cfarquhar has joined #kolla | 10:08 | |
| -openstackstatus- NOTICE: Gate back to normal, thanks to the backlisting of the problematic version | 10:14 | |
| *** weiyu_ has quit IRC | 10:15 | |
| *** weiyu_ has joined #kolla | 10:19 | |
| *** weiyu_ has quit IRC | 10:19 | |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Refactor set_configs.py https://review.openstack.org/224557 | 10:22 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Refactor set_configs.py https://review.openstack.org/224557 | 10:26 |
| SamYaple | inc0 mandre pbourke_ I would appreciate a review on https://review.openstack.org/224557 | 10:27 |
| SamYaple | right now mariadb is completely busted | 10:27 |
| SamYaple | need to get that fixed asap | 10:27 |
| inc0 | uhh you rewrote whole thing really | 10:27 |
| SamYaple | yep | 10:28 |
| SamYaple | i didnt really want to, but the problems were pretty deep and directory copies didnt work at all | 10:28 |
| *** dayou has joined #kolla | 10:35 | |
| SamYaple | going to step away for a bit | 10:35 |
| *** dayou has quit IRC | 10:40 | |
| *** tfukushima has quit IRC | 10:53 | |
| *** chandankumar has quit IRC | 10:55 | |
| *** jtriley has joined #kolla | 10:58 | |
| *** mfalatic_ has quit IRC | 11:00 | |
| *** mfalatic has joined #kolla | 11:00 | |
| *** jtriley has quit IRC | 11:03 | |
| *** daneyon_ has quit IRC | 11:03 | |
| inc0 | SamYaple, quick question - this whole refactor was about perms to files? | 11:05 |
| openstackgerrit | Merged openstack/kolla: Remove ansible_ssh_user from Kolla configs https://review.openstack.org/224005 | 11:29 |
| openstackgerrit | Merged openstack/kolla: Use Liberty cloudarchive repo for Ubuntu https://review.openstack.org/223986 | 11:33 |
| *** dwalsh has joined #kolla | 11:41 | |
| *** sdake_ has joined #kolla | 11:43 | |
| sdake_ | pbourke_ you around? | 11:44 |
| *** sdake has quit IRC | 11:47 | |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Ansible bits to rsyslog https://review.openstack.org/215062 | 11:53 |
| sdake_ | vbel around? | 11:53 |
| vbel | sdake_: yes | 11:53 |
| sdake_ | vbel i dont understand what you requested I do in the review about rdo from source | 11:54 |
| sdake_ | why not alter the top level docker file and get it working now? | 11:54 |
| sdake_ | (why dont you I mean0 | 11:54 |
| sdake_ | top level/base | 11:54 |
| vbel | I thought you add rdo to base image, was I wrong? :) | 11:54 |
| vbel | if not I will do that | 11:54 |
| SamYaple | inc0: no the refactor was already happening because the whole file was a mess, the perms just got rolled in with the fix | 11:56 |
| inc0 | SamYaple, any chances you could split making mariadb working from refactoring of code please? | 11:56 |
| vbel | sdake_: will do | 11:57 |
| SamYaple | inc0: nope, lots of change | 11:57 |
| SamYaple | it wasnt a single line | 11:57 |
| inc0 | but I really don't feel comfortable reviewing that much of code change with time pressure of broken codebase:/\ | 11:57 |
| SamYaple | you dont have to review it inc0 | 11:58 |
| inc0 | will try to do it, but well, next time please don't mix it | 11:58 |
| inc0 | I know I don't | 11:58 |
| inc0 | but I will | 11:58 |
| SamYaple | inc0: they are together | 11:58 |
| vbel | General question: should not we delete source archive and extracted files after install from most of containers? | 11:58 |
| SamYaple | i would have had to refactor to fix the perms | 11:58 |
| inc0 | still, you could do refactor first and perms later | 11:58 |
| SamYaple | refactoring fixed the perms | 11:58 |
| SamYaple | what am i not saying correctrly | 11:59 |
| inc0 | that would be easier to review as by refactoring you tend to keep same behaviour | 11:59 |
| vbel | i ask because if you do daily builds registry will blow up some day :) it always has difference in that layer | 11:59 |
| *** jtriley has joined #kolla | 11:59 | |
| SamYaple | inc0: i get what you are saying, but i am also not going to do the work twice to fix an issue | 12:00 |
| inc0 | also...we could use more test cases;) | 12:00 |
| inc0 | SamYaple, next time keep this in mind tho please | 12:00 |
| SamYaple | if we come up with another situation where an entire file needs a rewrite, probably still going to rewrite the entire file | 12:00 |
| SamYaple | but i dont normally bundle fixes together | 12:01 |
| sdake_ | well I highly recommend refactors done right, that is not rewriting a file | 12:01 |
| sdake_ | but what is done is done | 12:02 |
| sdake_ | if the code was more mature, and working and in use, the change would likely not be accepted ;) | 12:02 |
| vbel | Another question: we add bridge to compute node though it usually does not have spare nic and there is no need in it | 12:03 |
| SamYaple | yep | 12:03 |
| SamYaple | but the plan was always an immediate refactor anyway, or so say the original reviews | 12:03 |
| *** rhallisey has joined #kolla | 12:03 | |
| sdake_ | for example, if someone rewrote build.py in a non-incremental refactor, i'd pretty much -1 that change :) | 12:03 |
| SamYaple | if it wasnt for a time crunch i wouldnt have voted +2 on the orignal code | 12:03 |
| sdake_ | understood | 12:03 |
| sdake_ | so I think we are all in violent agreement :) | 12:04 |
| SamYaple | garrr | 12:04 |
| sdake_ | rewrites are bad and are not refactors typically :) | 12:04 |
| SamYaple | yea | 12:04 |
| sdake_ | but when its brand spanking new code, who cares | 12:04 |
| SamYaple | pbourke_: sdake_ i dont disagree about the sys.exit() comments, i dont like that either. but lets adress that in not such a large patch | 12:05 |
| sdake_ | samyaple i did a review, and paul looks to have covered all the points | 12:05 |
| *** jtriley has quit IRC | 12:05 | |
| SamYaple | i started on it and it was just too different | 12:05 |
| SamYaple | yea im addressign now | 12:05 |
| sdake_ | ya we can put it off to later | 12:05 |
| sdake_ | i'd like it fixed down the road if possible | 12:05 |
| sdake_ | not essential work atm | 12:05 |
| SamYaple | honestly i would be ok with a more better refactor right after this code lands, i jsut want it to be functional first | 12:05 |
| sdake_ | ya lets try to unbreak the codebase today if we can | 12:06 |
| sdake_ | what was the issue with permissionsbtw? | 12:06 |
| SamYaple | well for single file it would attempt to set the permissions on the parent directory | 12:07 |
| SamYaple | _and_ the permissions were not being converted to octal correctly | 12:07 |
| SamYaple | so the permissions were actually wrong | 12:07 |
| SamYaple | being applied to the wrong place | 12:07 |
| sdake_ | hmm bummer missed that in the review | 12:08 |
