*** sdake has joined #kolla | 00:01 | |
*** sdake_ has quit IRC | 00:04 | |
openstackgerrit | Martin André proposed stackforge/kolla: Add min env vars doc validation to pep8 gate https://review.openstack.org/197474 | 00:08 |
---|---|---|
openstackgerrit | Martin André proposed stackforge/kolla: Add validation script for min env vars documentation https://review.openstack.org/197473 | 00:08 |
openstackgerrit | Martin André proposed stackforge/kolla: Sort variables and remove duplicates for min env doc https://review.openstack.org/197586 | 00:08 |
openstackgerrit | Martin André proposed stackforge/kolla: Make tools/min-env.sh portable https://review.openstack.org/197806 | 00:08 |
mandre | sdake: can you approve this bp? https://blueprints.launchpad.net/kolla/+spec/add-min-env-vars-gate | 00:42 |
sdake | set assignee plz | 00:42 |
*** erkules has joined #kolla | 01:27 | |
*** erkules_ has quit IRC | 01:30 | |
SamYaple | vincent_vdk: yea were getting a bit funky with the include to support it being generic so we dont have to rewrite the exact same code repeatedly | 02:36 |
SamYaple | sdake: https://review.openstack.org/#/c/198393/ | 02:43 |
SamYaple | thats really a much higher priority than it seems | 02:43 |
SamYaple | config-internal is b0rked without that patch | 02:43 |
sdake | samyaple sup bro | 03:08 |
sdake | happy 4th! | 03:09 |
SamYaple | murica | 03:09 |
sdake | samyaple looking moment | 03:09 |
sdake | ugh I thought i saw that and made a comment and scored appropriately | 03:11 |
openstackgerrit | Merged stackforge/kolla: Fix CONFIG_STRATEGY var in bash scripts https://review.openstack.org/198393 | 03:12 |
sdake | either i didnt or comment wasn't addressed ;( | 03:12 |
sdake | oh well | 03:12 |
sdake | fixed now ;-) | 03:12 |
sdake | samyaple anything else needing attention? | 03:12 |
sdake | I'm having a rare moment of drinking at the house to celebrate the 4th ;) | 03:12 |
SamYaple | nothing requiring you to stop doing that | 03:13 |
sdake | how is your 4th treating you? | 03:14 |
SamYaple | loud explodig things with 2 babies and animals. clearly my favorite time of year | 03:14 |
*** sdake has quit IRC | 03:16 | |
*** sdake has joined #kolla | 03:17 | |
sdake | nothing like a war zone going on in my backyard in the DESERT in the middle of SUMMER | 03:22 |
sdake | hey guys, i've got a good idea, lets go light the neighrborhood on fire ;) | 03:22 |
*** dims_ has quit IRC | 03:40 | |
*** sdake_ has joined #kolla | 04:08 | |
*** sdake has quit IRC | 04:12 | |
*** sdake_ is now known as sdae | 04:20 | |
*** dims_ has joined #kolla | 04:41 | |
*** bmace_away is now known as bmace | 04:41 | |
*** sdae is now known as sdake | 04:45 | |
*** dims_ has quit IRC | 04:46 | |
openstackgerrit | Sam Yaple proposed stackforge/kolla: make merge_configs idempotent https://review.openstack.org/198519 | 07:05 |
openstackgerrit | Sam Yaple proposed stackforge/kolla: fix possible symlink attack with ansible https://review.openstack.org/198520 | 07:23 |
*** sdake has quit IRC | 08:21 | |
*** sdake has joined #kolla | 08:22 | |
*** sdake_ has joined #kolla | 08:30 | |
*** sdake has quit IRC | 08:34 | |
harmw | SamYaple: how does set_configs() not work in https://review.openstack.org/#/c/198091/4/docker/common/neutron/neutron-l3-agent/start.sh? | 08:59 |
SamYaple | harmw: oops! my apologies. i was checking neutron-server and linux-bridge (the ones I shouldnt be looking at ;) | 09:04 |
openstackgerrit | Merged stackforge/kolla: Add config-internal/config-external to Neutron containers https://review.openstack.org/198091 | 09:04 |
harmw | SamYaple: np :) | 09:04 |
SamYaple | I HAVE THE POWER | 09:04 |
harmw | and thanks | 09:04 |
harmw | yea, well, what else is new :P | 09:05 |
SamYaple | harmw: when oyu get a chance https://review.openstack.org/#/c/198519/ | 09:10 |
