Friday, 2015-08-21

openstackgerritMerged stackforge/os-ansible-deployment: Removes trailing whitespace for bashate  https://review.openstack.org/20766300:17
*** alop has quit IRC00:45
*** shoutm has joined #openstack-ansible01:08
*** klindgren has quit IRC01:11
*** bapalm has quit IRC01:12
*** bapalm has joined #openstack-ansible01:14
*** davidself has joined #openstack-ansible01:20
*** sdake_ has joined #openstack-ansible01:20
*** bapalm has quit IRC01:21
*** sdake has quit IRC01:24
*** tlian has quit IRC01:24
*** bapalm has joined #openstack-ansible01:33
*** bapalm has quit IRC01:38
*** bapalm has joined #openstack-ansible01:44
*** bapalm has quit IRC01:51
*** daneyon_ has quit IRC01:53
*** bapalm has joined #openstack-ansible01:56
*** bapalm has quit IRC02:03
*** bapalm has joined #openstack-ansible02:10
*** sdake_ is now known as sdake02:12
*** bapalm has quit IRC02:15
*** bapalm has joined #openstack-ansible02:23
*** bapalm has quit IRC02:30
*** bapalm has joined #openstack-ansible02:33
*** bapalm has quit IRC02:40
*** bapalm has joined #openstack-ansible02:42
*** bapalm has quit IRC02:52
*** bapalm has joined #openstack-ansible02:55
*** bapalm has quit IRC03:04
*** bapalm has joined #openstack-ansible03:10
*** shoutm_ has joined #openstack-ansible03:11
*** shoutm has quit IRC03:12
*** shoutm has joined #openstack-ansible03:52
*** shoutm_ has quit IRC03:54
*** shoutm_ has joined #openstack-ansible03:55
*** shoutm has quit IRC03:57
*** shoutm has joined #openstack-ansible03:58
*** shoutm_ has quit IRC03:59
*** fawadkhaliq has joined #openstack-ansible04:02
*** JRobinson__ is now known as JRobinson__afk04:04
*** yaya has joined #openstack-ansible04:05
*** shoutm_ has joined #openstack-ansible04:39
*** bapalm has quit IRC04:39
*** JRobinson__afk has quit IRC04:39
*** shoutm has quit IRC04:41
*** bapalm has joined #openstack-ansible04:41
*** JRobinson__afk has joined #openstack-ansible04:44
*** bapalm has quit IRC04:46
*** bapalm has joined #openstack-ansible04:47
*** yaya has quit IRC04:51
*** fawadkhaliq has quit IRC04:51
*** fawadk has joined #openstack-ansible04:51
*** sdake_ has joined #openstack-ansible04:52
*** bapalm has quit IRC04:54
*** sdake has quit IRC04:56
*** bapalm has joined #openstack-ansible04:56
*** bapalm has quit IRC05:03
*** bapalm has joined #openstack-ansible05:07
*** JRobinson__afk is now known as JRobinson__05:07
*** bapalm has quit IRC05:16
*** bapalm has joined #openstack-ansible05:20
*** sdake_ has quit IRC05:23
*** CheKoLyN has quit IRC05:30
*** bapalm has quit IRC05:34
*** bapalm has joined #openstack-ansible05:35
*** bapalm has quit IRC05:44
openstackgerritMerged stackforge/os-ansible-deployment: Set iptables-persistent install execution to append to log  https://review.openstack.org/21417205:56
*** bapalm has joined #openstack-ansible05:56
*** javeriak has joined #openstack-ansible06:00
*** javeriak has quit IRC06:10
*** bapalm has quit IRC06:15
*** bapalm has joined #openstack-ansible06:16
*** bapalm has quit IRC06:20
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Ensure that nova and neutron provide consistent DHCP domain values  https://review.openstack.org/21312706:35
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Add support for additional nova.conf options  https://review.openstack.org/21049206:35
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Removes tab indents for bashate  https://review.openstack.org/20766406:37
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Fixes deprecated arithmetic expansion for bashate  https://review.openstack.org/20766506:37
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Fixes function declarations for bashate  https://review.openstack.org/20766606:39
evrardjpgood morning06:59
evrardjpfor teardown.sh patch, this won't work for non AIO installations07:00
*** shoutm has joined #openstack-ansible07:03
*** shoutm_ has quit IRC07:05
evrardjpor for hosts with a name that don't start with aio_07:07
matttevrardjp: which patch is this?07:12
evrardjphttps://review.openstack.org/#/c/215357/07:12
evrardjpI'm using teardown for non AIO myself07:13
evrardjpdon't know if it was really meant for it, because there are others AIO stuff inside the script07:13
evrardjpso if you think it should be used for all the cases, then it could be improved07:13
matttevrardjp: it's my understanding that it can be used outside of AIO07:14
evrardjpit was mine too07:14
stevellehuh, interesting. I never considered that it would be used outside of an07:16
stevelleAIO07:16
evrardjpmaybe use ansible hosts --list-hosts to get ansible hostname07:16
evrardjploop over the results to do your sed07:16
matttreview updated :)07:16
stevellewould want to grab containers instead of hosts07:16
matttyou can get a pretty reliable search w/ "_container-"07:17
evrardjpyup07:17
evrardjpI could have added the review, but I wanted to know the focus first ;)07:17
matttbut tbh, i'm not sure how much faith i'd put in that teardown script07:18
evrardjpit works ok07:18
evrardjpnot perfect but good enough07:18
matttyeah i've tried it once and it worked, but i got burned by something that it left behind too07:18
evrardjpssh known hosts?07:18
matttit was a libvirt configuration, i was testing ceph migrations on a cluster which previously ran juno but was now running kilo after a teardown and migrations were working fine07:19
stevelleI thought that the teardown was like the load balancer, something that existed just to let us do gate stuff07:19
stevellesuppose there isn't a reason that can't change07:19
matttmade the assumption that all worked, then found out later it only worked because of some cruft juno left behind07:19