| SamYaple | that function wasn't easily reviewable | 12:08 |
| SamYaple | it started out as a simple fix but the refactor was actually cleaner | 12:08 |
| sdake_ | i read about 20 versions of that code - should have picked up on it :) | 12:08 |
| SamYaple | it changed quite a bit | 12:08 |
| SamYaple | and the octal thing isn't something people normally catch | 12:08 |
| SamYaple | i didn't catch it and i know about it explicitly from my other anible module | 12:09 |
| sdake_ | in mitaka we should probably not accept python code without unit tests | 12:09 |
| sdake_ | and writ eunit tests for the python we have | 12:09 |
| SamYaple | honestly, well be lucky to have unit tests for our current python | 12:10 |
| sdake_ | letting it slide now becausse unit tests are much harder to write then the code and I want the cod to land | 12:10 |
| SamYaple | it will probably be N before we can justify new python unit tests | 12:10 |
| sdake_ | ya unit tests for threaded code is very difficult | 12:10 |
| sdake_ | unit tests for that set_configs would be relatively easy to write | 12:11 |
| sdake_ | i find that 80% of my time writing python code in mature open stack projects is writing unit test code tho | 12:11 |
| sdake_ | the actual implementation is easy - unit tests are challenging | 12:11 |
| sdake_ | vbel i did add rdo to base image, but we are still waiting on mandre to weigh in on the thread re whether we would accept rhos into kolla | 12:12 |
| sdake_ | vbel and i dont want to block for several days the build gating for source for that issue | 12:13 |
| rhallisey | morning | 12:13 |
| *** sdake_ is now known as sdake | 12:13 | |
| inc0 | SamYaple, is there a bug for this perm stuff? | 12:14 |
| vbel | sdake_: i see, thanks | 12:14 |
| SamYaple | inc0: no the code just merged yesterday | 12:14 |
| inc0 | sdake, instead of making bindmount optional I've made log_data container instead | 12:14 |
| sdake | inc0 ok | 12:15 |
| sdake | inc0 so does it work? :) | 12:15 |
| inc0 | well, it works for me;) | 12:15 |
| sdake | inc0 it looks like you are host bind mounting /var/log | 12:15 |
| inc0 | (I mean mountig /var/log) | 12:15 |
| inc0 | /dev/log is still bindmount | 12:15 |
| sdake | anything that is directory:directory is a host bind mount | 12:16 |
| sdake | anything that is directory is not a host bind mount | 12:16 |
| SamYaple | oh god not this again | 12:16 |
| SamYaple | see what i mean inc0 | 12:16 |
| SamYaple | sdake: you explicitly agreed to this | 12:16 |
| sdake | its not a data container if it has a host bind mount imo | 12:17 |
| sdake | the idea i thought was to make that optional | 12:17 |
| inc0 | logs lands in data container | 12:17 |
| inc0 | /dev/log has to be shared across containers | 12:17 |
| sdake | i get the part about /dev/log | 12:17 |
| sdake | not addressing that, that is fine | 12:17 |
| inc0 | that's the bindmount, nothing else | 12:17 |
| SamYaple | ^ | 12:17 |
| SamYaple | i dont know what the current up code looks like | 12:18 |
| SamYaple | but thats the final outcome | 12:18 |
| inc0 | /var/log is data container /var/lib/kolla/dev/log is device | 12:18 |
| inc0 | device is shared and it's on host | 12:18 |
| sdake | volumes: 18 | 12:18 |
| sdake | - "/var/log/kolla:/var/log"19 | 12:18 |
| inc0 | ah in data container | 12:18 |
| SamYaple | yea so that will be removed before we merge | 12:18 |
| inc0 | my bad | 12:19 |
| sdake | a data container is /var/log/kolla | 12:19 |
| sdake | not /var/log/kolla:/var/log | 12:19 |
| SamYaple | actually wouldn't it just be /var/log? | 12:19 |
| SamYaple | i dont think we need to through kolla on there | 12:19 |
| inc0 | ok, makes sense | 12:19 |
| SamYaple | /var/log/keystone /var/log/nova | 12:19 |
| inc0 | let me fix that | 12:19 |
| sdake | if you want it written to the host filesystem, there is no reason for a data container | 12:20 |
| sdake | the data container's purpose is to contain the data, not export it via a host bind mount - but we had talked about making that host bind mounting optional | 12:20 |
| sdake | so people could cat logs from the data container or look on the host if it was enabled via config options | 12:21 |
| openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla: Ansible bits to rsyslog https://review.openstack.org/215062 | 12:21 |
| SamYaple | to make the to ideologies play nicely we can bindmount from the datacontainer | 12:21 |
| inc0 | this is what you mean? | 12:21 |
| SamYaple | then everything is fine for everyone | 12:21 |
| SamYaple | optionally of course | 12:21 |
| *** dwalsh has quit IRC | 12:21 | |
| sdake | inc0 ya that looks good | 12:21 |
| *** jruano has joined #kolla | 12:21 | |
| sdake | but that wont write to the host filesystem which I think is what you are wanting to do | 12:21 |
| SamYaple | yea sdake ill be submitting a patch for that kolla wide once we are ready to merge | 12:22 |
| SamYaple | we need to do it in lots of place | 12:22 |
| sdake | do what, host bindmount? | 12:22 |
| inc0 | start.yml line 40 does host writing | 12:22 |
| SamYaple | yea | 12:22 |
| sdake | if you use the data container, only the data ontainer ends up doing a bind mount | 12:23 |
| sdake | inc0 what i recommend doing is conditionalizating line 20 - not sure how you do that in ansible | 12:23 |
| sdake | and i dont think you need privileged: true for the data container - can you test without i t? | 12:24 |
| inc0 | SamYaple knows but doesn't want to tell;) | 12:24 |
| SamYaple | inc0: ive told before | 12:24 |
| SamYaple | dont make me get logs | 12:24 |
| inc0 | haha | 12:24 |
| SamYaple | i said it needs to be done in a variable | 12:24 |
| SamYaple | but its not a big thing | 12:24 |
| SamYaple | my biggest concern is doing multiple files | 12:24 |
| SamYaple | thats a much biegger issue | 12:24 |
| inc0 | afk for now, corpo stuff calling | 12:25 |
| sdake | if the container sall use volumes_from, then multiple files will just happen automatically when the services write to /var/log | 12:25 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Refactor set_configs.py https://review.openstack.org/224557 | 12:25 |
| SamYaple | sdake: no the services will be writing through syslog. | 12:26 |
| sdake | oh right | 12:26 |
| sdake | 4am, brain not booted :) | 12:27 |
| sdake | actually 530 now | 12:27 |
| SamYaple | pbourke_ pretty sure all your concerns were addressed | 12:28 |
| sdake | sam quick question | 12:30 |
| SamYaple | no | 12:30 |
| sdake | line 78 | 12:30 |
| SamYaple | no questions | 12:30 |
| sdake | why not just do a shutil.copytree | 12:31 |
| sdake | or shutil.copy | 12:31 |
| SamYaple | new patchset line 78? | 12:31 |