harmw | dude, it's sunday morning :P | 09:10 |
SamYaple | its always work time | 09:10 |
harmw | damn | 09:10 |
harmw | your right | 09:10 |
SamYaple | also this is my actual work time | 09:11 |
harmw | :) | 09:12 |
harmw | I had (well took) the last 4 days off | 09:12 |
SamYaple | with the patch above, ansible runs multiple times without reporting changes :D | 09:12 |
SamYaple | that means no changes happen when running repeatedly! | 09:12 |
harmw | I'll check out all your current ansible stuff this week, somewhere | 09:12 |
SamYaple | but change a file or a service and BAM itll recreate (and register a change) | 09:12 |
harmw | sounds efficient | 09:13 |
harmw | like it should be | 09:13 |
SamYaple | im big on making sure ansible doesnt report changes | 09:13 |
SamYaple | some other projects execute 1000 tasks and 100 of those always report change, whether it occured or not | 09:14 |
SamYaple | that makes running the playbooks scary | 09:14 |
harmw | yep | 09:14 |
harmw | ah its python, I should realy start doing more on that subject | 09:14 |
SamYaple | no changes makes you want to run it 2 or 3 times "just to make sure" | 09:14 |
harmw | it sucks not being able to judge what happens in: fake_dest = StringIO() | 09:16 |
SamYaple | magic | 09:16 |
harmw | yep, but I hate magic | 09:16 |
SamYaple | basically since i have no direct way to compare the config object directly to the file object, i make a fake in-memory only file (fake_dest) and write out the config object to that "file | 09:16 |
SamYaple | " | 09:16 |
harmw | ah, in memory | 09:17 |
SamYaple | then i compare fake_dest with the actual dest file | 09:17 |
harmw | I can see what happens anyway, it's nice | 09:17 |
harmw | yea | 09:17 |
vincent_vdk | What OS are you guys running on the host? | 09:22 |
SamYaple | i test and run mainly against ubuntu, then centos for verification | 09:22 |
harmw | fedora lately, but coreos as well | 09:23 |
SamYaple | vincent_vdk: to finish our conversation from before, here is context http://paste.openstack.org/show/344832/ | 09:23 |
vincent_vdk | cool | 09:23 |
SamYaple | what was your exact question again? | 09:23 |
SamYaple | because i am indluing ../../config.yml? | 09:24 |
vincent_vdk | yes | 09:24 |
SamYaple | yea. thats a bit wierd, I agree. the reason for it is to respect DRY. ansible has no way to share common playbooks atm, there is lots of discussion on it though | 09:25 |
*** pbourke has quit IRC | 09:26 | |
*** pbourke has joined #kolla | 09:26 | |
vincent_vdk | SamYaple: common playbooks should be roles | 09:26 |
SamYaple | basically without doing the out-of-directory include i take the LOC for each role from 50 to 250 | 09:26 |
SamYaple | vincent_vdk: no | 09:27 |
SamYaple | vincent_vdk: common playbooks that are USER BY each role | 09:27 |
SamYaple | USED* | 09:27 |
vincent_vdk | you could create a 'common' role and include it in all your playbooks | 09:28 |
SamYaple | nope | 09:28 |
vincent_vdk | like roles: | 09:28 |
SamYaple | that wont work | 09:28 |
vincent_vdk | - common | 09:29 |
vincent_vdk | - databes | 09:29 |
vincent_vdk | -... | 09:29 |
SamYaple | i understand, that wont work | 09:29 |
vincent_vdk | hmm | 09:29 |
SamYaple | basically it comes down to this, if we dont have some way to call a common playbook, then any time i start a container I have to have a block like https://github.com/stackforge/kolla/blob/master/ansible/roles/start.yml | 09:32 |
SamYaple | thats alot of options that are not needed to start the container since they are common options | 09:32 |
SamYaple | or dont need to be specified each time, they _are_ needed | 09:33 |
SamYaple | rather, I can abstract that away and call the start with only https://github.com/stackforge/kolla/blob/master/ansible/roles/keystone/tasks/start.yml | 09:33 |
