*** dims_ has joined #kolla | 00:02 | |
*** dimsum__ has quit IRC | 00:04 | |
*** shadower has quit IRC | 00:23 | |
*** shadower has joined #kolla | 00:23 | |
*** dims_ has quit IRC | 00:50 | |
*** dimsum__ has joined #kolla | 00:53 | |
*** achanda has joined #kolla | 01:25 | |
*** dimsum__ has quit IRC | 01:25 | |
*** dimsum__ has joined #kolla | 01:28 | |
*** erkules has joined #kolla | 01:54 | |
*** erkules_ has quit IRC | 01:56 | |
*** dimsum__ has quit IRC | 02:02 | |
openstackgerrit | Sam Yaple proposed stackforge/kolla: Adds push to build script https://review.openstack.org/208266 | 02:04 |
---|---|---|
*** jpeeler has joined #kolla | 02:49 | |
openstackgerrit | Sam Yaple proposed stackforge/kolla: Add log aggregation for Ansible into 'log_data' https://review.openstack.org/208275 | 03:08 |
*** diogogmt has quit IRC | 03:08 | |
*** ccrouch has left #kolla | 03:25 | |
*** coolsvap|away is now known as coolsvap | 03:40 | |
*** unicell1 has joined #kolla | 04:28 | |
*** unicell has quit IRC | 04:28 | |
*** daneyon has joined #kolla | 04:43 | |
*** daneyon has quit IRC | 04:43 | |
*** daneyon has joined #kolla | 04:44 | |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Remove service_hosts.sh https://review.openstack.org/208307 | 05:09 |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Remove service_hosts.sh https://review.openstack.org/208307 | 05:11 |
*** daneyon has quit IRC | 05:48 | |
*** daneyon has joined #kolla | 05:49 | |
*** daneyon has quit IRC | 05:58 | |
*** daneyon has joined #kolla | 06:02 | |
*** daneyon has quit IRC | 06:08 | |
*** daneyon has joined #kolla | 06:09 | |
*** daneyon_ has joined #kolla | 06:25 | |
*** daneyon has quit IRC | 06:28 | |
*** daneyon_ has quit IRC | 06:41 | |
*** achanda has quit IRC | 07:13 | |
*** achanda has joined #kolla | 07:17 | |
*** jmccarthy has quit IRC | 07:29 | |
*** pbourke has quit IRC | 07:29 | |
*** pbourke has joined #kolla | 07:33 | |
*** jmccarthy has joined #kolla | 07:33 | |
*** achanda has quit IRC | 07:36 | |
*** shadower has quit IRC | 08:06 | |
*** inc0 has joined #kolla | 08:21 | |
*** shadower has joined #kolla | 08:24 | |
inc0 | hi guys, happy sysadmin day | 08:29 |
inc0 | http://devopsreactions.tumblr.com/post/125543198764/happy-sysadmin-day | 08:29 |
SamYaple | its always sysadmin day | 08:33 |
SamYaple | inc0: i am writing a python wrapper for ansible | 08:33 |
SamYaple | its a PITA | 08:33 |
inc0 | SamYaple, ncurses stuff? | 08:34 |
inc0 | :P | 08:34 |
inc0 | btw...PITA? | 08:34 |
SamYaple | pain in the ass | 08:34 |
SamYaple | no inc0 python imports. I do things the right way | 08:34 |
*** akwasnie has joined #kolla | 08:34 | |
inc0 | you have problems with imports? | 08:35 |
inc0 | what kind of problems? | 08:35 |
SamYaple | nah its just im basically rewriting the ansible cli stuff | 08:35 |
inc0 | why would you do that? | 08:35 |
SamYaple | i have to to get our dependacies to be containerized | 08:36 |
inc0 | ahh...shade thing? | 08:36 |
SamYaple | not shade in particular | 08:36 |
SamYaple | just ansible in general | 08:36 |
inc0 | I mean one we've discussed | 08:36 |
SamYaple | i need to programatically pull out "changed" and "failed" | 08:36 |
inc0 | shade is an example | 08:36 |
SamYaple | i _could_ screen scrape for it | 08:37 |
SamYaple | i guess | 08:37 |
SamYaple | but thats horrible | 08:37 |
inc0 | no. | 08:37 |
inc0 | uhh...this all seems hacky:/ | 08:37 |
inc0 | and it won't work with other modules, which, unlike shade, has to be ran on host machine right? | 08:38 |
inc0 | have* | 08:38 |
inc0 | and run* | 08:38 |
SamYaple | itll work for all ansible modules when im done | 08:38 |
inc0 | damn english is hard. | 08:38 |
inc0 | idea is that you'll do a ansible pull from inside container right? | 08:38 |
SamYaple | it will appear as though you are calling ansible directly, only it will return the changed/failed correctly | 08:38 |
*** gfidente has joined #kolla | 08:39 | |
SamYaple | no | 08:39 |
*** gfidente has quit IRC | 08:39 | |
*** gfidente has joined #kolla | 08:39 | |
SamYaple | thats not going to work | 08:39 |
inc0 | then how will that work? | 08:39 |
SamYaple | the issue with that is I can't determine whether things changed | 08:39 |
SamYaple | so instead i hack up the cli and do a quick module that will accept ansible like input in a quick way | 08:40 |
SamYaple | since i know exactly wat options will be needed | 08:40 |
inc0 | ok...what problems do we solve now? It's about dependencies on target host? | 08:40 |
inc0 | so shade thing, which requires a lot of stuff we don't want? | 08:41 |
inc0 | or something else? | 08:41 |
SamYaple | shade is required for ansible modules | 08:41 |
SamYaple | python-mysqldb is required for mysql talking | 08:41 |
inc0 | I know, is that a problem you're solving? | 08:41 |
SamYaple | both pull in a bunch of crap and gcc | 08:41 |
SamYaple | yep | 08:41 |
inc0 | ok | 08:41 |
inc0 | and how do you solve it? | 08:42 |
inc0 | what's your idea? since ansible pull won't work as you say (btw, why?) | 08:42 |
SamYaple | 08:39 < SamYaple> the issue with that is I can't determine whether things changed | 08:42 |
inc0 | right, and you want to run ansible stuff inside container and scrap it's output right? | 08:43 |
SamYaple | well, i _dont_ want to screen scrape the output. | 08:43 |
inc0 | and btw, how do you run things inside container? you don't ssh to it do you? | 08:43 |
SamYaple | im using python and importing it so i can know for sure | 08:43 |
inc0 | ok, I know, just want to pinpoint problem itself | 08:44 |
SamYaple | i just spin up the "ansible" container and call the module i want | 08:44 |
SamYaple | no ssh | 08:44 |
*** Haomeng has quit IRC | 08:44 | |
*** Haomeng has joined #kolla | 08:45 | |
inc0 | as docker-exec call? | 08:45 |
SamYaple | no there is no "ansible" container running | 08:46 |
SamYaple | it gets geenrated and destroyed as needed | 08:46 |
inc0 | ah, ok | 08:46 |
inc0 | makes sense | 08:46 |
inc0 | but that means it'll always change right? | 08:46 |
SamYaple | yea i guess thats right. so well need it running always | 08:47 |
inc0 | one idea would be to add some file "bootstraped" | 08:47 |
inc0 | and check for this one | 08:47 |
inc0 | but that would require to keep contaier | 08:48 |