evrardjpI think there is an easy way to avoid this07:20
evrardjpwhen creating an install playbook, there should be a uninstall playbook that does the opposite07:20
evrardjp;)07:20
evrardjpdoubles the work though07:20
matttyeah that'd be awesome, but a massive amount of work :(07:21
stevellesounds like an awesome blueprint :)07:21
evrardjpit would be useful for all the containers targetted by is_metal07:21
evrardjpif it's inside a container, there is no use, because you trash the container anyway07:22
evrardjpstevelle: if it's awesome, you'll assign it to yourself, right? ;)07:22
stevelleevrardjp: I'll definately review it ;)07:23
stevellehttp://i.imgur.com/dJCANhe.jpg  and with this I am checking out for the night.  Have a good Friday folks.07:26
evrardjp:)07:27
matttnight stevelle !07:27
evrardjpnight07:27
evrardjpis there a way to see why my apt seems locked?07:29
evrardjpI don't see anything in my apt logs07:29
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Set iptables-persistent install execution to append to log  https://review.openstack.org/21549507:29
*** shoutm has quit IRC07:29
evrardjpI just see this process runing07:29
evrardjp/usr/lib/apt/methods/https07:30
evrardjpfor like 15 minutes07:30
evrardjpdoing just an apt-get update07:30
*** javeriak has joined #openstack-ansible07:32
matttevrardjp: hrrmph, i think that happens after containers are built07:35
matttif it's actually running but just slow, you may want to override with a local belgium ubuntu mirror07:35
evrardjpI'd gladly do that, because we host one07:35
evrardjpbut there is no easy way right now, IIRC07:36
evrardjpbut I found the culprint07:37
evrardjpsometimes if you Ctrl-C the ansible process, the instruction is not sent to kill the command sent07:37
evrardjpso I'm stuck with apt-get that seem to run but are in a bad state07:37
evrardjpso I kill the processes and re-run the playbook and it works fine07:38
*** gparaskevas has joined #openstack-ansible07:38
evrardjpno one had that issue before?07:40
mattti've not seen that07:40
matttwhy are you ctrl-c'ing ?07:40
evrardjpgood question07:42
evrardjptoo bad I don't have a decent answer... I try to do things too fast sometimes07:43
mattt:)07:43
openstackgerritMatt Thompson proposed stackforge/os-ansible-deployment: Add nova_libvirt_live_migration_flag variable  https://review.openstack.org/21245207:46
*** shoutm has joined #openstack-ansible07:48
gparaskevasHey Goodmorning07:51
gparaskevasmattt: Hello one quick question, can i set a variable like haproxy_stats_password in the role/defaults/main.yml and then override it from user_variables.yml? is that how it works?07:53
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client  https://review.openstack.org/20793907:55
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client  https://review.openstack.org/20793907:55
matttgparaskevas: in theory yes!07:56
matttgparaskevas: that is the idea07:56
matttgparaskevas: my ansible isn't great, and evrardjp and i witnessed some variables that weren't over-writable07:56
matttgparaskevas: but generally speaking you can override most things07:57
gparaskevasmattt: ok so is there a better approach? I mean i have been asked by Jesse to default a variable in case its not overiden by the user.07:57
evrardjpdefaults/main.yml should definitely be overriden by user_variables07:58
gparaskevasok i will then give it a try07:58
evrardjpit's good that way07:58
gparaskevasthank you both07:58
evrardjpif odyssey4me isn't happy with the change he'll just say it in the review, don't worry anyway :p07:59
*** javeriak has quit IRC08:01
openstackgerritGeorge Paraskevas proposed stackforge/os-ansible-deployment: Enable HAProxy Stats Web UI  https://review.openstack.org/21501908:06
*** fawadk has quit IRC08:06
*** fawadkhaliq has joined #openstack-ansible08:06
*** shoutm_ has joined #openstack-ansible08:11
*** javeriak has joined #openstack-ansible08:12
*** shoutm has quit IRC08:13
*** fawadkhaliq has quit IRC08:21
*** JRobinson__ has quit IRC08:25
evrardjpdoes someone know what would be the best practice for the SSL certificates I'll put behind HAProxy? Having everywhere the same certificate seems weird08:26
evrardjpbut as we are using IPs for all the communications, it doesn't make sense to create good certificates for it08:26
evrardjpwould you give a fake name for every component, and generate ssl certificates for them?08:27
evrardjplike keystone.myclouddomain, horizon.myclouddomain08:28
evrardjpand the load balancer would have myclouddomain certificate08:28
odyssey4meevrardjp use an internal CA to sign IP-based certs08:32
evrardjpthis feels weird, but ok08:32
evrardjpI don't want to maintain an internal CA while I have an external valid CA, that's also what I meant08:33
evrardjpIP based certs doesn't mean anything to me08:33
odyssey4methe internal CA is not a lot to maintain08:33
odyssey4mebut otherwise you can use self-signed certs, but that's not worth much in real security terms08:33
evrardjpgiving a fake name to components would also work I guess08:34
odyssey4meI've been thinking that it may be a good idea to add a CA role for this purpose, but I'd rather do that as an optional role in a seperate repo.08:35
odyssey4meevrardjp yes, except that we're not really instrumented to use names internally at this point08:35
evrardjphttp://i1.kym-cdn.com/photos/images/newsfeed/000/353/279/e31.jpg08:35
odyssey4mehorizon and keystone (pending backport currently in review) are both instrumented to set self-signed certs08:36