SamYaple | since I start 4-10 containers per role, if i suplicated the first code, any change would touch ALOT of files | 09:34 |
SamYaple | when (not if) common playbooks can be called, this issue goes away | 09:34 |
SamYaple | instead of "include: ../../start.yml" we can do "include_common: start.yml" or something | 09:35 |
vincent_vdk | i see | 09:40 |
vincent_vdk | still need to figure out how it all works together though | 09:40 |
vincent_vdk | we are looking at deploying openstack to replace a kvm based "cloud" | 09:40 |
vincent_vdk | but i think containers are the way to deploy openstack so we are looking at things like kolla and OSAD | 09:41 |
vincent_vdk | i like kolla's Docker approach a bit better | 09:41 |
harmw | SamYaple: re: https://etherpad.openstack.org/p/kolla-standard-start, how far along are Galera and Rabbit regarding that bp? | 09:42 |
SamYaple | I do as well, but only because OSAD doesnt use flat containers, nor do they use containers for all the services. some services run on bare-metal | 09:42 |
SamYaple | harmw: it was commited before the BP started | 09:42 |
vincent_vdk | since everything is more atomic (container image has everything already) | 09:42 |
harmw | ok SamYaple, Ill mark them as completed | 09:42 |
vincent_vdk | and you can tag/version your images | 09:42 |
SamYaple | vincent_vdk: OSAD is great for bare-metal, there really is no other choice | 09:42 |
vincent_vdk | docker on baremetal is the same | 09:43 |
SamYaple | vincent_vdk: yea that is the idea! I have also made these playbooks super flexible when it comes to that. so you can be running the "official" kolla images for all services, and then decide to build your own keystone image. then just deploy that keystone image, not all the services | 09:43 |
SamYaple | vincent_vdk: no i mean the compute services dont run in any containers at all for OSAD | 09:44 |
vincent_vdk | ah yes | 09:44 |
vincent_vdk | i've seen that | 09:44 |
SamYaple | OSAD on bare-metal without any containers, there isn't another option that works nearly as well | 09:44 |
vincent_vdk | found it a bit weird indeed | 09:44 |
SamYaple | for containers, i do like Docker more than LXC for nothing more than the flexibility and existing infrastructure | 09:45 |
vincent_vdk | advantage imho is tags and the images containing everything already | 09:45 |
SamYaple | yep | 09:46 |
vincent_vdk | basically they're packages | 09:46 |
SamYaple | containers as containers were meant to exist | 09:46 |
vincent_vdk | just the hypervisor par in kolla is something i need to check | 09:46 |
vincent_vdk | part* | 09:46 |
SamYaple | As I am aware, it is slow if you don't use a backend like ceph | 09:46 |
harmw | coolsvap|away: how's your progress on https://etherpad.openstack.org/p/kolla-standard-start, specifically Cinder and Horizon? | 09:47 |
SamYaple | in my container project that I am merging into Kolla i used ceph as a backend for everything and it was as quick as running openstack on baremetal | 09:47 |
vincent_vdk | maybe you could use ansible to deploy baremetal hosts | 09:47 |
vincent_vdk | that is nice | 09:47 |
SamYaple | vincent_vdk: thats what OSAD is for! | 09:47 |
vincent_vdk | i mean it might be an idea to have an option to just deploy the hypervisors on metal | 09:52 |
vincent_vdk | all other things can live in Docker containers | 09:52 |
SamYaple | vincent_vdk: that won't happen in kolla | 09:52 |
vincent_vdk | ok | 09:52 |
SamYaple | 1 there is no performance benefit to doing that, two we do _everything_ in containers, that is our thing | 09:52 |
SamYaple | we dont want to know or care about the host | 09:52 |
vincent_vdk | i understand | 09:53 |
SamYaple | OSAD doesnt deploy on baremetal for performance reasons, btw | 09:53 |