inc0 | or have it on some volume thing | 08:48 |
inc0 | if that would be host volume, you can use when statement to run it once | 08:48 |
inc0 | but, that would also break on multihost | 08:48 |
SamYaple | no that wont work inc0 | 08:49 |
SamYaple | we have to run it always | 08:49 |
SamYaple | i use some modules as healthchecks | 08:49 |
SamYaple | the mysql one for example is used to block until mysql is available | 08:49 |
inc0 | ok...why do you run healthcheck there? | 08:49 |
SamYaple | so it runs always | 08:49 |
SamYaple | because I have to block until mysql is available? | 08:50 |
inc0 | sometimes I have problem following you... you're solving problem of shade having a lot of dependencies (yeah, potentially other packages too, but let's focus on shade) | 08:50 |
inc0 | you run container and block it until mysql is available, then you run endpoint bootstraping and stuff using shade right? | 08:51 |
SamYaple | well its not a 'then' its an also | 08:51 |
SamYaple | i use it to bootstrap users for database, I also use it to do openstack keystone things | 08:51 |
inc0 | ok | 08:52 |
inc0 | let's call it a bootstrap container | 08:52 |
inc0 | so you run bootstrap contaner and wait till mysql is ready | 08:52 |
SamYaple | ok | 08:52 |
inc0 | then you bootstrap mysql users, then you wait for keystone | 08:52 |
inc0 | and bootstrap endpoints | 08:53 |
inc0 | is there anything else this container does? | 08:53 |
SamYaple | i dont think so, not at the moment at least | 08:53 |
inc0 | so why do you want it to run always? in case of new service being added? | 08:53 |
inc0 | for examle | 08:53 |
SamYaple | well i didnt _want_ it to run always, but as you pointed out starting and destroying a container means I awlys have to report a change which is not idempotent | 08:54 |
SamYaple | so it needs to be running always | 08:54 |
inc0 | it can be | 08:54 |
inc0 | you can use user in database as lock | 08:54 |
inc0 | like "if this user exists, I don't run at all" | 08:55 |
SamYaple | how would i know if the user exists? | 08:55 |
inc0 | SELECT user from mysql_users (can't remember exact table name); | 08:55 |
SamYaple | so i ouwld run that in the always running bootstrap container? | 08:56 |
inc0 | or just try to add it, then it will fail because username is unique | 08:56 |
SamYaple | what problem does that solve? | 08:56 |
inc0 | you'll get idempotency | 08:56 |
SamYaple | i have idemptonecy already | 08:56 |
*** erkules has quit IRC | 08:56 | |
*** erkules has joined #kolla | 08:56 | |
SamYaple | deleting and recreating the bootstrap container is the problem | 08:56 |
SamYaple | therefore it must always run | 08:56 |
inc0 | why is that a problem? | 08:57 |
SamYaple | because if i create a container and delete it, i am not being idemptoent | 08:57 |
inc0 | why? | 08:57 |
SamYaple | because those are changed on the system | 08:57 |
inc0 | this particular container has to be idempotent across whole cluster anyway | 08:57 |
inc0 | so you must assume it will run multiple times anyway | 08:58 |
SamYaple | i dont think you understand | 08:58 |
inc0 | I might not | 08:58 |
SamYaple | if i have to create and delete this container every time i run ansible then i am not idempotent | 08:58 |
inc0 | brb, unplugging laptop | 08:58 |
SamYaple | i should run ansible without changing anything if nothing needs changing | 08:59 |
*** inc0_ has joined #kolla | 09:00 | |
*** inc0_ has quit IRC | 09:02 | |
*** inc0_ has joined #kolla | 09:03 | |
*** inc0 has quit IRC | 09:03 | |
*** inc0_ is now known as inc0 | 09:03 | |
inc0 | ok, back | 09:03 |
inc0 | I hate vifi | 09:03 |
inc0 | wifi even | 09:03 |
inc0 | soo...ok, running container isn't idempotent, it will never be because its distributed | 09:04 |
SamYaple | whether its distributed or not it wont be idempotent | 09:04 |
SamYaple | to stop and recreate that is | 09:04 |
inc0 | but things this container does is idempotent | 09:04 |
SamYaple | correct | 09:05 |
inc0 | so you can run it every time | 09:05 |
inc0 | and nothnig will break | 09:06 |
SamYaple | no one said anything breaks | 09:06 |
SamYaple | but starting a container is a change to the system | 09:06 |
SamYaple | so it needs to stay running | 09:06 |
SamYaple | otherwise its not idempotent | 09:06 |
inc0 | uhh...its all just for ansible to show not-changed? | 09:07 |
SamYaple | if all i cared was for ansible ot show "not_changed" i would set changed_when to false | 09:08 |
SamYaple | i care about starting a container on a system since that is a true change | 09:08 |
SamYaple | that has potential to break things | 09:08 |
inc0 | how this will break things? | 09:08 |
SamYaple | answer me this, are you saying starting a contaienr should not count as a change? | 09:09 |
inc0 | if it's just bootstrap which tries to do stuff and dies | 09:10 |
inc0 | I'd argue it has to be run each time | 09:10 |
SamYaple | that doesnt answer my question | 09:10 |
inc0 | I'd consider it as normal script run | 09:10 |
SamYaple | that doesnt answer my question | 09:10 |
inc0 | therefore it's not a change per se | 09:10 |
SamYaple | it is a change | 09:10 |
inc0 | why did you ask me if you already have answer?:P | 09:11 |
SamYaple | its not running `ls` its starting a container | 09:11 |
SamYaple | so we could stop the conversation | 09:11 |
SamYaple | its a change and therefore must be prevented | 09:11 |
inc0 | from OS point of view there is little difference | 09:11 |
SamYaple | lies | 09:11 |
inc0 | it's a process | 09:11 |
inc0 | which lives for a moment and dies | 09:11 |
inc0 | even VM is a process | 09:12 |
SamYaple | its processes with cgroups and changes to the docker daemon | 09:12 |
inc0 | only usually continous | 09:12 |
SamYaple | and by your logic starting a vm isnt a change if it exists like a script | 09:12 |
inc0 | my point is | 09:12 |
inc0 | we *need* to run bootstraps | 09:12 |
inc0 | each time we change something | 09:13 |
inc0 | because you can add service dynamically | 09:13 |
inc0 | and if containers stays, it won't bootstrap it | 09:13 |
SamYaple | i dont think you understand the issues at hand, so ill do this and we can talk about the implementation | 09:13 |
SamYaple | i have idempotency right now | 09:13 |