odyssey4methe keystone role allows for haproxy to run ssl and also have keystone run ssl (so ssl front and back)08:36
evrardjpthat's what I'll do08:37
odyssey4mewe're missing the functionality to have user-provided certs for haproxy right now though08:37
evrardjpI plan to have an EV cert on frond (HAProxy)08:37
evrardjpand have like a standard cert on back08:37
evrardjpI'll add this08:37
evrardjpit's like a 10 min job right?08:37
odyssey4mecool :)08:38
odyssey4mewell, horizon provides a pattern to use - if you go with that it'll be pretty quick, yes08:38
evrardjpthe hardest part is always the variable name08:39
odyssey4meevrardjp the pattern is there, so it's easy :)08:39
openstackgerritMerged stackforge/os-ansible-deployment: Add support for additional nova.conf options  https://review.openstack.org/21049208:40
*** shoutm_ has quit IRC08:40
evrardjpok will check with horizion08:40
*** fawadkhaliq has joined #openstack-ansible08:43
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Removes trailing whitespace for bashate  https://review.openstack.org/21551708:43
*** javeriak has quit IRC08:59
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Ensure that nova and neutron provide consistent DHCP domain values  https://review.openstack.org/21312709:00
odyssey4memattt now that master gating is back on track, you may wish to revisit https://review.openstack.org/21317009:02
openstackgerritHugh Saunders proposed stackforge/os-ansible-deployment: Fixes deprecated arithmetic expansion for bashate  https://review.openstack.org/20766509:04
matttodyssey4me: revisit how ?09:05
openstackgerritJean-Philippe Evrard proposed stackforge/os-ansible-deployment: Adds the ability to provide user certificates to HAProxy  https://review.openstack.org/21552509:06
matttodyssey4me: oh this one, yeah i want to test this quickly then +209:06
openstackgerritJean-Philippe Evrard proposed stackforge/os-ansible-deployment: Adds the ability to provide user certificates to HAProxy  https://review.openstack.org/21552509:07
evrardjpsorry for the spam09:07
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Enable admin level on the haproxy stats socket  https://review.openstack.org/21411009:08
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Enable admin level on the haproxy stats socket  https://review.openstack.org/21411009:08
odyssey4meevrardjp patch spam is the best spam :p09:10
odyssey4meI adjusted your commit message to add a bug, after reporting the wishlist bug09:10
evrardjpoh ok09:10
odyssey4meevrardjp can you please do a wishlist bug for https://review.openstack.org/21552509:10
evrardjpwill do09:11
*** sdake has joined #openstack-ansible09:11
*** benwh4 has joined #openstack-ansible09:14
benwh4o/ good morning !09:15
evrardjpgood morning benwh409:16
benwh4I am confused reading the doc ... in http://openstack-ansible-deployment.readthedocs.org/en/stable/install-guide/configure-haproxy.html09:16
benwh4the IP@ used are on the subnet 172.29.23609:16
benwh4wheras on the http://openstack-ansible-deployment.readthedocs.org/en/stable/install-guide/configure-hostlist.html09:16
benwh4the IP@ used are on the subnet 10.240.009:17
benwh4and it is mentionned that the subnet to used must be the one defined by the br-mgmt interface09:17
openstackgerritJean-Philippe Evrard proposed stackforge/os-ansible-deployment: Adds the ability to provide user certificates to HAProxy  https://review.openstack.org/21552509:18
evrardjpper se, you must not really be on the same network. Because ansible will use the ip to connect09:20
evrardjphowever09:20
evrardjpit's really best if you have one network that can speak to the other components09:20
evrardjpso it's really best if your internal ip of your haproxy is in the mgmt network09:21
evrardjpthe external ip could used also for ansible09:21
evrardjpI use the internal ip myself09:21
evrardjphope it helps benwh409:21
benwh4so in my config file I must set up node infra1: using an ip on the subnet speaking to my br-mgmt network ?09:22
evrardjpyou mean for haproxy right?09:23
evrardjpit depends on the name you want to give to your machine ;)09:23
*** davidself has quit IRC09:23
benwh4it is just that in your doc in the example you used the host-mgmt network and not the container management network but the doc says to use br-mgmt net09:23
evrardjpSam-I-Am: I guess that one is for you ;)09:24
evrardjpyou could use both, for haproxy the most important things are: the haproxy destination machine must be reachable through ssh (whatever the ip you give) from the deployment machine and the haproxy destination machine must be able to reach the mgmt network09:26
evrardjpI guess it's discrepency in the doc09:26
evrardjpif possible, use mgmt network09:26
odyssey4meevrardjp yeah, we really need to revisit docs - perhaps in a sprint where we all focus on using them to do a deployment and find the gaps09:27
odyssey4mealthough I'm a little torn around what audience the docs should cater for09:27
evrardjpfresh eyes are always good to have, they read the docs differently than those who wrote them and re-read them many times09:28
evrardjpodyssey4me: what do you mean?09:28
odyssey4meOn one hand, I think it's better to do some basic docs which give someone a simple orientation - eg: here's where you find options that can be set, this is where you set them, etc09:28
evrardjpyup, that will bring deployers that will help you09:28
odyssey4methen describe a few example environments with detailed config examples09:28
evrardjplike me for example ;)09:28
odyssey4meand those config examples should be what we gate test, so that people also then understand the gate09:29
evrardjpoh09:29