| sdake | yup | 12:31 |
| openstackgerrit | Vladislav Belogrudov proposed openstack/kolla: Fix broken centos-source- build: add rdo repository https://review.openstack.org/223963 | 12:31 |
| sdake | just curious why the for src in os.listdir | 12:31 |
| SamYaple | because that would not work like you think | 12:31 |
| SamYaple | we need to copy the folder contents, not the folder | 12:31 |
| sdake | right | 12:31 |
| SamYaple | cp folder/* dest | 12:31 |
| sdake | i just figured it out prior to you answering | 12:31 |
| sdake | got i t | 12:31 |
| SamYaple | not cp folder dest | 12:31 |
| SamYaple | yea | 12:31 |
| SamYaple | i couldn't find a shutil to do that | 12:32 |
| SamYaple | but then i realized copy * was globing anyway | 12:32 |
| SamYaple | it basically does that function | 12:32 |
| SamYaple | is anyone going to review ceph :( | 12:33 |
| SamYaple | i know its a pita to review | 12:33 |
| sdake | question about line 105 | 12:34 |
| sdake | default permission is 0? | 12:34 |
| SamYaple | thats not what that does | 12:34 |
| sdake | would you mind explaining so i can make sure the review is solid :) | 12:35 |
| SamYaple | https://docs.python.org/2/library/functions.html#int | 12:35 |
| sdake | i see thanks | 12:35 |
| *** achanda has joined #kolla | 12:35 | |
| sdake | line 39, how do you enforce that in the code? | 12:36 |
| sdake | int(data.get('perm') without a perm key I think would be zero even in base 10 | 12:37 |
| SamYaple | oh that should be a note | 12:37 |
| SamYaple | like only source and dest _should_ | 12:37 |
| SamYaple | be required | 12:37 |
| SamYaple | thatsn ot how its currently writeen | 12:37 |
| SamYaple | the required_keys above have owner and perm | 12:38 |
| SamYaple | it will break without them | 12:38 |
| sdake | ok good enough | 12:38 |
| SamYaple | it is going to need to change later i think because we will be privilege dropping in each container | 12:38 |
| SamYaple | so we wont be able to set the owner anymore | 12:39 |
| SamYaple | but thats later, like M1 | 12:39 |
| sdake | line 121, 128 can be deduped i think | 12:40 |
| sdake | ya that should be fine samyaple | 12:40 |
| SamYaple | looks like they could be, youre right. moved directly above the if | 12:40 |
| *** dayou has joined #kolla | 12:40 | |
| sdake | well its not really worthy of a -1 | 12:41 |
| sdake | but if you want to fix it might as well do it now | 12:41 |
| SamYaple | ill fix if another patchset is needed | 12:41 |
| sdake | that is all I have for review comments | 12:42 |
| sdake | looks good | 12:42 |
| SamYaple | otherwise the whole thing is still in need of another round of refactor, but once the functionality works | 12:42 |
| SamYaple | ideally on a failure it would call a seperate function to print out lost of useful info | 12:43 |
| sdake | ok sam i'll review ceph today after i wake up a bit | 12:43 |
| sdake | thats a good idea | 12:44 |
| sdake | not sure how it woudl be implemented but seems like a good approach - true diags :) | 12:45 |
| SamYaple | if that failure function could call sys.exit, it could be implemented easily | 12:45 |
| SamYaple | since this is a util called only from commandline, not the worst idea | 12:45 |
| SamYaple | i mean thats what sys.exit() was built for | 12:46 |
| sdake | yup | 12:46 |
| sdake | the thing I dont like about the current code is it has a million exit points | 12:46 |
| sdake | but one function with one exit point sounds good | 12:46 |
| SamYaple | yea that would tie them up to two | 12:46 |
| SamYaple | success and failure | 12:46 |
| sdake | yup that sounds good for future work to me | 12:47 |
| sdake | vbel are you going to address https://review.openstack.org/207925 prior to release? | 12:54 |
| sdake | of rc1 | 12:54 |
| sdake | patch is almost 2 mo old | 12:54 |
| sdake | oh 1 month | 12:55 |
| sdake | SamYaple looks like pretty much everyone is in agreement re rhos implmeentation in the code base | 12:59 |
| sdake | you had a technical concern over the naming | 12:59 |
| sdake | shall I go with jpeeler's suggestion? | 12:59 |
| SamYaple | refresh my memory, whats was his suggestion? | 13:00 |
| sdake | second let me find the email so i can represent it accurately | 13:01 |
| sdake | So I see no reason to not support all the types suggested, albeit perhaps with a slightly different naming scheme: | 13:02 |
| sdake | source - obviously, using tarballs as defined in build.ini | 13:02 |
| sdake | binary - for whenever there is no distinction necessary for what type of packages are in use | 13:02 |
| sdake | binary-rdo - for the community version of RPM OpenStack packaging | 13:02 |
| sdake | binary-rhos - for the paid version of RPM OpenStack packaging | 13:02 |
| sdake | (cut and paste from email) | 13:02 |
| SamYaple | wouldn't that make binary non-existant? | 13:03 |
| sdake | ya i guess his proposal could be interepreted that way which is not internally consistent | 13:03 |
| sdake | what I liked about it was the naming scheme binary-rdo vs just rdo | 13:03 |
| sdake | perhaps binary | binary-rdo | 13:04 |
| SamYaple | what i dont want to see is just a bunch of types everywhere | 13:04 |
| sdake | or binary-rhos | 13:04 |
| SamYaple | oh i know | 13:04 |
| SamYaple | this would be very clean | 13:04 |
| SamYaple | we could have them all reduce to a single binary type in the base file | 13:05 |
| SamYaple | that would retain the ability to call them for specifics but have that generic type | 13:05 |
| SamYaple | that would be almost no change to the dockerfiles | 13:05 |
| SamYaple | and allow us to support n-number of types | 13:05 |
| sdake | ya only the base dockr file ends up with bloat which is ok, that thing is always going tobe a beast :) | 13:06 |
| SamYaple | sdake: as soon as someone fixes centos it wont be | 13:06 |
| sdake | how would you get specifics if you needed it? | 13:06 |
| SamYaple | centos isn't using openstack-base really at all :/ | 13:06 |
| sdake | ya I intend to do that shortly | 13:06 |
| sdake | waiting on my rebase patches first | 13:06 |
| SamYaple | if someone actually fixed centos base would be reasonable | 13:06 |
| pbourke_ | SamYaple: inc0: is 'docker logs' still the default way to get logs, even once rsyslog merges? | 13:06 |
| vbel | sdake: nope, I hope we all get good kernels so that sudo won't be necessary :) | 13:07 |
| SamYaple | pbourke_: no that wont be used | 13:07 |
| sdake | vbel if your going to abaondon the patch, can you abaondon the patch | 13:07 |
| pbourke_ | SamYaple: which wont, 'docker logs'? | 13:07 |
| vbel | sdake: ok | 13:07 |
| SamYaple | pbourke_: correct | 13:07 |