inc0 | just do a patchset | 09:14 |
SamYaple | im just moving the dependacies into the container | 09:14 |
inc0 | and we'll discuss there | 09:14 |
inc0 | SamYaple, other thing, do you have Dockerfile templating bp submitted? | 09:18 |
SamYaple | i dont think so | 09:19 |
SamYaple | maybe | 09:20 |
SamYaple | ive been in and out of conciousness since the midcycle | 09:20 |
inc0 | it's that bad?:P | 09:20 |
SamYaple | cant normally sleep lots of stuff happening | 09:21 |
inc0 | wait till I'll get there, that will break you totally | 09:21 |
SamYaple | is that confirmed? | 09:22 |
*** diga has quit IRC | 09:34 | |
*** inc0_ has joined #kolla | 09:48 | |
*** inc0 has quit IRC | 09:50 | |
openstackgerrit | Swapnil Kulkarni (coolsvap) proposed stackforge/kolla: Temporary remove gnocchi from kolla-compose & functional tests https://review.openstack.org/208403 | 09:57 |
inc0_ | SamYaple, I'll try different approach - virtualenv in container | 10:06 |
inc0_ | this will fix other problems as well it it'll work | 10:06 |
SamYaple | inc0_: what problems will a virtualenv in a container solve? | 10:07 |
*** Ti-mo has quit IRC | 10:08 | |
*** walters has quit IRC | 10:08 | |
*** jpeeler has quit IRC | 10:08 | |
*** Ti-mo- has joined #kolla | 10:08 | |
*** jpeeler has joined #kolla | 10:09 | |
*** walters has joined #kolla | 10:10 | |
*** daneyon has joined #kolla | 10:18 | |
*** daneyon has quit IRC | 10:23 | |
*** dimsum__ has joined #kolla | 10:34 | |
SamYaple | inc0_: aha | 10:59 |
SamYaple | so i can parse json out | 10:59 |
SamYaple | i can have ansible wrap the output of ansible and catch the changed/failed | 11:00 |
*** coolsvap is now known as coolsvap|away | 11:06 | |
*** dims_ has joined #kolla | 11:18 | |
*** dimsum__ has quit IRC | 11:20 | |
*** rhallisey has joined #kolla | 11:42 | |
inc0_ | SamYaple, sorry, I was out | 11:51 |
SamYaple | np | 11:51 |
akwasnie | hi | 11:51 |
inc0_ | venv in container will allow you to run shade stuff in normal ansible flow | 11:51 |
SamYaple | doubtful | 11:51 |
inc0_ | it should be possible | 11:51 |
SamYaple | akwasnie: hello! | 11:51 |
SamYaple | inc0_: how will this help at all? | 11:52 |
inc0_ | not sure about mysql stuff, as its C-based, but with enough PYTHONPATH tinkering, it should be doable | 11:52 |
akwasnie | SamYaple have you done anything about templating Dockerfiles or can I start from scratch? | 11:52 |
SamYaple | but i mean, you still have the container | 11:52 |
SamYaple | akwasnie: i have... give me a second let me see if its around | 11:52 |
inc0_ | yeah, but this one can just run, will not change, and will be much simpler to work with | 11:53 |
inc0_ | it will be volume container essentialy | 11:53 |
akwasnie | ofc sorry | 11:53 |
SamYaple | inc0_: see i think we are talking the same language here. i have a essneitally "volume" container to do things | 11:54 |
SamYaple | akwasnie: this is what we agreed on during the midcyle, this is finished teplating for all of our keystone code | 11:54 |
SamYaple | http://paste.openstack.org/show/406297/ | 11:54 |
*** dims_ has quit IRC | 11:54 | |
SamYaple | to keep things simple, we would like to stick with if blocks | 11:55 |
inc0_ | yeah, but you'll have logic inside container, and I'll have it in normal ansible, that's basically only difference | 11:55 |
*** dimsum__ has joined #kolla | 11:55 | |
SamYaple | inc0_: how so? you cant execute normal ansible stuff in a container | 11:56 |
akwasnie | SamYample is there any bp for this? | 11:56 |
SamYaple | akwasnie: no | 11:56 |
inc0_ | you'll run container and run sh script inside | 11:56 |
inc0_ | to bootstrap stuff | 11:56 |
SamYaple | inc0_: wrong | 11:56 |
SamYaple | hold on | 11:56 |
inc0_ | ansible pull then? | 11:56 |
SamYaple | ill wip this thing | 11:56 |
inc0_ | please do this, will help a lot with discussion | 11:56 |
openstackgerrit | Sam Yaple proposed stackforge/kolla: Reduce dependencies on the dest nodes https://review.openstack.org/208451 | 11:57 |
SamYaple | btw i determined we _can_ do connection draining on haproxy | 11:57 |
SamYaple | so we can do service restarts with )% loss | 11:58 |
SamYaple | 0% | 11:58 |
*** coolsvap|away is now known as coolsvap | 11:58 | |
inc0_ | whats connection draining? | 11:59 |
SamYaple | i change the wieght on the server in the loadbalancer so no new connections go to it, then i wait until all existing connections close | 12:00 |
SamYaple | restart the service then change the weight back | 12:00 |
coolsvap | SamYaple, akwasnie I also want to work on templating work | 12:00 |
akwasnie | coolsvap: great | 12:00 |
SamYaple | coolsvap: ok. would you like to coordinate the process of templating like harmw did for the config-external/internat stuff? | 12:01 |
SamYaple | coolsvap: managing an etherpad and all? | 12:01 |
coolsvap | SamYaple, sure | 12:01 |
SamYaple | coolsvap: ok well you and akwasnie take the lead on this then | 12:01 |
coolsvap | let me create a etherpad | 12:01 |
SamYaple | I am unsure if a blueprint was started | 12:01 |
akwasnie | I will add bp in a minute | 12:01 |
SamYaple | wonderful | 12:01 |
SamYaple | go team! | 12:02 |
inc0_ | I thought as much, we can add this logic to upgrade flow | 12:02 |
SamYaple | so for the record, we will need to create a second toplevel folder named docker_templates | 12:02 |
inc0_ | but well, this will only turn off APIs | 12:02 |
SamYaple | so we will have docker and docker_templates | 12:02 |
SamYaple | we have t ostage the templates beside the existing code until its ready | 12:02 |
inc0_ | RPC calls will still balance over nodes | 12:02 |
coolsvap | SamYaple, yes, i have a bit of context from midcycle for this | 12:03 |
SamYaple | cool | 12:03 |
SamYaple | inc0_: i send rabbit through the LB so i can do this too | 12:03 |
inc0_ | but not consume through LV | 12:03 |
inc0_ | LB | 12:03 |
SamYaple | yea it does | 12:03 |
SamYaple | i mean i ensure no client is connecting directly through the node going down | 12:04 |
SamYaple | so it will get proxeied through the other nodes if the node thats going down is master | 12:04 |
SamYaple | _but_ the other nodes know how to safely retry | 12:04 |
inc0_ | yeah, I mean indirect connections | 12:05 |
SamYaple | like i said the other nodes know how to properly retry and the client never has to | 12:05 |