evrardjpthat's smart09:29
evrardjpdo you plan to have multiple deployments to gate, to broaden test coverage?09:30
*** shoutm has joined #openstack-ansible09:30
odyssey4meI don't see the long hand descriptions in docs as being very sustainable. It's great and important for downstream deployers to have for their own teams, but for openstack-ansible I'm starting to think that we need to ask our direct consumers to understand ansible and to understand openstack already. We can't take on the task of teaching them both of those.09:31
odyssey4meevrardjp yep09:31
evrardjpodyssey4me: that's not really what you do with the docs09:32
evrardjpyou don't teach ansible/openstack09:32
evrardjpso that's good09:32
odyssey4methat said, having a base of long hand docs is great, but we need more volunteers to maintain them09:32
evrardjpthe more user-friendly the documentation is, the more ppl you'll have to look at it09:33
odyssey4meperhaps having the base of docs as mentioned above will be enough to give people a head start, and more volunteers will get involved in improving the long hand docs09:33
evrardjpThat's also why I like your docs for now, it will feed and replace our documentation system, except for real specifics09:33
evrardjp(at my company I mean)09:34
odyssey4meok, cool -happy to know that it's useful then :)09:34
evrardjpit's definitely useful09:34
odyssey4mebut clearly it's not good enough to help benwh4, so we need to figure out how to do that better09:34
*** javeriak has joined #openstack-ansible09:34
evrardjpit was my decision factor to go to osad instead of doing it myself09:34
evrardjpit's enough, he just found a bug ;)09:35
evrardjpthe best would be to convert him to fix this doc issue when he'll have a first deployment working :) (because I guess that's his focus right now)09:36
odyssey4mebenwh4 it'd be really helpful if you could register documentation bugs for us as you find them :)09:43
odyssey4meit'd be even better if you submit patches to fix them :)09:43
odyssey4mehmm, that reminds me - I don't think that we're doing regular republishing to readthedocs - we need to fix that09:44
odyssey4mewhen palendae is back online I'll find out how to do that\09:44
odyssey4meevrardjp re: the different gate check scenarios, that's what https://blueprints.launchpad.net/openstack-ansible/+spec/split-aio-gates is about09:44
odyssey4mehughsaunders and I will pull together a spec before the next meeting09:45
*** shoutm has quit IRC09:45
evrardjpwith all these environment variables, I have the feeling that I'm working with a devstack09:46
evrardjp:p09:47
hughsaundersevrardjp: devstack of the future09:47
odyssey4mehaha, only for gate checks09:47
openstackgerritHugh Saunders proposed stackforge/os-ansible-deployment: Use dict args for ceph_config slurp  https://review.openstack.org/21555109:53
*** javeriak_ has joined #openstack-ansible09:55
*** javeriak has quit IRC09:56
*** fawadkhaliq has quit IRC09:57
evrardjphughsaunders: Is the = sign deprecated somehow?09:57
evrardjpI prefer this style, it's more compact ;)09:57
hughsaundersevrardjp: nope, but dict args are preferred over string args09:57
odyssey4memore compact, like perl?09:58
evrardjpodyssey4me: that's an insult, right?09:58
evrardjp:p09:58
hughsaundersevrardjp: https://github.com/stackforge/os-ansible-deployment/blob/master/CONTRIBUTING.rst#style-guide10:00
odyssey4meevrardjp yeah, the dict args are much more readable10:03
evrardjpI'll try to remember that10:03
*** javeriak has joined #openstack-ansible10:06
*** javeriak_ has quit IRC10:07
benwh4odyssez4me ok I will take note and come back later with a patch10:10
benwh4BTW do you all work on aio on multinode env ?10:18
javeriakguys, question. Trying to configure glance to use CloudFiles, I dont understand what the "glance_swift_store_endpoint_type: publicURL" is for and what do I change it to10:28
*** fawadkhaliq has joined #openstack-ansible10:36
*** fawadkhaliq has quit IRC10:51
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client  https://review.openstack.org/20793910:57
openstackgerritJean-Philippe Evrard proposed stackforge/os-ansible-deployment: [WIP] Fixing haproxy-playbook fails when installing on multiple hosts  https://review.openstack.org/21557910:57
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Add galera_address to all playbooks installing galera_client  https://review.openstack.org/20793910:58
*** fawadkhaliq has joined #openstack-ansible10:59
odyssey4mebenwh4 the aio is for development and testing - we use it for that, but also do multinode environments for development and testing10:59
odyssey4mejaveriak I think that's for when you're using a swift deployment that's managed independently for a glance back-end11:00
odyssey4mewhen in doubt, leave it as-is11:00
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Fixes function declarations for bashate  https://review.openstack.org/20766611:03
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Enable Neutron lbaas-haproxy  https://review.openstack.org/20851911:03
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment-specs: Add spec for the Liberty cycle upgrade path  https://review.openstack.org/20771311:06
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment-specs: Add spec for the Liberty cycle upgrade path  https://review.openstack.org/20771311:07
openstackgerritMatt Thompson proposed stackforge/os-ansible-deployment: [WIP] Update how neutron migrations are handled  https://review.openstack.org/21558411:26
*** javeriak has quit IRC11:35
openstackgerritJean-Philippe Evrard proposed stackforge/os-ansible-deployment: Fixing haproxy-playbook fails when installing on multiple hosts  https://review.openstack.org/21557911:36