SamYaple | it does it because it can't use contaienrs for compute | 09:53 |
SamYaple | there is a bug with iscsi connections in a network namespace so they cant use cinder if compute is in a container (with its own network namespace) | 09:54 |
SamYaple | we dont have that issue in kolla because we dont use a seperate network namespace | 09:54 |
SamYaple | a seperate namespace adds a good bit of latency and we like to go fast | 09:54 |
vincent_vdk | i see | 09:54 |
vincent_vdk | i'm also very new to Openstack | 09:54 |
SamYaple | welcome to the show | 09:55 |
vincent_vdk | been learning about it for a month now | 09:55 |
SamYaple | we are hoping Kolla becomes a tool to help people learn it | 09:55 |
vincent_vdk | i'm starting to understand how it all works together | 09:55 |
SamYaple | setup for openstack is a big task for people who UNDERSTAND openstack | 09:55 |
SamYaple | we want to make it as simple as a package install | 09:56 |
vincent_vdk | kolla looks very interesting | 09:56 |
vincent_vdk | i hope we can go with a container besed deploy | 09:56 |
vincent_vdk | i'm currently working with Helion | 09:56 |
vincent_vdk | which uses TripleO | 09:56 |
vincent_vdk | i don't really like it | 09:57 |
SamYaple | im not sure if triple-o is using kolla containers in that case | 09:57 |
SamYaple | i know triple-o uses kolla containers somewhere | 09:57 |
vincent_vdk | no | 09:57 |
vincent_vdk | it's baremetal with ironic | 09:57 |
SamYaple | ah | 09:57 |
vincent_vdk | which requires a lot of machines to begin with | 09:57 |
SamYaple | yea | 09:58 |
vincent_vdk | that is not really an issue, but still | 09:58 |
SamYaple | probably 2-3 for the setup, then all baremetal? | 09:58 |
vincent_vdk | yes | 09:58 |
vincent_vdk | and 1 undercloud host | 09:58 |
vincent_vdk | and one to run the seed vm | 09:58 |
vincent_vdk | so at least 5 machines for services | 09:59 |
vincent_vdk | and they add 2 for swift | 09:59 |
vincent_vdk | have you guys looked at consul for service discovery | 09:59 |
vincent_vdk | can be used for load balancing too | 10:00 |
vincent_vdk | maybe even HA | 10:00 |
SamYaple | vincent_vdk: yea its been talked about but we are trying _not_ to add in external deps for things like that | 10:10 |
SamYaple | im not saying we wont, but we are trying ot avoid it | 10:10 |
vincent_vdk | i guess you'll need HA in sone way | 10:12 |
vincent_vdk | some* | 10:12 |
SamYaple | keepalived and haproxy vincent_vdk | 10:15 |
SamYaple | https://github.com/stackforge/kolla/blob/master/specs/high-availability.rst | 10:15 |
vincent_vdk | yeah, that's a common setup | 10:17 |
SamYaple | vincent_vdk: you can be my beta tester for ansible things | 10:28 |
SamYaple | right now database rabbit and keystone work, glance is in review | 10:29 |
vincent_vdk | ok | 10:30 |
vincent_vdk | i just need to setup a host somwhere to test on | 10:31 |
SamYaple | dont worry vincent_vdk, all the ansible bits arent finished yet | 10:39 |
vincent_vdk | i've been using Ansible for quite some time. I might add some things when i have some time left | 11:18 |
SamYaple | vincent_vdk: once all the initial components are added and an instance is bootable then we have a bunch of containers that need to be added | 11:22 |
SamYaple | If you are interested in patching some thing in, I would be more than happy to wlak your through the project | 11:23 |
SamYaple | so far I have been the only one doing ansible things | 11:23 |
vincent_vdk | that would be nice | 11:27 |
vincent_vdk | but i need to deploy it first | 11:27 |
vincent_vdk | it's the best way to learn | 11:28 |
openstackgerrit | Sam Yaple proposed stackforge/kolla: Have bootstrap script create the initial endpoint https://review.openstack.org/198533 | 11:51 |