SamYaple | inderect is ok | 12:05 |
SamYaple | cutting off the one you are talking to is bad | 12:05 |
inc0_ | if they know how to retry:(.. | 12:05 |
SamYaple | well for openstack | 12:05 |
SamYaple | they do actually a good job of it | 12:06 |
inc0_ | cinder doesn't for example;) | 12:06 |
SamYaple | but now it doesnt have too! | 12:06 |
inc0_ | but yeah, connection draining is cool | 12:06 |
*** daneyon has joined #kolla | 12:07 | |
*** dimsum__ has quit IRC | 12:08 | |
coolsvap | akwasnie, https://etherpad.openstack.org/p/kolla-dockerfile-template | 12:12 |
*** daneyon has quit IRC | 12:12 | |
coolsvap | akwasnie, not sure if you got the earlier msg, here's the etherpad: https://etherpad.openstack.org/p/kolla-dockerfile-template | 12:15 |
*** dimsum__ has joined #kolla | 12:22 | |
akwasnie | coolsvap thanks | 12:24 |
akwasnie | here is the bp https://blueprints.launchpad.net/kolla/+spec/dockerfile-template | 12:24 |
SamYaple | synergy, we have achieved it | 12:24 |
SamYaple | inc0_: did you look at the review? | 12:25 |
*** dims_ has joined #kolla | 12:27 | |
*** dims__ has joined #kolla | 12:29 | |
*** dimsum__ has quit IRC | 12:29 | |
*** dimsum__ has joined #kolla | 12:32 | |
*** dims_ has quit IRC | 12:32 | |
*** dims__ has quit IRC | 12:34 | |
*** dwalsh has joined #kolla | 12:38 | |
inc0_ | mstachow, could you link 3o poc for rhallisey | 12:39 |
inc0_ | ? | 12:39 |
inc0_ | config external stuff | 12:39 |
mstachow | you mean link to github? | 12:39 |
inc0_ | SamYaple, yeah, I did, seems simple enough now | 12:40 |
inc0_ | yup | 12:40 |
rhallisey | mstachow, yea where are you working | 12:40 |
mstachow | rhallisey https://github.com/intelsdi-x/puppet_kolla.git | 12:41 |
mstachow | but this is only PoC | 12:41 |
rhallisey | cool thx | 12:41 |
mstachow | it's dirty as hell | 12:41 |
mstachow | >:) | 12:41 |
rhallisey | no problem | 12:41 |
*** dims_ has joined #kolla | 12:45 | |
*** dimsum__ has quit IRC | 12:46 | |
SamYaple | morning rhallisey | 12:52 |
rhallisey | SamYaple, morning | 12:52 |
*** dims_ has quit IRC | 12:53 | |
SamYaple | got the kolla dependacies situation solved for ansible | 12:53 |
SamYaple | I would bet yo ucan do the same thing or puppet | 12:53 |
rhallisey | oh nice! What's the solution | 12:53 |
rhallisey | I was trying some stuff out on Friday with no success | 12:53 |
SamYaple | its a wip but it does work. I containerize all of the deps and then have ansible istalled in that container | 12:54 |
SamYaple | I then call ansible in that container and parse the output of _that_ ansible for changed/failed | 12:54 |
*** dimsum__ has joined #kolla | 12:54 | |
SamYaple | so i maintain idempotency | 12:54 |
SamYaple | no scripts on my part | 12:54 |
SamYaple | the output is json too, so no screen scraping | 12:55 |
SamYaple | no passwords hitting the disk | 12:55 |
SamYaple | https://review.openstack.org/#/c/208451/ | 12:55 |
*** dimsum__ is now known as dims | 12:55 | |
rhallisey | I'll have to try it out | 13:00 |
SamYaple | yea im not sure how puppet works, but im positive we could adopt the same model | 13:00 |
SamYaple | if needed | 13:00 |
SamYaple | this reduces the deps down to only docker and docker-py o nthe dest hosts | 13:00 |
rhallisey | essentially what you doing is exaclty what we do with heat | 13:01 |
SamYaple | i made tripleo! w00t | 13:02 |
inc0_ | we're playing with kok now | 13:02 |
inc0_ | kolla on kolla | 13:02 |
rhallisey | ya I haven't don KoK yet | 13:02 |
SamYaple | kolla in dind on tripleo | 13:02 |
inc0_ | kolla-setup heat to deploy tripleo templates+puppets that deploy kolla-based overcloud | 13:03 |
rhallisey | inc0_, right | 13:03 |
SamYaple | I, for one, welcome our new Kolla overlord | 13:03 |
inc0_ | we need ironic container tho... | 13:03 |
inc0_ | someone have done that I think | 13:03 |
inc0_ | I mean was doing it | 13:04 |
inc0_ | jpeeler ... | 13:04 |
rhallisey | ya I was going to take a stab at it | 13:09 |
SamYaple | wont ironic require nova and then keystone? | 13:09 |
rhallisey | yea, what are you asking though? | 13:12 |
SamYaple | i dont really understand triple-o i guess | 13:13 |
SamYaple | i mean i get it on the highlevel :) | 13:13 |
SamYaple | but the initial openstack deploy i dont understand | 13:13 |
*** jtriley has joined #kolla | 13:34 | |
-openstackstatus- NOTICE: The Gerrit service on review.openstack.org has been restarted in an attempt to improve performance. | 13:39 | |
*** sdake has joined #kolla | 13:41 | |
sdake | morning | 13:41 |
inc0_ | morning sdake | 13:44 |
*** ccrouch has joined #kolla | 13:51 | |
*** daneyon has joined #kolla | 13:56 | |
coolsvap | morning sdake :) | 13:56 |
*** daneyon has quit IRC | 14:01 | |
*** ccrouch has quit IRC | 14:04 | |
sdake | samyaple around | 14:05 |
*** ccrouch has joined #kolla | 14:05 | |
*** ccrouch has left #kolla | 14:05 | |
*** ccrouch has joined #kolla | 14:11 | |
*** ccrouch has left #kolla | 14:11 | |
*** ccrouch1 has joined #kolla | 14:11 | |
openstackgerrit | Steven Dake proposed stackforge/kolla: Add Ansible support for Heat https://review.openstack.org/205528 | 14:15 |
openstackgerrit | Steven Dake proposed stackforge/kolla: Bootstrap the heat images properly https://review.openstack.org/208248 | 14:15 |
openstackgerrit | Steven Dake proposed stackforge/kolla: Include oslo_reports in installation of heat https://review.openstack.org/208226 | 14:15 |
*** sdake has quit IRC | 14:16 | |
*** sdake has joined #kolla | 14:24 | |
sdake | https://review.openstack.org/#/c/208226/ | 14:24 |
sdake | can i get quick reviews on this ^^ harmw samyaple rhallisey jpeeler | 14:24 |
sdake | pbourke? | 14:25 |
SamYaple | sdake: ping | 14:26 |
sdake | wound me samyaple | 14:26 |
*** unicell has joined #kolla | 14:26 | |
rhallisey | sdake, shouldn't that be in the base image ? | 14:26 |
rhallisey | cinder also needs oslo reports | 14:26 |
sdake | it is in the base image of heat | 14:26 |
rhallisey | no base centos | 14:26 |
sdake | i wasn't aware cinder needed that | 14:27 |
*** unicell1 has quit IRC | 14:27 | |
rhallisey | let me find the old bug.. | 14:27 |
sdake | i will edit in a moemnt after conversation with sam is done | 14:27 |
rhallisey | kk | 14:27 |