*** smallbig has joined #openstack-ansible11:48
svgHow can I mark a bug as 'Wishlist'? Or perhaps I don;t have the auth to do that?11:53
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Add default user role for Keystone & Horizon and tasks to create it  https://review.openstack.org/20219411:53
*** fawadkhaliq has quit IRC11:57
*** fawadkhaliq has joined #openstack-ansible11:57
odyssey4mesvg you should be able to11:58
openstackgerritSerge van Ginderachter proposed stackforge/os-ansible-deployment: Allow Horizon setup with external SSL termination  https://review.openstack.org/21464711:58
odyssey4mesvg you set it by changing 'importance'11:59
svgI don;t have access to that field it seems11:59
odyssey4meah ok, will set it now then12:00
svgI think I know why, I wanted to lower the number of emails I got from launchpad, and must have removed me from the team12:00
odyssey4meah, that makes sense12:01
evrardjpif I define something in conf.d and env.d (let's say keepalived for the conversation)12:02
evrardjpand the child items are already defined somewhere else in my inventory (for example haproxy)12:02
evrardjpwhen I look into openstack_inventory.json, I see that my haproxy host has its definition changed12:03
svgodyssey4me: is that the openstack-ansible-maintainers I should re-join?12:03
odyssey4mesvg the first substitution filter for horizon_secure_proxy_ssl_header_django doesn't make sense to me - when would it have a space in it?12:03
odyssey4mesvg hmm, I don't think so?12:03
evrardjpnow I have "physical_host_group": "keepalived_hosts", instead of "physical_host_group":"haproxy_hosts", is that a problem/isn't it supposed to merge?12:03
odyssey4mesvg https://launchpad.net/~openstack-ansible12:04
*** pradk has quit IRC12:04
svgk12:04
svgodyssey4me: I'm not totally sure if a space is allowed in tha apache header actually, that is when doing e.g. "HTTP: X-..."12:06
odyssey4meevrardjp personally I have no idea how the dynamic inventory fits together :/12:06
svgWith this update of course this won't happen anymore, so yes, superflous12:06
odyssey4mesvg so you're just making sure, just in case a deployer does something silly?12:07
svgThat was the idea, but in my first version the 'HTTP' was not hardcoded actually, which did not make sense, as this will always be a HTTP header, so I changed it to be hardcoded12:08
svgit still could be usefull if deployer sets horizon_secure_proxy_ssl_header_apache12:09
*** sigmavirus24_awa is now known as sigmavirus2412:11
openstackgerritMerged stackforge/os-ansible-deployment: Removes tab indents for bashate  https://review.openstack.org/20766412:17
openstackgerritMerged stackforge/os-ansible-deployment: Allow a log dir that isn't the "service_name"  https://review.openstack.org/21508112:17
*** sigmavirus24 is now known as sigmavirus24_awa12:18
*** prad_ has joined #openstack-ansible12:18
*** alejandrito has joined #openstack-ansible12:20
odyssey4meevrardjp do you want to backport  https://review.openstack.org/207664 and https://review.openstack.org/215081 to see how it's done?12:28
odyssey4methe second will be easier as a first12:29
mgariepyodyssey4me, hard to get hpcloud to build it now.12:29
odyssey4memgariepy we've fixed the issue for hpcloud, so the only issue now is that it's slow12:29
mgariepyok12:30
odyssey4memgariepy your patch has been approved, so it's doing the merge test now12:30
mgariepyyeah i saw that :)12:31
mgariepywill it be added to https://review.openstack.org/#/c/211151/ ? or should it be a different patch ?12:33
odyssey4memgariepy the backports will be done within the next week by the guys working on the ceph feature12:34
odyssey4methey'll chain the backports in dependent patches12:34
mgariepyok12:34
odyssey4meit's all in preparation for the 11.2.0 release, which is due on friday next week12:35
evrardjpodyssey4me: why not :)12:35
odyssey4meevrardjp ok, open https://review.openstack.org/21508112:36
odyssey4meyou should see a 'cherry pick' button12:36
odyssey4meclick on that12:36
evrardjpif it takes me more than 2 hours It's going to be hard12:36
evrardjpok12:36
evrardjpfor kilo right?12:36
openstackgerritJean-Philippe Evrard proposed stackforge/os-ansible-deployment: Allow a log dir that isn't the "service_name"  https://review.openstack.org/21560212:37
odyssey4meyou'll see that it includes the 'cherry picked from ...' at the bottom of the proposed message -this is the same as using 'git cherry-pick -x <sha>' on the cli12:37
odyssey4methere we go, you got it :)12:37
evrardjpok12:37
odyssey4meif gerrit can't cherry-pick cleanly, it'll error out12:37
evrardjpthere is no topic anymore, normal?12:37
odyssey4methat'll mean there's a dependent patch which hasn't merged into the selected branch yet - so you have to find it12:37
odyssey4meno topic when using this method is normal, and fine12:38
evrardjpok12:38
odyssey4methe topic is just to make it easier to see reviews that are covering a topci12:38
odyssey4me*topic12:38
odyssey4meif you try to cherry pick https://review.openstack.org/207664 you'll find that it'll error12:39
evrardjpI'll try if you don't mind12:39
odyssey4meit errors because it depends on https://review.openstack.org/215517 merging first12:39
odyssey4meso if you want to backport it, you have to do so from the cli12:40
evrardjpwait a sec12:40
odyssey4mewant to do that so that you see how to do dependent patches?12:40
evrardjpfirst question, is: except from a git blame on the files, how do you do it?12:40
evrardjpok12:41