| sdake | and set teh bug state to invalid | 13:07 |
| sdake | pls | 13:07 |
| sdake | or wont fix | 13:07 |
| sdake | wont fix makes more sense here | 13:07 |
| SamYaple | pbourke_: i mean some garbage would pop out there, probably. but it would be garbage and should probably be cleaned up | 13:08 |
| pbourke_ | SamYaple: ok thanks | 13:08 |
| SamYaple | if its worth recording, it goes through the rsyslog container | 13:08 |
| SamYaple | if its not, it should be suppressed | 13:08 |
| sdake | at some point if docker ever figures out how to do logging properly we may need to revisit this discussion :) | 13:08 |
| pbourke_ | sounds fine | 13:08 |
| SamYaple | sdake: even if docker _does_ get logging right, it wont matter | 13:09 |
| sdake | so SamYaple how do I get specifics from inside a docker file if I needit? I don't think we will need it, but justcurious how that owuld be done | 13:09 |
| SamYaple | docker logs shows output, but not syslog out | 13:09 |
| SamYaple | we need syslog | 13:09 |
| sdake | i mean if they integrate with /dev/log | 13:09 |
| SamYaple | which they definetely shouldnt | 13:09 |
| sdake | ok dont care, want answer to my othe question :) | 13:10 |
| SamYaple | yea the docker logs things is a huge big ugly mess | 13:10 |
| SamYaple | best to ignore it forever | 13:10 |
| sdake | ya it blows donkey tails | 13:10 |
| SamYaple | i said it would have to be in base, i did not mean that | 13:11 |
| SamYaple | i meant in build.py | 13:11 |
| SamYaple | so we could collapse the types there into a single 'binary' and provide an additional variable with the specific type of binary (not a new cli option) | 13:11 |
| sdake | cool I like it, I'll make that happen today | 13:12 |
| SamYaple | so cli doesnt change and docker files dont change | 13:12 |
| sdake | i wanted to do somethign like that, but could only see to do with a cli addition which seemed sucky | 13:12 |
| SamYaple | yea thats no good | 13:12 |
| sdake | but shouldn't need that | 13:12 |
| SamYaple | not at all. could even be dynamic instead of an explict table | 13:13 |
| SamYaple | if we do the binary, binary-rdo, binary-rhos, binary-other scheme | 13:13 |
| sdake | i am pretty sure for rhos we dont need the explicit type | 13:14 |
| sdake | justneed the repos enabled | 13:14 |
| sdake | but just in case for the future might as well get it right now | 13:14 |
| SamYaple | well no i mean you would still need the explict type | 13:15 |
| SamYaple | it just would get converted into the generic type | 13:15 |
| SamYaple | and a second var would be produced | 13:15 |
| sdake | so the build.pt would take aprefix of binary- and generate a subtype variable based upon that? | 13:15 |
| SamYaple | meta-type or something | 13:15 |
| sdake | well base.j2 woudl need the subtype, but not the rest of the tree | 13:16 |
| sdake | and maybe openstack.j2 | 13:16 |
| SamYaple | sdake: the rest of the tree would need it as well | 13:16 |
| SamYaple | in case a package name is ever different | 13:16 |
| sdake | shouldn't, could just use the generic | 13:16 |
| SamYaple | who knows what tweaks will be needed between the teo | 13:16 |
| SamYaple | two* | 13:16 |
| SamYaple | thats way to optimistic | 13:17 |
| sdake | well the idea would be to make the variable available but only use if needed i think | 13:17 |
| SamYaple | thats my point yea | 13:17 |
| sdake | i think where we get into different package names is fedora, not centos + variants | 13:17 |
| sdake | and fedora is busted | 13:18 |
| SamYaple | `build.py --type binary` should still exist, with a default to rdo stuff | 13:18 |
| sdake | build.py --type binary-rdo should be rdo as well then? | 13:18 |
| SamYaple | but binary-rhos or binary-rdo still sets type==binary, but provides a metatype variable of rhos or rdo | 13:18 |
| SamYaple | yea | 13:18 |
| sdake | images get called distro-binary-rhos-foo? | 13:19 |
| sdake | or distro-binary-foo | 13:19 |
| sdake | i think distro-binarry-rhos-foo | 13:19 |
| sdake | SamYaple thanks for the suggestions, I like em, I'll get cracking in a bit on the implementation | 13:21 |
| *** openstackgerrit has quit IRC | 13:31 | |
| *** openstackgerrit has joined #kolla | 13:31 | |
| *** inc0 has quit IRC | 13:33 | |
| SamYaple | sdake: yea i dont know about the image naming, but the second one i guess is all we can do | 13:34 |
| *** banix has joined #kolla | 13:34 | |
| sdake | ya it needs to be unique | 13:34 |
| sdake | goign to call the variable install_metatype | 13:35 |
| sdake | the current variable is install_type | 13:35 |
| sdake | install_type = source|binary | 13:35 |
| sdake | install_metatype = rdo, rhos, nothing? | 13:36 |
| SamYaple | it would be much cleaner if we moved the specific types to the namespace :/ | 13:36 |
| SamYaple | sdake: yea | 13:36 |
| sdake | what do you mean to the namespace? | 13:36 |
| SamYaple | kolla-binary-rhos/mariadb, kolla-binary-rdo/mariadb, kolla-ubuntu-source/mariadb | 13:36 |
| pbourke_ | sdake: would you be interested in shipping images for the ostk cli clients? | 13:37 |
| sdake | pbourke_ I dont know - community project - people can do whatever they like :) | 13:37 |
| sdake | what is the use case? | 13:37 |
| pbourke_ | dockerise all the things | 13:37 |
| SamYaple | pbourke_: like a 'utility' type container? | 13:37 |
| sdake | ya a toolbox container would rock | 13:38 |
| pbourke_ | hmm didnt think about putting them all in one | 13:38 |
| sdake | still would rock ;) | 13:38 |
| SamYaple | one bad thing about toolbox conatiner is no network namespace access | 13:38 |
| SamYaple | would love to get that figured out | 13:38 |
| *** jtriley has joined #kolla | 13:38 | |
| SamYaple | sdake: did you ever send an email to the docker guy? | 13:38 |
| pbourke_ | SamYaple: are you against one container per tool? | 13:39 |
| sdake | i could neve rfind the bug to harass him about | 13:39 |
| sdake | if you can provide it, I can contact him | 13:39 |
| SamYaple | pbourke_: yea. cause youll likely want to access more than one tool :) | 13:39 |
| sdake | or the atch | 13:39 |
| sdake | or whatever | 13:39 |
| sdake | something to reference | 13:39 |
| SamYaple | sdake: you mean the patch i posted in the commit message? That you said you emailed him about like many months ago? | 13:39 |
| sdake | ya an AIO toolbox container would be higly useful I think | 13:40 |
| sdake | i talked toh im about it | 13:40 |
| *** achanda has quit IRC | 13:40 | |
| SamYaple | shouldn't you have that in a previous email then? | 13:40 |
| sdake | told him we needed something undone to get a fully fucntioanl system | 13:40 |