sdake | when i connect to my intrenal network i dont have irc access for some annoying reqason | 14:28 |
sdake | shoot samyaple whats up | 14:29 |
SamYaple | sdake: i dknt know | 14:30 |
SamYaple | you said my name | 14:30 |
sdake | oh wanted to let you know heat does work | 14:30 |
sdake | but whaty ousee is unfinished job | 14:30 |
SamYaple | i left reviews | 14:30 |
sdake | ok | 14:30 |
sdake | i left responses | 14:30 |
sdake | i'll brb | 14:31 |
SamYaple | its not actually bootstraping, you know that right | 14:31 |
sdake | absolutely know that | 14:31 |
SamYaple | ok cool | 14:31 |
sdake | i just wantd to push my work | 14:31 |
*** absubram has joined #kolla | 14:31 | |
sdake | inc case, i got hit by a bus or something | 14:31 |
rhallisey | sdake, I take that back, cinder needs oslo-policy not oslo-reports | 14:31 |
SamYaple | rhallisey: i think the issue with heat is it doesnt have athe package dep on it, even though it needs it | 14:32 |
SamYaple | so its a broken package, this doesnt prevent it from being used in cinder | 14:32 |
*** echoingumesh has joined #kolla | 14:33 | |
*** ccrouch1 has quit IRC | 14:36 | |
sdake | if folks could just approve that change, no other packae needs oslo reports | 14:36 |
sdake | i have tested a build last night at ike midnight | 14:36 |
*** ccrouch1 has joined #kolla | 14:38 | |
SamYaple | started on the logging container | 14:42 |
SamYaple | it appears we will need to use rsyslog after all :/ | 14:42 |
SamYaple | not everything can log to a file (keepalived and haproxy) | 14:42 |
*** jtriley has quit IRC | 14:43 | |
*** dwalsh has quit IRC | 14:44 | |
sdake | comon gate | 14:53 |
sdake | jesh if we hae to wait ike 4 hours toay for that patch to merge i'm going to strangle someone | 14:53 |
*** achanda has joined #kolla | 14:53 | |
SamYaple | i want my stackalytics stuff to merge | 14:53 |
sdake | dims may be able to help | 14:53 |
sdake | i think they do weekly maintenaceo n the wekends tho so it wont go live until the weekend | 14:54 |
SamYaple | yea i got it in late on friday :( | 14:54 |
SamYaple | 6 patches | 14:54 |
SamYaple | _six_ | 14:55 |
SamYaple | i shouldnt be an openstack dev | 14:55 |
sdake | heat going on 15 patches ;-) | 14:55 |
sdake | rhallisey are you going to have time tthis cycle to do ansible for cinder | 14:55 |
SamYaple | but i was just adding my name to the stackatylics thing | 14:55 |
sdake | after heat is working i guess i'll do magnum | 14:56 |
sdake | which wil default to n | 14:56 |
sdake | we need some cat to take on horizon as well | 14:57 |
sdake | i guess i'll make blueprints for all the remaining containers | 14:57 |
sdake | rather services | 14:57 |
* sdake needs to have a 1:1 with launchpad | 14:58 | |
*** barra204 has joined #kolla | 14:58 | |
sdake | jasonsb did you make a blueprint for logging | 14:58 |
SamYaple | sdake: he did | 14:58 |
SamYaple | its in discussion | 14:58 |
sdake | got a link? | 14:58 |
sdake | so I can put it in l3 | 14:58 |
SamYaple | blueprints kolla | 14:59 |
sdake | groan | 14:59 |
rhallisey | sdake, ya I'll give it a go | 14:59 |
sdake | rhallisey nice :) | 15:00 |
sdake | so I know I am not big on exceptions | 15:00 |
sdake | but openstack has a standard policy of allowing features in past the feature freeze deadline | 15:00 |
sdake | these are called FFEs | 15:00 |
sdake | I'm not usre if the TC or PTL decides or if its a combination | 15:01 |
sdake | (feature freeze exceptions) | 15:01 |
SamYaple | alright im going home, maybe for the night ill be off | 15:03 |
*** achanda has quit IRC | 15:04 | |
sdake | ok guys FFEs are decided by ptl | 15:06 |
sdake | unless we have the release_managed tag which we dont have | 15:06 |
sdake | jpeeler you around | 15:06 |
sdake | jpeeler i talked to most of the core team about the mission change, wanted toget your ack/nack | 15:07 |
* jpeeler is still catching up on email | 15:07 | |
jpeeler | was it posted to the list? i'll get to it soon if so | 15:07 |
sdake | please the deadline is today to veto | 15:09 |
*** barra204 has quit IRC | 15:09 | |
sdake | https://review.openstack.org/#/c/206789/ | 15:10 |
sdake | jpeeler how his married life ;) | 15:10 |
sdake | cna you look at the mission that review | 15:10 |
sdake | thats what the tc trimmed us down to | 15:10 |
sdake | it actually expands out scope not reduces it | 15:10 |
sdake | which is ok | 15:10 |
sdake | since we have ultimate decision factor on our scope by what we do | 15:10 |
jpeeler | married life is great, still adjusting though heh | 15:11 |
* jpeeler looks now | 15:11 | |
sdake | our midcycle rocked | 15:11 |
sdake | too bd you missed it | 15:11 |
*** jtriley has joined #kolla | 15:12 | |
*** dwalsh has joined #kolla | 15:13 | |
*** thumpba has joined #kolla | 15:14 | |
*** shakamunyi has joined #kolla | 15:14 | |
sdake | mandre you around | 15:15 |
*** shakamunyi is now known as barra204 | 15:16 | |
jpeeler | sdake: this review seems fine. is there a +1 vote i need to add anywhere? | 15:24 |
sdake | no just tell me your happy with the mission statement change | 15:25 |
*** barra204 has quit IRC | 15:26 | |
jpeeler | has it changed since i last looked at it? you're talking about this specifically - https://etherpad.openstack.org/p/kolla-manifesto | 15:26 |
sdake | that has changed yes | 15:30 |
sdake | its inthe review | 15:30 |
sdake | in the yaml file | 15:30 |
sdake | mission: >1045 | 15:31 |
sdake | To provide production-ready containers and deployment tools for operating1046 | 15:31 |
sdake | OpenStack clouds.1047 | 15:31 |
openstackgerrit | Merged stackforge/kolla: Include oslo_reports in installation of heat https://review.openstack.org/208226 | 15:35 |
SamYaple | sdake: techinically the deadline passed | 15:35 |
SamYaple | you said 00:00 Aug 3rd UTC | 15:35 |
SamYaple | its 15:36 | 15:35 |
jpeeler | ah ok, yeah it's fine anyway | 15:37 |
* jpeeler has forgotten how to use gerrit and forgot to read the actual file | 15:37 | |
*** SourabhP has joined #kolla | 15:38 | |
sdake | you wer eoverwhelmed by the rocking commit message we created I know ;-) | 15:39 |
SamYaple | jpeeler: does anyone really _know_ how to use gerrit? | 15:39 |
*** tfukushima has joined #kolla | 15:39 | |
jpeeler | true on both points | 15:40 |