odyssey4meso, on the cli from your clone of the repo12:41
odyssey4meexecute: git review -d <review number> for the review to depend on12:41
odyssey4me\ie in this case: git review -d 21551712:41
odyssey4methat will pull down and checkout to a copy of that review12:42
evrardjpok12:42
evrardjpcontinue please :)12:42
odyssey4methen open the review you want to cherry pick on top of that, ie https://review.openstack.org/20766412:43
evrardjpok12:43
odyssey4methere's a "download" drop-down which contains options to checkout/cherry-pick, etc12:43
evrardjpnow i click on cherry pick manually12:43
evrardjppaste in my terminal12:43
odyssey4meyep, but before you execute that12:43
odyssey4mepaste into terminal, but add '-x' just after cherry-pick12:44
odyssey4meie change git fetch https://review.openstack.org/stackforge/os-ansible-deployment refs/changes/64/207664/6 && git cherry-pick FETCH_HEAD to git fetch https://review.openstack.org/stackforge/os-ansible-deployment refs/changes/64/207664/6 && git cherry-pick -x FETCH_HEAD12:44
*** jwagner is now known as jwagner_away12:44
odyssey4methat'll ensure that the commit it's cherry-picked from is in the commit message12:44
odyssey4meassuming the cherry pick is successful (check the git log), you can execute 'git review' to push that up12:45
odyssey4meit'll ask you if you want to submit both patches, answer yes12:45
openstackgerritJean-Philippe Evrard proposed stackforge/os-ansible-deployment: Removes tab indents for bashate  https://review.openstack.org/21560912:45
odyssey4methere we go :)12:46
evrardjpI don't get it12:46
evrardjpI understand it's easy12:46
odyssey4methe second patch is now dependent on the first12:46
evrardjpjust to be sure12:47
odyssey4meopen up the review in your browser and you'll see that the dependence is the same as in your git log12:47
evrardjpI see that, and it's a cool interface to show all that12:48
evrardjpand it has updated the first commit cool12:48
evrardjpok quick question12:49
odyssey4mesure12:49
evrardjpis there an easy way to find the dependency?12:50
evrardjpexcept the git blaming of files?12:50
odyssey4menot to my knowledge12:50
evrardjpor you have to know the history of stuff12:51
evrardjpwhich makes sense12:51
odyssey4meI usually manually try the cherry pick, see where it conflicts, then look at the history of the conflicting file and try to find the missing cherry-pick12:51
evrardjpok12:51
odyssey4methis is why we prefer to have bugs logged for all commits, then we can track what's merged and what hasn't been backported yet12:52
evrardjphope I'm not the only one learning that, it's precious to see that12:52
evrardjpyup12:52
odyssey4meI'm still learning how to use git effectively, so it's likely you might learn a better way and can teach me. :)12:53
*** prad_ is now known as pradk12:53
evrardjpI was thinking to rebase at first12:53
evrardjpbut this isn't a good idea12:53
evrardjpI mean it won't get the job done12:54
odyssey4meif the patches depend, it's usually better to create the backports in the chain as they are, then you can rebase later12:54
odyssey4megerrit provides a simple rebase button, but it'll only show when a build of the review has completed once12:54
evrardjpok12:54
evrardjpnice piece of software this gerrit12:56
openstackgerritgit-harry proposed stackforge/os-ansible-deployment: WIP: Generic ini conf template  https://review.openstack.org/21207913:06
*** scarlisle has joined #openstack-ansible13:06
*** shoutm has joined #openstack-ansible13:08
openstackgerritMatt Thompson proposed stackforge/os-ansible-deployment: Add Ceph/RBD support to playbooks  https://review.openstack.org/21115113:15
openstackgerritMatt Thompson proposed stackforge/os-ansible-deployment: Allow nova key/partition/password injection  https://review.openstack.org/21562513:19
*** javeriak has joined #openstack-ansible13:23
odyssey4meevrardjp see my comment on https://review.openstack.org/215579 - I think there's a better way :)13:32
evrardjpI was planning to do that with keepalived ;)13:33
openstackgerritMatt Thompson proposed stackforge/os-ansible-deployment: Add support for additional nova.conf options  https://review.openstack.org/21563213:34
evrardjpI don't think that haproxy should restart if he gets the ip13:34
evrardjpif he doesn't get*13:34
evrardjpit should stay in the same state as before13:34
odyssey4meevrardjp but you can implement that in keepalived - ie you can make it stop/start as the vip migrates13:35
evrardjpyup13:35
evrardjpI wanted to make sure that, without a decent failover with keepalived (so switchover instead of failover), you won't turn 2 haproxy on, because they both will be doing checks on the backend13:37
evrardjpit's not bad, but it's not nice for the logs13:37
evrardjpI'll remove my change then13:37
openstackgerritMatt Thompson proposed stackforge/os-ansible-deployment: Set lxc.autodev=0 for cinder_volumes in containers  https://review.openstack.org/21563413:37
cloudnullmorning13:47
mattthowdy cloudnull13:48
cloudnullhows it mattt13:48
matttcloudnull: not bad, jet lagged at all?13:56
cloudnullnah. im tired but im staying on eu time because im headed back there tomorrow13:57
matttcloudnull: fair enough!13:57
openstackgerritMerged stackforge/os-ansible-deployment: Ensure that nova and neutron provide consistent DHCP domain values  https://review.openstack.org/21312713:57
matttcloudnull: thanks for your neutron-db-manage update yesterday, i put through https://review.openstack.org/#/c/215584/ but deliberately marked it WIP incase anyone wants to make any edits13:57
cloudnullthank you for raising the issue and fixerating it :)14:00