| pbourke_ | what about the whole entrypoint mechanism though | 13:40 |
| sdake | i guess, my inbox has 4k unread emails :) | 13:40 |
| sdake | pbourke_ ya that needs to be thought out | 13:40 |
| sdake | the worst part is I've onlybeen at cisco for 8 months :( | 13:41 |
| SamYaple | pbourke_: well if its a utility container it should always be running, no? | 13:41 |
| SamYaple | pbourke_: any utility container with all the tools should be started with sleep infinity | 13:41 |
| pbourke_ | SamYaple: hmm no I think we have different ideas on how this would work | 13:41 |
| SamYaple | i mean we arent going to spin up a container each time we call the tool i dont think | 13:41 |
| sdake | typically toolbox containers spin up the container on each toolbox tool access | 13:42 |
| SamYaple | ugh thats really bad | 13:42 |
| SamYaple | i would not be ok with that | 13:42 |
| pbourke_ | its how docker works | 13:42 |
| sdake | that is what everyone does with toolbox containers | 13:42 |
| pbourke_ | or I should say how its typically used | 13:42 |
| sdake | (spin up each invocation) | 13:42 |
| pbourke_ | docker run keystoneclient token-get | 13:42 |
| SamYaple | not when they need to access a socket or a volume from another container | 13:42 |
| sdake | not saying its right, I really dont have enough experience with it to know | 13:43 |
| pbourke_ | most of them are standard http | 13:43 |
| pbourke_ | things like ovs are more involved yes | 13:43 |
| SamYaple | yea im not cool with a container per | 13:43 |
| pbourke_ | im not cool with an all in one :) | 13:43 |
| SamYaple | but if you want you can do that right now i guess | 13:43 |
| SamYaple | thats cool we dont need to have either | 13:43 |
| sdake | typically toolbox containes dont access sockets or volumes | 13:44 |
| SamYaple | rabbitmqctl is only going to talk over a socket | 13:44 |
| sdake | i prefer AIO toolbox containers | 13:44 |
| SamYaple | same with ovs | 13:44 |
| sdake | it puts all the things in one place | 13:44 |
| sdake | i dont see the value of separate toolbox containers per client | 13:44 |
| sdake | pbourke_ could you expalin your thinking there? | 13:44 |
| SamYaple | given that all of this already exists in seperate images, i dont think we need anything else to achieve what you want pbourke_ | 13:44 |
| pbourke_ | SamYaple: wow didnt even think of that :o | 13:45 |
| SamYaple | honestly pbourke_ all the utils exist in openstack-base, so you may just be able to call that | 13:45 |
| sdake | typically toolbox containers have some well thought out entrypoint rather then only binaries in them | 13:45 |
| SamYaple | pass your auth strings in as environmental variables | 13:45 |
| SamYaple | just fyi, what i have right now guys is alias commands that execute the appropriate tools in a container | 13:46 |
| SamYaple | so my box has `rabbtimqctl` alias'd to `docker exec -it rabbitmq rabbitmqctl` | 13:46 |
| SamYaple | in case that intrests anyone. works well for me | 13:47 |
| sdake | ya i'm not pro enough to write aliases, i just type the shit out each time :) | 13:47 |
| SamYaple | considering a single toolbox rung command would look like `docker run --rm -it 192.168.31.12:8082/kollaglue/ubuntu-source-toolbox command` i think thats way to much work to type out | 13:48 |
| pbourke_ | SamYaple: good idea | 13:48 |
| *** jpeeler has joined #kolla | 13:49 | |
| *** jpeeler has joined #kolla | 13:49 | |
| sdake | one ting that could be done is a kolla-toolbox command which calls the various operations in one container for example | 13:50 |
| sdake | kolla-toolbox endpoint list | 13:50 |
| sdake | would print out the endpoints | 13:50 |
| sdake | kolla-toolbox rabbitmqctl would do whtever rabbitmqctl stuff was needed | 13:51 |
| sdake | but from a container runtime rather then a local install of the tools | 13:51 |
| sdake | this is typically how peopel implement toolbox containers | 13:51 |
| sdake | or how i have seen them implemented | 13:51 |
| *** jruano has quit IRC | 13:52 | |
| sdake | red hat atomic has a pretty solid toolbox container implementation | 13:52 |
| *** absubram has joined #kolla | 13:57 | |
| openstackgerrit | sean mooney proposed openstack/kolla: Regression: git is required to build from source on ubuntu https://review.openstack.org/224288 | 13:58 |
| [Leeloo | Hello | 13:59 |
| [Leeloo | Hello, yesterday I finaly succeffuly install kolla (Fedora 21 Core). I could use Openstask for a moment, then can't log anymore. | 13:59 |
| [Leeloo | So I re-install my server with Debian8...build/push the images, but got these errors --> "https://paste.fedoraproject.org/268441/49817814/" | 13:59 |
| *** diogogmt has joined #kolla | 14:00 | |
| sdake | [Leeloo you would have to past ethe full og | 14:00 |
| sdake | clearly somethign is not working for you with teh build ;) | 14:01 |
| sdake | if it was yesterday and you were buildign the default ( centos binary) the rdo repositories were down | 14:01 |
| [Leeloo | sdake: could I add something to "tools/build.py" to log ? | 14:01 |
| [Leeloo | no today ubuntu/source | 14:01 |
| sdake | can't you past ethe full log? | 14:02 |
| sdake | it shoudl hve a aslew of data beyond what youpasted | 14:02 |
| sdake | what i would recommend iss the following | 14:02 |
| sdake | run the command script | 14:02 |
| sdake | run the build.py operation | 14:02 |
| sdake | type exit | 14:02 |
| sdake | this will save afile called "typescript" | 14:02 |
| sdake | paste the typescript file | 14:02 |
| [Leeloo | sdake: I do it... and give the log in 2/3 hours | 14:04 |
| sdake | cool I can help debug from that info | 14:04 |
| sdake | i wonder if your push is failing | 14:04 |
| sdake | or your build is failing | 14:04 |
| sdake | do you know? | 14:04 |
| [Leeloo | sdake: no :( | 14:06 |
| [Leeloo | Do you know why I can login to openstack then alter a click... and being logout..can't re-log ? Could only test openstak for 10 mins. No password change for the admin ... Try to remove the cookie for Horizon.. | 14:08 |
| sdake | was only horizon non-operational? | 14:08 |
| sdake | or was the entire thing bust | 14:08 |
| sdake | were you able to run openstack endpoint list? | 14:09 |
| sdake | it is possible horizon is broken in some way, although it works perfectl yfor me | 14:09 |
| sdake | note a patch hit the repo yesterday which broke kolla badly | 14:09 |
| sdake | and we are still working on fixing it | 14:09 |
| [Leeloo | sdake: i quite knew to openstack/kolla, I was loking only to try to repeat kolla/openstack install..not digging a lot for this pb...:( | 14:10 |