*** banix has joined #kolla | 15:40 | |
*** barra204 has joined #kolla | 15:41 | |
*** daneyon has joined #kolla | 15:45 | |
*** blahRus has joined #kolla | 15:47 | |
*** daneyon has quit IRC | 15:50 | |
*** inc0_ has quit IRC | 15:50 | |
*** daneyon has joined #kolla | 15:51 | |
*** daneyon has quit IRC | 15:51 | |
*** daneyon has joined #kolla | 15:51 | |
*** openstackgerrit_ has joined #kolla | 15:53 | |
*** daneyon_ has joined #kolla | 15:54 | |
*** daneyon has quit IRC | 15:56 | |
*** ccrouch1 has quit IRC | 16:03 | |
*** ccrouch has joined #kolla | 16:04 | |
*** ccrouch has left #kolla | 16:04 | |
*** ccrouch has joined #kolla | 16:05 | |
*** unicell has quit IRC | 16:13 | |
*** sdake has quit IRC | 16:16 | |
*** rhallisey has quit IRC | 16:18 | |
*** sdake has joined #kolla | 16:21 | |
sdake | damn forgot to put a bug id in oslo-reports | 16:21 |
sdake | next time remind me plz :) | 16:21 |
*** diga has joined #kolla | 16:23 | |
*** diogogmt has joined #kolla | 16:32 | |
SamYaple | sdake: bug id | 16:36 |
*** unicell has joined #kolla | 16:37 | |
sdake | too late | 16:41 |
sdake | the release has been done | 16:41 |
sdake | enjoy ;) | 16:41 |
*** sdake has quit IRC | 16:43 | |
bmace | that is some real dropping the mic stuff there.. release done, sdake out! :) | 16:43 |
*** sdake has joined #kolla | 16:47 | |
sdake | I am loki of Asgard | 16:47 |
sdake | and I am burdened with glorious purpose ;) | 16:47 |
*** vbel has quit IRC | 16:50 | |
*** vbel has joined #kolla | 16:50 | |
*** rhallisey has joined #kolla | 16:51 | |
*** jasonsb has quit IRC | 16:54 | |
*** banix has quit IRC | 16:55 | |
*** coolsvap is now known as coolsvap|away | 16:58 | |
bmace | i see sdake, global control and it all started from being the ptl on an open source project. nobody will ever see it coming, like the spanish inquisition! | 17:00 |
sdake | yikes | 17:00 |
*** dwalsh has quit IRC | 17:00 | |
bmace | that was a monty python gag btw.. if you haven't watched it you should. | 17:01 |
*** akwasnie has quit IRC | 17:03 | |
*** dwalsh has joined #kolla | 17:03 | |
*** jtriley has quit IRC | 17:11 | |
*** diogogmt has quit IRC | 17:15 | |
*** unicell1 has joined #kolla | 17:16 | |
*** unicell has quit IRC | 17:16 | |
*** banix has joined #kolla | 17:18 | |
*** echoingumesh has quit IRC | 17:20 | |
*** echoingumesh has joined #kolla | 17:20 | |
*** echoingumesh has quit IRC | 17:21 | |
*** echoingumesh has joined #kolla | 17:22 | |
*** achanda_ has joined #kolla | 17:26 | |
*** SourabhP has quit IRC | 17:27 | |
openstackgerrit | Vikas Jain proposed stackforge/kolla: Fix incorrect config-external in horizon https://review.openstack.org/208570 | 17:29 |
*** echoingumesh has quit IRC | 17:33 | |
*** diogogmt has joined #kolla | 17:34 | |
*** jtriley has joined #kolla | 17:36 | |
sdake | samyaple ping re templating installs | 17:43 |
*** echoingumesh has joined #kolla | 17:45 | |
sdake | https://blueprints.launchpad.net/kolla/+spec/dockerfile-template | 17:46 |
*** ccrouch has quit IRC | 17:52 | |
*** ccrouch has joined #kolla | 17:56 | |
sdake | my tube amp and feickart record player are so fantastic | 18:00 |
sdake | just need a cabinet to put all this stuff on | 18:00 |
sdake | and i'll be "done" with my gear setup :) | 18:01 |
*** jasonsb has joined #kolla | 18:01 | |
sdake | my record player has ateast 100 hours of breakin left too so its just going to get better :) | 18:02 |
sdake | and some of my tubes have 50+ hours left of breakin | 18:02 |
*** openstackgerrit_ has quit IRC | 18:07 | |
*** achanda has joined #kolla | 18:14 | |
*** achanda_ has quit IRC | 18:16 | |
*** SourabhP has joined #kolla | 18:17 | |
*** Haomeng has quit IRC | 18:22 | |
daneyon_ | sdake we need to add the kollaglue/centos-rdo-haproxy to dockerhub | 18:27 |
daneyon_ | ^ repo | 18:27 |
sdake | I will push liberty2 images today | 18:28 |
daneyon_ | sdake the kollaglue/centos-rdo-haproxy repo need to be created on docker hub before you can push the L2 image | 18:28 |
daneyon_ | I do not have permissions to create the repo | 18:29 |
sdake | when we push the liberty2 images it will create the repo | 18:29 |
daneyon_ | ok, I'll mod my config to use local images until L2 is pushed | 18:36 |
*** tfukushima has quit IRC | 18:38 | |
*** bmace has quit IRC | 18:42 | |
*** gfidente is now known as gfidente|afk | 18:46 | |
*** bmace has joined #kolla | 18:55 | |
daneyon_ | sdake fyi: https://bugs.launchpad.net/kolla/+bug/1481054 | 19:00 |
openstack | Launchpad bug 1481054 in kolla "Can't Deploy Kolla Ansible All-in-One" [Undecided,New] | 19:00 |
sdake | i know what that is | 19:01 |
sdake | internal_network must be a free ip address on your network | 19:01 |
rhallisey | I just skip haproxy :P | 19:01 |
sdake | or internal_inetwork_address | 19:01 |
sdake | or whatever it is | 19:01 |
sdake | what happens is haproxy forwards allt he requests to the appropriate services | 19:01 |
sdake | from the global ip address | 19:02 |
sdake | that haproxy manages | 19:02 |
sdake | I know, intuitiive huh? :) | 19:02 |
sdake | daneyon_ feel free to close that out ;) | 19:04 |
sdake | or keep it open and we can use it for a doc fix against the /etc directory documentation | 19:05 |
daneyon_ | sdake let me try that. rhallisey I set the enable_haproxy to "no" and ansible is still processing the haproxy role | 19:10 |
daneyon_ | sdake maybe another option is to bind the services being proxied by haproxy to the loopback interface | 19:10 |
rhallisey | daneyon_, ya that didn't work for me either so I edited inventory/all-in-one and commented out haproxy | 19:11 |
rhallisey | that prevented it for being used | 19:11 |
sdake | you need haproxy | 19:11 |
sdake | just set it to a empty ip address | 19:11 |
rhallisey | k | 19:11 |
sdake | kolla will have undefined results without the haproxy container operational | 19:12 |
rhallisey | sdake, where do I set that | 19:14 |
sdake | rhallisey int he etc directory dude | 19:14 |
sdake | did you rtfm :) | 19:14 |
rhallisey | what's rtfm | 19:14 |
sdake | red the foxtrot manual | 19:15 |
sdake | read | 19:15 |
rhallisey | ok I am setting it.. I didn't know what it was for | 19:16 |
rhallisey | sorry still my 2nd time attempting to deploy.. | 19:17 |