matttcloudnull: it's not quite fixed yet, it needs more work14:00
matttbut i wanted to put the review up so we can figure it out14:00
matttcloudnull: but liberty introduces some nice things w/ online/offline migrations, so we should def. leverage that14:01
cloudnull+1 i agree14:01
cloudnullim presently trying to catch up on all the work that has gone on. theres been a bunch of good changes. which is awsome!14:01
matttcloudnull: cool, maybe we can bash heads on this next week14:01
cloudnullfor sure14:01
cloudnulli look forwrad to it ;)14:02
openstackgerritMerged stackforge/os-ansible-deployment: Use slurp to get the content of the ceph.conf file  https://review.openstack.org/21317014:03
*** Mudpuppy has joined #openstack-ansible14:04
matttmgariepy: i'll get that backported14:04
cloudnullgreat work on all the things everyone ! seems like master/kilo have progressed nicely !14:04
palendaeodyssey4me: The readthedocs publishing is a manual process right now, I'll regenerate them14:05
mgariepymattt, ok perfect maybe include this as well https://review.openstack.org/#/c/215551/14:06
palendaeI'm not too interested in putting work into getting them automated since presumably we're going to be putting a lot of our content on openstack.org "soon"14:06
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Ensure that nova and neutron provide consistent DHCP domain values  https://review.openstack.org/21565014:06
*** KLevenstein has joined #openstack-ansible14:06
matttmgariepy: sure once it merges i'll get that done14:07
odyssey4mepalendae that process for publishing to openstack will probably only be for the rename, which is in a month or two14:07
openstackgerritMatt Thompson proposed stackforge/os-ansible-deployment: Use slurp to get the content of the ceph.conf file  https://review.openstack.org/21565214:10
*** woodard has joined #openstack-ansible14:14
svgWas there a final project name decided upon yet?14:17
palendaeopenstack-ansible14:18
palendaeInfra's just dealing with the stackforge 'retirement' decision14:18
palendaeWhich caught us at an unlucky time14:18
*** pradk has quit IRC14:21
*** prad_ has joined #openstack-ansible14:22
*** prad_ is now known as pradk14:23
odyssey4meyeah, we just missed the last scheduled rename slot by a day or two14:24
odyssey4meso now we wait for them to bundle up before it happens14:24
palendaeYep :(14:27
*** k_stev has joined #openstack-ansible14:30
*** pradk has quit IRC14:31
*** benwh4 has quit IRC14:32
*** fawadkhaliq has quit IRC14:41
*** gparaskevas has quit IRC15:02
*** fawadkhaliq has joined #openstack-ansible15:19
*** Mudpuppy has quit IRC15:20
*** bapalm has joined #openstack-ansible15:21
*** javeriak has quit IRC15:24
*** Mudpuppy has joined #openstack-ansible15:25
palendaeodyssey4me: I'll have to look at readthedocs's docs again. iirc, we needed to add some things to the git repos (openstack.org and github) that I couldn't do, but I'll double check15:26
odyssey4mepalendae ok, thanks - otherwise is there an option to just schedule a build once per day or something?15:27
palendaeI'll have to look. cloudnull, KLevenstein, and myself have access to the project on rtd15:27
palendaeI can easily add new maintainers if they make an account on readthedocs.org15:28
KLevensteinpalendae: I can help, though maybe not today15:28
palendaeI did trigger another build this morning15:28
palendaeWhich takes like, 2-3 minutes at the most15:28
palendaeThe builds tab looks like it's just a history15:28
palendaeodyssey4me: Ah, yeah - I'd either need access to stackforge's GitHub project or put commit hooks in the openstack one: https://read-the-docs.readthedocs.org/en/latest/webhooks.html15:36
odyssey4mepalendae alternatively, perhaps -infra can help just publish the docs to the right place on -infra :)15:40
palendaeYeah, asking15:40
palendaeIf they can do it, cool beans and no worries15:40
palendaeI'd rather do it all in reo15:40
palendaeI asked about rtd for now, hoping someone will come back and say, "No, you can't do RTD from here, use this instead."15:40
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Fixes loops for bashate  https://review.openstack.org/20766715:44
*** Mudpuppy has quit IRC15:51
*** tlian has joined #openstack-ansible15:52
palendaeAha, there's an existing infra job15:54
palendaebuuuut it uses the name of the project directly, and we don't own the os-ansible-deployment one >.<15:55
*** Mudpuppy has joined #openstack-ansible15:55
palendae...now we do :)15:56
*** javeriak has joined #openstack-ansible16:00
palendaeKLevenstein, cloudnull, I've recreated the project on readthedocs so the name is the full project name (git://git.openstack.org/stackforge/os-ansible-deployment), and proposed a change on infra to get it autopushed (https://review.openstack.org/215699)16:04
palendaeEr, http://os-ansible-deployment.readthedocs.org/en/latest/16:04
*** Mudpuppy has quit IRC16:05
KLevensteinpalendae: cool16:07
*** javeriak has quit IRC16:07
palendaeodyssey4me, cloudnull: Do either of you remember the title of the governance patch that got us moved into big tent?16:12
*** Mudpuppy has joined #openstack-ansible16:13
*** KLevenstein has quit IRC16:13
*** javeriak has joined #openstack-ansible16:18
odyssey4mepalendae https://review.openstack.org/19110516:20
palendaeThanks, found it.16:20
palendaehttps://review.openstack.org/#/c/215699/ is updated16:20
*** Mudpuppy has quit IRC16:25
*** Mudpuppy has joined #openstack-ansible16:26
*** daneyon has joined #openstack-ansible16:31
*** yaya has joined #openstack-ansible16:38
*** javeriak_ has joined #openstack-ansible16:42
openstackgerritJesse Pretorius proposed stackforge/os-ansible-deployment: Update documentation for multiple VLAN ranges  https://review.openstack.org/20959816:42