| pbourke_ | SamYaple: surely the rpm family will be left git-less with this change? https://review.openstack.org/#/c/224288/4/docker/kolla-ansible/Dockerfile.j2 | 14:11 |
| *** diogogmt has quit IRC | 14:11 | |
| *** dwalsh has joined #kolla | 14:12 | |
| openstackgerrit | Steven Dake proposed openstack/kolla: Add sudo for RHEL builds https://review.openstack.org/222882 | 14:13 |
| openstackgerrit | Steven Dake proposed openstack/kolla: Install which if its not installed in base https://review.openstack.org/222870 | 14:13 |
| openstackgerrit | Steven Dake proposed openstack/kolla: Significant refactor of base image https://review.openstack.org/222893 | 14:13 |
| pbourke_ | what is the distinction between base and openstack-base | 14:15 |
| pbourke_ | that change slipped by me | 14:15 |
| sdake | openstack-base is for openstack containers base is for non-openstack containers | 14:15 |
| sdake | openstack-base inherits from base | 14:15 |
| sdake | so for example, openstack-bae would contain a base set of requirements to make the total downlaod time smaller for a big deploy | 14:15 |
| pbourke_ | hmm ok. kolla-ansible is a non ostk service but seems to use openstack-base ? | 14:18 |
| sdake | I dont know why that particular container uses openstack-base | 14:18 |
| sdake | i pointed out general guidlines, not hard rules :) | 14:19 |
| pbourke_ | fair enough | 14:19 |
| *** inc0 has joined #kolla | 14:20 | |
| *** achanda has joined #kolla | 14:22 | |
| *** achanda has quit IRC | 14:25 | |
| *** inc0 has quit IRC | 14:35 | |
| SamYaple | pbourke_: isnt it installed in the base? | 14:36 |
| pbourke_ | SamYaple: yeah sorry it seems it is. just getting mixed up what goes where | 14:36 |
| SamYaple | pbourke_: oh yea kolla-ansible uses openstack-base because it has alot of the deps on that (namely all of the openstack clients for shade) | 14:37 |
| pbourke_ | SamYaple: I mean, why is git for centos etc in base, but for ubuntu its in openstack-base | 14:37 |
| SamYaple | pbourke_: i cant help there | 14:37 |
| SamYaple | i know its only needed in openstack-base ofr ubuntu | 14:37 |
| pbourke_ | SamYaple: you +2'ed the review though :) | 14:37 |
| SamYaple | i dont know centos structure anymore | 14:37 |
| SamYaple | thats because it works | 14:38 |
| sdake | it is only needed in openstack-base for centos too | 14:38 |
| sdake | nobody moved it :) | 14:38 |
| SamYaple | i mean we arent _removing_ git | 14:38 |
| SamYaple | we are just not installing it twice | 14:38 |
| SamYaple | where git is actually installed in the stack isn't really part of that review for centos because centos has had it in the base for a long time | 14:38 |
| sdake | there are likely to be minor inconsistencies between distros | 14:39 |
| sdake | this is not a terrible thing at this early stage | 14:39 |
| sdake | over time we should try to sort it out | 14:39 |
| sdake | but in the big scheme of priorities, it seems pretty low :) | 14:40 |
| SamYaple | feel free to look as ubuntu as the guide. i spent alot of time in yaodu getting all of the non-needed deps out of the way | 14:40 |
| sdake | define non-needed dep? | 14:40 |
| SamYaple | i guess not really a term you are used to in centos land :D | 14:41 |
| sdake | you mean deps for openstack services in the right place vs for the base? | 14:41 |
| SamYaple | with the debian based stuff you can not install the recommended packages | 14:41 |
| sdake | ya a dependency is by definition needed :) | 14:41 |
| SamYaple | centos you dont get that choice | 14:41 |
| SamYaple | only if you are going to be using that thing, which alot of times you arent | 14:42 |
| sdake | ya - so sounds like ubuntu has a way of idenitfying those optional dependencies | 14:42 |
| SamYaple | like zfs-fuse is a dependacy for some libvirt thing _if_ you want to use that libvirt feature | 14:42 |
| SamYaple | debian did packaging like this in the begining | 14:42 |
| sdake | ya glusterfs is a dep for libvirt | 14:42 |
| sdake | libvirt has like 300 dependencies in centos | 14:42 |
| sdake | its pure cracksauce | 14:43 |
| SamYaple | but i dont install that in the ubuntu/debian stuff cause we dont use it | 14:43 |
| SamYaple | thats my point | 14:43 |
| sdake | there is nothing we can do about that in centos | 14:43 |
| sdake | for ibnary | 14:43 |
| SamYaple | anyway what goes where, a good guide is what i have ubuntu setup as | 14:43 |
| *** dayou has quit IRC | 14:44 | |
| [Leeloo | sdake: here is the log --> http://paste.fedoraproject.org/268466/14425010/ | 14:44 |
| sdake | v2 ping attempt failed with error: Get https://5.39.71.21:4000/v2/: tls: oversized record received with length 20527 | 14:45 |
| sdake | somethign is busted with your registry configuration | 14:45 |
| sdake | no idea what | 14:45 |
| sdake | your build is succeeding, your push is failing | 14:45 |
| SamYaple | wierd indeed. would recommend v1 registry if at all possible | 14:45 |
| *** jasonamyers has joined #kolla | 14:46 | |
| SamYaple | we havent really tests https pushing with the buid script too much | 14:46 |
| SamYaple | ive done it with v1, but not v2 private | 14:46 |
| [Leeloo | so "docker run -d -p 4000:5000 --restart=always --name registry registry:1" instead of "docker run -d -p 4000:5000 --restart=always --name registry registry:2" ? | 14:47 |
| SamYaple | registry without a tag or the 'latest' tag defaults to the v1 registry (for a reason) | 14:47 |
| SamYaple | i dont know the explict v1 registy tag though | 14:48 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Refactor set_configs.py https://review.openstack.org/224557 | 14:48 |
| openstackgerrit | Martin André proposed openstack/kolla: Update documentation for Vagrant environment https://review.openstack.org/223420 | 14:48 |
| *** jruano has joined #kolla | 14:51 | |
| *** achanda has joined #kolla | 14:52 | |
| *** dims_ has quit IRC | 14:53 | |
| *** dims_ has joined #kolla | 14:55 | |
| *** unicell1 has joined #kolla | 15:00 | |
| *** unicell has quit IRC | 15:00 | |
| *** vbel has quit IRC | 15:03 | |
| *** vbel has joined #kolla | 15:04 | |
| [Leeloo | I got this error now "ERROR:__main__:unable to ping registry endpoint https://5.39.71.21:4000/v0/ | 15:06 |
| [Leeloo | v2 ping attempt failed with error: Get https://5.39.71.21:4000/v2/: EOF" | 15:06 |
| *** diogogmt has joined #kolla | 15:06 | |
| SamYaple | silly question here.. are you able to ping 5.39.71.21? | 15:06 |
| SamYaple | can you open port 4000 | 15:06 |
| SamYaple | ? | 15:06 |
| *** achanda has quit IRC | 15:09 | |
| *** diogogmt has quit IRC | 15:11 | |
| *** diogogmt has joined #kolla | 15:13 | |
| [Leeloo | SamYaple/ i am, it's the ip of the server | 15:20 |