sdake | there is a deployment manual in the docs directoroy | 19:21 |
sdake | granted its imperfect | 19:21 |
sdake | also set registry to "missing" | 19:21 |
sdake | or pull or whatever it is | 19:22 |
sdake | kolla_external_address: "openstack.example.com" | 19:22 |
sdake | kolla_internal_address: "10.10.10.254" | 19:22 |
sdake | kolla_external_address: "openstack.example.com" | 19:22 |
sdake | kolla_internal_address: "10.10.10.254" | 19:22 |
sdake | Isorry for double paste | 19:22 |
sdake | mistyle | 19:22 |
sdake | I set these to a free ip address in my network | 19:23 |
sdake | docker_pull_policy: "always" | 19:23 |
sdake | set this to missing | 19:23 |
sdake | network_interface: "eth0" | 19:23 |
sdake | neutron_external_interface: "eth1" | 19:23 |
sdake | set those to your eth0 and eth1 on your box | 19:23 |
*** sdake has quit IRC | 19:28 | |
*** bmace has quit IRC | 19:33 | |
rhallisey | I set internal and external to my host ip | 19:39 |
rhallisey | don't know if that's bad | 19:39 |
daneyon_ | sdake here is my latest problem trying to deploy: https://gist.github.com/danehans/95409886ced6b34f200b | 19:50 |
*** echoingumesh has quit IRC | 19:50 | |
*** echoingumesh has joined #kolla | 19:51 | |
*** banix has quit IRC | 19:54 | |
*** mfalatic has joined #kolla | 20:01 | |
*** sdake has joined #kolla | 20:01 | |
*** banix has joined #kolla | 20:02 | |
sdake | daneyon_ did that work for you | 20:05 |
*** sdake_ has joined #kolla | 20:10 | |
*** sdake has quit IRC | 20:13 | |
daneyon_ | sdake changing int and ext addresses to 127.0.0.1 worked, but I then ran into another issue: https://gist.github.com/danehans/95409886ced6b34f200b | 20:32 |
sdake_ | ccrouch around | 20:33 |
sdake_ | no, you want to change to an unused interfce | 20:33 |
sdake_ | 127.0.0.1 is used | 20:33 |
sdake_ | just pick an ip address out of your network that has nothing assigned to it | 20:33 |
sdake_ | for example, I have 192.168.1.148 | 20:33 |
sdake_ | nothign else will use .148 | 20:33 |
sdake_ | looking at gist moment | 20:33 |
sdake_ | daneyon_ I have never seen that trace before | 20:34 |
sdake_ | daneyon_ and i've been beating on this stuff daily for 3 weeks | 20:34 |
sdake_ | my first inclination would be to fixt the 127.0.01 thing, then to a cleanup-containers -fv | 20:35 |
sdake_ | then reboot | 20:35 |
sdake_ | which distro arey ou using? | 20:35 |
sdake_ | did you install the shade depepndencies? | 20:35 |
sdake_ | not installing the sahde dep may result in something like that, I'm not sure i've never tried running without it prior | 20:36 |
sdake_ | I use fedora 21, and AIO works like a champ | 20:36 |
sdake_ | I use ubuntu 14..04, AIO works like a champ | 20:37 |
sdake_ | but I had to install the shade deps first | 20:37 |
sdake_ | in some cases including the entire toolchain to build shade | 20:37 |
sdake_ | after you get through with my Q's if you coudl pot your globals.yml for me to have a look :) | 20:38 |
sdake_ | post | 20:38 |
sdake_ | FRUEDEN SLIP! | 20:38 |
sdake_ | note you should probbly reboot if you chnage the internal address | 20:39 |
sdake_ | it mya damage things, I am not sure | 20:39 |
sdake_ | after cleanup-containers -fv | 20:39 |
daneyon_ | sdake_ i am using F21 on the host and centos rdo containers | 20:40 |
sdake_ | cool that should work like a champ | 20:40 |
daneyon_ | sdake_ I installed all the pkgs/dep identified in the ansible deploy doc | 20:40 |
sdake_ | so pip show shade shows shade installed? | 20:41 |
daneyon_ | one min | 20:41 |
sdake_ | it is possible you have a newer version of shade then me or a differnet version of ansible which are broke | 20:41 |
sdake_ | oh, one key thing | 20:41 |
sdake_ | not in the documentation | 20:42 |
sdake_ | I run ansible as sudo | 20:42 |
sdake_ | running as sudo is mandatory unfortunately atm | 20:42 |
rhallisey | what version and ansible and shade do we need | 20:43 |
sdake_ | one moment | 20:43 |
sdake_ | i'm running shade 0.11.0 | 20:45 |
sdake_ | docker-py 1.2.3 (this is mandatory!) | 20:46 |
sdake_ | ansible-1.9.2-1.fc21.noarch | 20:46 |
sdake_ | docker-io-1.7.0-1.git21cb6eb.fc21.x86_64 | 20:47 |
sdake_ | newer versions of docker-py will not work | 20:47 |
sdake_ | like the 1.3 series will not operate - problems with the api versioning | 20:47 |
daneyon_ | sdake: https://etherpad.mozilla.org/y3JRiEFC0e | 20:49 |
daneyon_ | sdake i have been running ansible as root | 20:50 |
sdake_ | running ansible from pip | 20:50 |
sdake_ | ? | 20:50 |
sdake_ | I am using ansible from packaging | 20:50 |
daneyon_ | sdake i tried ansible last week from yum pkgs and had problems. SamYaple suggested i try the pip version | 20:50 |
sdake_ | i think that was bad advice tbh :) | 20:51 |
sdake_ | try pip remove ansible | 20:51 |
sdake_ | or maybe its uninstall | 20:51 |
daneyon_ | sdake are u running docker 1.7? | 20:51 |
sdake_ | ya docker 1.7 | 20:52 |
daneyon_ | ok | 20:52 |
sdake_ | I have a different build the nyou | 20:52 |
sdake_ | I found that build to be flakey | 20:52 |
sdake_ | I cant tell you where I got my fc21 version, it apepars missing on koji | 20:54 |
sdake_ | someone probably deleted it | 20:54 |
sdake_ | I think docker 1.6.2 should do the job | 20:55 |
sdake_ | (stock yum upgrade version) | 20:55 |
daneyon_ | but docker seems to be working fine outside of ansible deploy | 20:55 |
sdake_ | ya its probably not docker | 20:55 |
sdake_ | my guess is it s your pip installed docker or your 127.0.0.1 address | 20:56 |
sdake_ | I dont' even know what that 127.0.0.1 for kolla_internal_address would do, but nothing good for certain | 20:56 |
*** thumpba has quit IRC | 20:56 | |
sdake_ | the kolla_internal_address is a HAProxy/keepalived VIF | 20:56 |
daneyon_ | aand binding to 127.0.0.1 seems to work and should. haproxy binds the load-balnced services to the real ip/port and then hits the services on the backend using 127.0.0.1 | 20:56 |
daneyon_ | right | 20:57 |
daneyon_ | bith int and ext are virtual IP's used by haproxy | 20:57 |
sdake_ | I would highly recommend changing that from 127.0.0.1 before continuing | 20:57 |
sdake_ | what can it hurt? | 20:57 |
sdake_ | do you not have a free ip to use? | 20:57 |
sdake_ | the external_interface_address i htnk is just a DNS mapping to the internal address | 20:58 |