*** woodard has quit IRC16:42
*** javeriak has quit IRC16:44
*** shoutm has quit IRC17:08
*** tlian has quit IRC17:12
openstackgerritMerged stackforge/os-ansible-deployment: Removes trailing whitespace for bashate  https://review.openstack.org/21551717:19
openstackgerritMerged stackforge/os-ansible-deployment: Removes tab indents for bashate  https://review.openstack.org/21560917:19
openstackgerritMerged stackforge/os-ansible-deployment: Allow a log dir that isn't the "service_name"  https://review.openstack.org/21560217:19
*** javeriak has joined #openstack-ansible17:24
*** javeriak_ has quit IRC17:25
*** javeriak_ has joined #openstack-ansible17:25
*** KLevenstein has joined #openstack-ansible17:26
*** javeriak has quit IRC17:28
*** prad_ has joined #openstack-ansible17:31
*** alop has joined #openstack-ansible17:36
*** woodard has joined #openstack-ansible17:43
*** woodard has quit IRC17:48
*** prad_ is now known as pradk17:55
*** jaypipes is now known as leakypipes18:00
*** javeriak has joined #openstack-ansible18:10
*** javeriak_ has quit IRC18:12
meteorfoxodyssey4me: ping18:20
*** sdake has quit IRC18:46
*** mpmsimo has joined #openstack-ansible18:52
openstackgerritMerged stackforge/os-ansible-deployment: Add Ceph/RBD support to playbooks  https://review.openstack.org/21115118:56
*** javeriak has quit IRC18:57
*** javeriak has joined #openstack-ansible18:58
*** fawadkhaliq has quit IRC18:58
cloudnullo/ meteorfox18:59
cloudnulli think odyssey4me is out for the day .18:59
*** yaya has quit IRC19:02
*** sdake has joined #openstack-ansible19:03
meteorfoxcloudnull: ah, ok, thanks19:03
*** mpmsimo has quit IRC19:10
*** yaya has joined #openstack-ansible19:12
*** yaya has quit IRC19:17
*** woodard has joined #openstack-ansible19:29
*** b8sell has joined #openstack-ansible19:32
*** b8se11 has quit IRC19:33
*** woodard has quit IRC19:34
*** alejandrito has quit IRC19:48
*** fawadkhaliq has joined #openstack-ansible20:03
*** javeriak has quit IRC20:18
openstackgerritMerged stackforge/os-ansible-deployment: Use slurp to get the content of the ceph.conf file  https://review.openstack.org/21565220:22
openstackgerritMerged stackforge/os-ansible-deployment: Allow nova key/partition/password injection  https://review.openstack.org/21562520:23
*** woodard has joined #openstack-ansible20:30
*** woodard has quit IRC20:34
*** Mudpuppy has quit IRC20:43
lbragstado/ hello! I'm curious if there is any way to find/generate a sample inventory file, just to see how the architecture is laid out for osad?20:57
*** fawadkhaliq has quit IRC20:58
*** yaya has joined #openstack-ansible21:03
cloudnulllbragstad: the inventory file looks like https://gist.github.com/cloudnull/e586ef61f0edc8684cf621:05
lbragstadcloudnull: awesome, thanks!21:06
cloudnullanytime21:06
lbragstadcloudnull: I take it that is generated via the dynamic inventory21:07
cloudnullyes21:07
lbragstadand the vars that are available here - https://github.com/stackforge/os-ansible-deployment/tree/master/etc/openstack_deploy21:07
cloudnullcorrect21:07
lbragstadwow21:08
cloudnullwow?21:08
lbragstadcloudnull: s/wow/cool/21:10
cloudnullah :)21:10
lbragstadcloudnull: dynamic inventory is pretty neat, I haven't played with it much, but this seems like a pretty good example21:10
cloudnullit works but it should be better...21:11
cloudnullnot ansibles dyn inv but how we create it and maintain it.21:11
cloudnullbut itll get there in time21:11
lbragstadso, with the inventory you pasted me, that's all json and that can be passed via command line directory to ansible (-i)21:11
cloudnull`ansible-playbook` via stdin21:12
cloudnullie http://docs.ansible.com/ansible/developing_inventory.html21:12
*** KLevenstein has quit IRC21:14
lbragstadcloudnull: you guys don't have anyone that uses just straight up inventory files for this (because osad expects certain things that can't be detailed with regular inventory?)21:16
cloudnullnot that im aware of21:19
lbragstadok21:19
cloudnullit will work with stock inv but idk anyone who is using it21:19
lbragstadgotcha,21:19
*** yaya has quit IRC21:23
*** yaya has joined #openstack-ansible21:31
openstackgerritMerged stackforge/os-ansible-deployment: Add support for additional nova.conf options  https://review.openstack.org/21563221:36
openstackgerritMerged stackforge/os-ansible-deployment: Set lxc.autodev=0 for cinder_volumes in containers  https://review.openstack.org/21563421:36
openstackgerritMerged stackforge/os-ansible-deployment: Ensure that nova and neutron provide consistent DHCP domain values  https://review.openstack.org/21565021:36
openstackgerritMerged stackforge/os-ansible-deployment: Update the documented ceph user variables  https://review.openstack.org/20913021:37
openstackgerritMerged stackforge/os-ansible-deployment: Fixes deprecated arithmetic expansion for bashate  https://review.openstack.org/20766521:37
openstackgerritMerged stackforge/os-ansible-deployment: Fixes function declarations for bashate  https://review.openstack.org/20766621:37
*** yaya has quit IRC21:41
*** scarlisle has quit IRC22:21
openstackgerritMerged stackforge/os-ansible-deployment: Fixes loops for bashate  https://review.openstack.org/20766722:30
*** woodard has joined #openstack-ansible22:46
*** alejandrito has joined #openstack-ansible22:46
*** woodard has quit IRC22:50
*** daneyon has quit IRC23:34
*** mpmsimo has joined #openstack-ansible23:36
*** k_stev has quit IRC23:58

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!