| *** daneyon has joined #kolla | 15:30 | |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Nova https://review.openstack.org/218505 | 15:36 |
| *** mfalatic_ has joined #kolla | 15:36 | |
| *** mfalatic has quit IRC | 15:36 | |
| *** devlaps has joined #kolla | 15:41 | |
| *** jasonamyers has quit IRC | 15:45 | |
| *** dwalsh has quit IRC | 15:46 | |
| *** chandankumar has joined #kolla | 15:53 | |
| *** unicell1 has quit IRC | 16:10 | |
| *** mduggan has quit IRC | 16:12 | |
| *** mduggan has joined #kolla | 16:13 | |
| *** unicell has joined #kolla | 16:13 | |
| *** dwalsh has joined #kolla | 16:19 | |
| *** sdake_ has joined #kolla | 16:19 | |
| *** unicell1 has joined #kolla | 16:23 | |
| *** sdake has quit IRC | 16:23 | |
| *** unicell has quit IRC | 16:25 | |
| *** daneyon_ has joined #kolla | 16:25 | |
| *** daneyon has quit IRC | 16:28 | |
| *** unicell1 has quit IRC | 16:32 | |
| *** sdake has joined #kolla | 16:34 | |
| *** sdake_ has quit IRC | 16:38 | |
| *** unicell has joined #kolla | 16:52 | |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Add missing libvirt section to nova.conf https://review.openstack.org/224825 | 16:55 |
| sdake | pbourke_ i dont like that change | 16:55 |
| pbourke_ | sdake: how have you been doing it? | 16:56 |
| sdake | pbourke_ we want to keep templating out of the ansible playbooks as much as possible because it creates a dev/test cycle | 16:56 |
| sdake | i dont use virt, but the wy to do it is via an override in nova-compute.conf | 16:56 |
| sdake | just add [libvirt] | 16:56 |
| sdake | type=qemu | 16:56 |
| sdake | or wahtever the stuff is | 16:56 |
| pbourke_ | that works and is fine, but this is a required setting | 16:57 |
| pbourke_ | as in it wont work out of the box without it | 16:57 |
| pbourke_ | afaik | 16:57 |
| sdake | you mean it wont work in a hypervisor out of the box | 16:57 |
| sdake | if you use that setting, it will emulate virtualization which will give you shit performance on bare metal | 16:57 |
| sdake | i really dont want any templating unless absolutely necessary | 16:58 |
| sdake | the one example I can think of where a config option may need templating is MTU | 16:58 |
| sdake | this is because mtu needs all kinds of custom calculations all over the place | 16:59 |
| sdake | and expecting an operator to sort that out sounds sort of painful | 16:59 |
| vbel | why not to have kvm as default then | 16:59 |
| vbel | people at least will know where to switch it if they play with vbox | 16:59 |
| vbel | or whatever hypervisor | 17:00 |
| sdake | vbel it would be fine to specify kvm in the libvirt section in the j2 | 17:00 |
| pbourke_ | but no variable? | 17:00 |
| sdake | but not the playbooks | 17:00 |
| sdake | right no variable | 17:00 |
| sdake | here is why | 17:00 |
| sdake | if we start down that path, we will end up with hundreds of config options | 17:00 |
| sdake | or could potentially | 17:00 |
| pbourke_ | i see | 17:00 |
| sdake | you notice nothing is configurable like that - that is the rationale | 17:00 |
| pbourke_ | vbel: does hardcoding it to kvm work for you? | 17:00 |
| sdake | I think MTU may be an optional we want to add at some point | 17:01 |
| sdake | hardcoding to kvm will nto work in virtualized environments | 17:01 |
| vbel | i replace with qemu because I am on vbox | 17:01 |
| vbel | but it is good to have a place to switch it | 17:01 |
| *** dwalsh has quit IRC | 17:01 | |
| vbel | instead of digging internet for error messages from nova-compute | 17:01 |
| sdake | i hope folks understand why we really dont want to do that as a community | 17:02 |
| pbourke_ | yeah know I think it makes sense | 17:02 |
| sdake | i think it is something that should be documented as a common occurance type thing though :) | 17:02 |
| sdake | i am fine with all kinds of blaot in the documentation | 17:02 |
| sdake | because that publishes live without a dev cycle | 17:02 |
| vbel | qemu is good for starters :) so it should be documented at least | 17:02 |
| sdake | and we can reorganize it easily in the field | 17:02 |
| pbourke_ | I think its easier for someone to figure out, "oh, I can't use kvm, but here's the place to change it" rather than having to google and then conclude the template is flat out missing it | 17:03 |
| sdake | or rather the docs canbe reorganized without affecting the field | 17:03 |
| sdake | if you put it in an ovverride it will be set properly | 17:03 |
| sdake | but it wouldn't hurt to have kvm set as the default in the j2 | 17:03 |
| pbourke_ | ok cool | 17:03 |
| sdake | nova defaults to that anyway I think | 17:03 |
| vbel | yes, i would vote for hardcoded kvm in j2 which i can just replace with kvm,xen,.. | 17:04 |
| sdake | i'm open to a small minimal set of configurable key/value pairs | 17:04 |
| vbel | sorry, qemu,xen... | 17:04 |
| sdake | totally fine with that approach | 17:04 |
| sdake | as long as its not a variable ;) | 17:04 |
| vbel | ok | 17:05 |
| sdake | but the key value pairs that are configurable should offer some serious pain to not be configurable in the standard override way | 17:05 |
| openstackgerrit | Paul Bourke proposed openstack/kolla: Add missing libvirt section to nova.conf https://review.openstack.org/224825 | 17:05 |
| sdake | mtu is one example | 17:05 |
| pbourke_ | changed everything to vars in latest patch... | 17:05 |
| pbourke_ | :p | 17:05 |
| sdake | pbourke_ ar eyou say8ing nova doesnt work for you without that section? | 17:05 |
| sdake | or doesn't work in a virt environment? | 17:05 |
| sdake | nova works on my bare metal fine without it | 17:05 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Glance https://review.openstack.org/218501 | 17:05 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Keystone https://review.openstack.org/218497 | 17:05 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Rabbitmq https://review.openstack.org/218496 | 17:06 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Nova https://review.openstack.org/218505 | 17:06 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Neutron (thin) https://review.openstack.org/224829 | 17:06 |
| pbourke_ | sdake: vbel had to add it for qemu, others internally have been confused about it missing. Again we can improve docs but the above patch should help things I think | 17:06 |
| sdake | pbourke_ I'll write documentation right now to address it | 17:07 |
| sdake | give me 1 hour | 17:07 |
| sdake | but we can also merge your patch | 17:07 |
| sdake | it doesn't hurt to have that config option in there | 17:07 |
| sdake | and if nova ever changes defaults atleast we wont be suprrised by it :) | 17:07 |