sdake_ | but samyaple was not clear on that point | 20:58 |
daneyon_ | sdake_ added by ansible configs to the etherpaad | 20:59 |
*** SourabhP has quit IRC | 20:59 | |
*** SourabhP has joined #kolla | 21:00 | |
daneyon_ | sdake_ i only have the 1 ip. i can try using an ip on the subnet that doesn't respond, but it may cause problems down the line.. Christine Hicks said all other IPs have been assigned. | 21:00 |
sdake_ | do you hae any other interfaces inthe mchine? | 21:01 |
sdake_ | if so give one a fake addess like 10.100.1.1 | 21:01 |
sdake_ | and then set your internal_interface to 10.100.1.2 | 21:01 |
sdake_ | your config looks good | 21:02 |
daneyon_ | sdake_ i can create a fake interface and give it an ip... i'll give that a try | 21:03 |
sdake_ | what I would try just to get down the road a bit is to sue an ip that doens't repsond to ping | 21:03 |
sdake_ | just temporary like | 21:03 |
sdake_ | then we can try workaround | 21:03 |
sdake_ | we can try the other workarounds today after we use this temporary non-responsive ip address | 21:03 |
sdake_ | what could possibly go wrong ;-) | 21:03 |
sdake_ | there is one other thing I did before installing the dependencies that seemed to help alot | 21:05 |
sdake_ | I ran pip install -U wheel pip | 21:06 |
sdake_ | then reinstalled all my dependencies | 21:06 |
*** rhallisey has quit IRC | 21:07 | |
sdake_ | so step 1 is to use an ip address that is non responsive on your main interface | 21:09 |
sdake_ | clenaup-containers -fv and reboot | 21:09 |
sdake_ | step 2 is to run pip uninstall ansible | 21:09 |
sdake_ | step 3 is to yum install ansible | 21:09 |
sdake_ | it would be great if we could get to the root cause of the problem you ahve | 21:10 |
*** bmace has joined #kolla | 21:13 | |
*** daneyon_ has quit IRC | 21:14 | |
*** daneyon has joined #kolla | 21:15 | |
sdake_ | rhallisey are you likewise jammed up? | 21:18 |
*** jtriley has quit IRC | 21:19 | |
sdake_ | jpeeler when you get unstuck from email | 21:21 |
sdake_ | jpeeler I'd like you tog ive the ansible deployment a spin | 21:21 |
sdake_ | you can use a vm if you like | 21:21 |
daneyon | sdake_ take a look at https://etherpad.mozilla.org/y3JRiEFC0e | 21:26 |
daneyon | i can't seem to get mysql to bind to my real ip and have haproxy use a fake ip | 21:26 |
sdake_ | i set external and internal address to the same thing | 21:27 |
sdake_ | you have them set to 85 and 86 | 21:28 |
sdake_ | I have them both set to .148 | 21:28 |
sdake_ | kolla_internal_address should be .85 if that is your fake address | 21:28 |
sdake_ | is your machine's ip .86? | 21:28 |
sdake_ | daneyon ^ | 21:29 |
daneyon | I just set my kolla_external_address and kolla_internal_address to the fake ip and set database_address: "172.29.74.86" (real ip) | 21:30 |
sdake_ | leave database address at whatever it defaults to | 21:30 |
sdake_ | and make your internal and external address match | 21:31 |
sdake_ | as a non-used ip in your network | 21:31 |
sdake_ | or one you ahe borrwed temporarily | 21:31 |
sdake_ | if you do what you are doing, none of the services will work | 21:32 |
sdake_ | because there are other addresses besides "datbasae_address" used tobind the services to the correct ip addresses | 21:32 |
daneyon | sdake_ no luck: https://etherpad.mozilla.org/y3JRiEFC0e | 21:37 |
sdake_ | so 85 is an unused interface correct? | 21:37 |
*** banix has quit IRC | 21:37 | |
daneyon | yes | 21:37 |
sdake_ | her eis my database address: | 21:39 |
sdake_ | database_address: "{{ kolla_internal_address }}" | 21:39 |
sdake_ | your etherpad is a different database address | 21:39 |
sdake_ | you ahve database_address: "{{ network_interface }}" | 21:40 |
sdake_ | i htink you ened to clenaup-containers and reboot between failed attempts too ;-) | 21:40 |
sdake_ | kolla atm in ansible does not restart failed containers properly imo | 21:41 |
daneyon | sdake let me change the db addr to "{{ kolla_internal_address }}" | 21:41 |
sdake_ | if a container fails to start, and then you change the config, it doesn't really restart it even though th econfig has changed | 21:42 |
sdake_ | that has been my experience atleast | 21:42 |
sdake_ | or rather starts then fails | 21:43 |
sdake_ | and then kolla gets jammed up | 21:43 |
daneyon | sdake_ changed db addr to match your but still same problem. | 21:44 |
sdake_ | did you cleanup-containers -fv | 21:45 |
sdake_ | and reboot | 21:45 |
daneyon | i did clean -f | 21:45 |
sdake_ | just tobe clear, 85 is the VIF interface, and .86 i your local box? | 21:45 |
daneyon | let me try -fv | 21:45 |
sdake_ | -fv just deletes volumes | 21:45 |
daneyon | yes .85 is a fake ip .86 is how i manage my box | 21:46 |
sdake_ | network_interface: "em1" | 21:46 |
sdake_ | neutron_external_interface: "enp1s0" | 21:46 |
sdake_ | you hae those set? | 21:47 |
sdake_ | one thing I noticed is you have the interfaces backwards from boot order | 21:47 |
sdake_ | this may be a problem for the implementation | 21:47 |
*** dwalsh has quit IRC | 21:47 | |
sdake_ | network_interface: "enp10s0" | 21:48 |
sdake_ | neutron_external_interface: "enp9s0" | 21:48 |
sdake_ | can you cd to /opt/kolla | 21:48 |
sdake_ | and find a configuration file relating to an openstack service | 21:48 |
sdake_ | and see what it sets bind to? | 21:48 |
sdake_ | bind_host = 192.168.1.100 | 21:48 |
*** jasonsb has quit IRC | 21:51 | |
*** bmace has quit IRC | 21:55 | |
*** bmace has joined #kolla | 21:56 | |
*** thumpba has joined #kolla | 22:19 | |
*** thumpba has quit IRC | 22:29 | |
*** thumpba has joined #kolla | 22:41 | |
*** thumpba has quit IRC | 22:41 | |
*** jasonsb has joined #kolla | 22:45 | |
*** dims_ has joined #kolla | 22:51 | |
*** dims has quit IRC | 22:52 | |
*** dims has joined #kolla | 22:53 | |
*** dims_ has quit IRC | 22:56 | |
*** absubram has quit IRC | 23:05 | |
*** dims has quit IRC | 23:29 | |
*** echoingumesh has quit IRC | 23:40 | |
*** echoingumesh has joined #kolla | 23:41 | |
*** echoingumesh has quit IRC | 23:45 | |
*** blahRus has quit IRC | 23:45 | |
*** banix has joined #kolla | 23:47 | |
*** banix has quit IRC | 23:50 | |
*** vinkman has joined #kolla | 23:51 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!