*** bmace has quit IRC | 11:53 | |
*** bmace has joined #kolla | 12:06 | |
openstackgerrit | Sam Yaple proposed stackforge/kolla: Change restart policy for bootstrap container https://review.openstack.org/198534 | 12:14 |
*** dims_ has joined #kolla | 12:23 | |
openstackgerrit | Sam Yaple proposed stackforge/kolla: Ansible Glance support https://review.openstack.org/198494 | 12:34 |
SamYaple | sorry harmw i had you review an old glance patchset. my bad I updated and never pushed | 12:40 |
SamYaple | well i didnt _ask_ you to, so I suppose im in the clear | 12:41 |
*** dims_ has quit IRC | 12:41 | |
*** britthou_ has joined #kolla | 13:40 | |
*** britthouser has quit IRC | 13:43 | |
*** mandre has quit IRC | 13:48 | |
*** dims_ has joined #kolla | 14:00 | |
*** dims_ has quit IRC | 14:00 | |
*** mandre has joined #kolla | 14:06 | |
*** dims_ has joined #kolla | 14:13 | |
*** dims_ has quit IRC | 14:59 | |
openstackgerrit | Sam Yaple proposed stackforge/kolla: Have bootstrap script create the initial endpoint https://review.openstack.org/198533 | 15:36 |
SamYaple | sdake_: https://review.openstack.org/#/c/198520/ | 15:37 |
SamYaple | thats going to ensure a filed with proper ownership and perms. and it will ENSURE a file, that work for you? | 15:38 |
*** mandre has quit IRC | 15:54 | |
*** mandre has joined #kolla | 16:23 | |
*** sdake_ has quit IRC | 16:31 | |
*** nihilifer has joined #kolla | 16:50 | |
*** mandre has quit IRC | 16:50 | |
*** dims_ has joined #kolla | 17:01 | |
*** dims_ has quit IRC | 17:01 | |
*** dims_ has joined #kolla | 17:12 | |
*** dims_ has quit IRC | 17:12 | |
*** dims_ has joined #kolla | 17:13 | |
*** britthou_ has quit IRC | 17:25 | |
*** dims_ has quit IRC | 17:41 | |
*** mandre has joined #kolla | 17:42 | |
*** mandre has quit IRC | 17:49 | |
*** mandre has joined #kolla | 17:52 | |
*** britthouser has joined #kolla | 18:26 | |
*** sdake has joined #kolla | 18:31 | |
*** sdake_ has joined #kolla | 18:31 | |
*** sdake has quit IRC | 18:35 | |
sdake_ | morning | 18:42 |
vincent_vdk | hi | 18:42 |
harmw | sdake_: there is a container spec targetting Trove, you may assign that to me | 18:44 |
harmw | there's a time Ill get to it, so rather have it assigned to make sure I don't forget :) | 18:45 |
harmw | (and given the absense of Fang) | 18:45 |
sdake_ | you should be able to assign it to yourself i tink | 18:45 |
harmw | ok? couldn't find the button to do that though | 18:45 |
*** mandre has quit IRC | 18:49 | |
*** mandre has joined #kolla | 18:56 | |
*** mandre has quit IRC | 19:03 | |
openstackgerrit | Harm Weites proposed stackforge/kolla: Make current mariadb container use start.sh https://review.openstack.org/198567 | 19:16 |
*** britthouser has quit IRC | 19:18 | |
openstackgerrit | Harm Weites proposed stackforge/kolla: Make current mariadb container use start.sh https://review.openstack.org/198567 | 19:19 |
harmw | sdake_: instead of wooping things around I took a sidestep into moving mariadb to the start.sh model first | 19:21 |
harmw | when this is acked, I can quickly move it over to config-internal/external | 19:21 |
harmw | *quicker | 19:21 |
*** britthouser has joined #kolla | 19:27 | |
*** sdake_ has quit IRC | 19:28 | |
*** dims_ has joined #kolla | 19:40 | |
*** mandre has joined #kolla | 19:54 | |
*** dims_ has quit IRC | 20:02 | |
*** mandre has quit IRC | 20:14 | |
*** mandre has joined #kolla | 20:41 | |
*** sdake_ has joined #kolla | 20:51 | |
*** mandre has quit IRC | 20:56 | |
*** akscram has quit IRC | 21:04 | |
*** akscram has joined #kolla | 21:05 | |
*** sdake has joined #kolla | 21:21 | |
*** sdake_ has quit IRC | 21:25 | |
*** sdake has quit IRC | 21:35 | |
*** nihilifer has quit IRC | 22:26 | |
*** dims_ has joined #kolla | 22:28 | |
bmace | s | 23:43 |
*** sdake has joined #kolla | 23:53 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!