| pbourke_ | sdake: ok thanks. I have no problem adding more docs during the m cycle either | 17:07 |
| pbourke_ | assuming there will be a drive in that area | 17:08 |
| sdake | i think its time to document how to override config options :) | 17:08 |
| pbourke_ | +1 | 17:08 |
| sdake | and i'm waiting on a build atm | 17:08 |
| sdake | so have nothing else to do but watch tv or intertoobs | 17:08 |
| sdake | pbourke_ master is busted atm | 17:10 |
| sdake | can you review https://review.openstack.org/#/c/224557/ | 17:10 |
| pbourke_ | i think SamYaple addressed all the important stuff so will approve | 17:11 |
| *** vbel has quit IRC | 17:11 | |
| sdake | cool | 17:11 |
| *** chandankumar has quit IRC | 17:20 | |
| *** mfalatic_ has quit IRC | 17:33 | |
| *** loth has joined #kolla | 17:35 | |
| openstackgerrit | Steven Dake proposed openstack/kolla: Significant refactor of base image https://review.openstack.org/222893 | 17:36 |
| openstackgerrit | Steven Dake proposed openstack/kolla: Implement a install_type and install_metatype https://review.openstack.org/224841 | 17:36 |
| *** achanda has joined #kolla | 17:39 | |
| openstackgerrit | Steven Dake proposed openstack/kolla: Add a customize deployment document https://review.openstack.org/224847 | 17:51 |
| *** dwalsh has joined #kolla | 17:57 | |
| *** chandankumar has joined #kolla | 18:12 | |
| *** sdake has quit IRC | 18:19 | |
| *** akwasnie has quit IRC | 18:19 | |
| *** athomas has quit IRC | 18:20 | |
| *** sdake has joined #kolla | 18:23 | |
| *** sdake has quit IRC | 18:31 | |
| *** sdake has joined #kolla | 18:33 | |
| *** huikang has joined #kolla | 18:58 | |
| *** banix has quit IRC | 18:58 | |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Cinder https://review.openstack.org/224872 | 18:58 |
| huikang | Hi, how can I disable authentication when run ./tool/build --push glance. I have already add my local registry Ip as the namespace | 19:00 |
| *** devlaps has quit IRC | 19:08 | |
| *** diogogmt has quit IRC | 19:25 | |
| *** diogogmt has joined #kolla | 19:26 | |
| *** chandankumar has quit IRC | 19:26 | |
| *** diogogmt has quit IRC | 19:27 | |
| *** banix has joined #kolla | 19:40 | |
| *** banix has quit IRC | 19:45 | |
| *** sdake has quit IRC | 19:46 | |
| *** dwalsh has quit IRC | 19:47 | |
| *** sdake has joined #kolla | 19:48 | |
| *** sdake has quit IRC | 19:57 | |
| *** sdake has joined #kolla | 20:00 | |
| *** diogogmt has joined #kolla | 20:03 | |
| SamYaple | is zuul borkeed? | 20:10 |
| openstackgerrit | Sam Yaple proposed openstack/kolla: Refactor set_configs.py https://review.openstack.org/224557 | 20:10 |
| SamYaple | ok guys i think zuul was busted earlier | 20:12 |
| SamYaple | its working now but lots of patches need rekicked | 20:12 |
| *** diogogmt has quit IRC | 20:12 | |
| *** diogogmt_ has joined #kolla | 20:12 | |
| *** smeisner is now known as smeisner|afk | 20:20 | |
| *** jerryz has joined #kolla | 20:29 | |
| jerryz | sdake: ping | 20:29 |
| jerryz | sdake: https://github.com/ansible/ansible/issues/10906 are you aware of this bug | 20:30 |
| jerryz | sdake: it breaks the mariadb boostrap task for me | 20:30 |
| SamYaple | jerryz: we don't have ansible with sudo working at this time | 20:32 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Cinder https://review.openstack.org/224872 | 20:33 |
| jerryz | SamYaple: you mean i need to connect to hosts as root ? | 20:33 |
| *** jtriley has quit IRC | 20:33 | |
| SamYaple | jerryz: correct | 20:33 |
| SamYaple | the push for non-root wont happen until Mitaka | 20:34 |
| SamYaple | thats also when container privelges will be dropping so they wont be running as root either | 20:34 |
| *** diogogmt_ has quit IRC | 20:35 | |
| jerryz | SamYaple: got it. | 20:35 |
| *** jtriley has joined #kolla | 20:42 | |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Cinder https://review.openstack.org/224872 | 20:44 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Cinder https://review.openstack.org/224872 | 20:48 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Heat https://review.openstack.org/224891 | 20:49 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Heat https://review.openstack.org/224891 | 20:50 |
| openstackgerrit | Ryan Hallisey proposed openstack/kolla: Replace config-external with a JSON file for Cinder https://review.openstack.org/224872 | 20:50 |
| SamYaple | rhallisey: you should really rebase off of this patchset https://review.openstack.org/#/c/224557/ | 20:51 |
| SamYaple | rhallisey: none of your permissions are working, I can assure you | 20:51 |
| rhallisey | right.. | 20:51 |
| SamYaple | i dont know what havoc that would wreak for you | 20:52 |
| SamYaple | i imagine testing is hard | 20:52 |
| rhallisey | oh you changed the name | 20:55 |
| rhallisey | command_options -> run_command | 20:55 |
| rhallisey | kk | 20:55 |
| SamYaple | oh right, yea you had it named command_options from when it was just going to be the ARGS part | 20:55 |
| SamYaple | sorry should have told you that | 20:55 |
| rhallisey | ya I need to stop adnd rebase | 20:56 |
| rhallisey | I didn't even notice you patched it | 20:56 |
| SamYaple | i patched up the 'source is directory' thing so it copies correctly too | 20:57 |
| SamYaple | and sets perms recursively | 20:57 |
| rhallisey | ok cool | 20:58 |
| rhallisey | wfm | 20:58 |
| *** dwalsh has joined #kolla | 20:59 | |
| *** jruano has quit IRC | 21:00 | |
| *** blakec has joined #kolla | 21:05 | |
| *** absubram has quit IRC | 21:16 | |
| *** gfidente has quit IRC | 21:17 | |
| *** dwalsh has quit IRC | 21:19 | |
| *** rhallisey has quit IRC | 21:19 | |
| *** dims_ has quit IRC | 21:23 | |
| *** diogogmt has joined #kolla | 21:27 | |
| *** loth has left #kolla | 21:27 | |
| *** jtriley has quit IRC | 21:41 | |
| *** dims_ has joined #kolla | 21:49 | |
| *** jruano has joined #kolla | 21:53 | |
| sdake | samyaple I believe pbourke had implemented non-root based sudo but you reverted his change this morning | 22:02 |
| sdake | damn so tired :( | 22:02 |
| *** achanda has quit IRC | 22:11 | |
| *** dims_ has quit IRC | 22:17 | |
| *** dims_ has joined #kolla | 22:20 | |
| *** mfalatic has joined #kolla | 22:25 | |
| *** achanda has joined #kolla | 22:26 | |
| *** diogogmt has quit IRC | 22:51 | |
| *** dims_ has quit IRC | 22:59 | |
| *** alisonh has quit IRC | 23:17 | |
| *** jruano has quit IRC | 23:33 | |
| *** alisonh has joined #kolla | 23:35 | |
| *** jruano has joined #kolla | 23:41 | |
| *** alisonh has quit IRC | 23:51 | |
| *** alisonh has joined #kolla | 23:54 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!