| *** sacharya_ has quit IRC | 00:01 | |
| *** devl has quit IRC | 00:11 | |
| *** mss_ has quit IRC | 00:13 | |
| *** scarlisle has quit IRC | 00:13 | |
| *** jaybeers has quit IRC | 00:14 | |
| *** CheKoLyN has quit IRC | 00:35 | |
| *** BjoernT has quit IRC | 00:39 | |
| *** Mudpuppy has quit IRC | 00:58 | |
| *** woodard has joined #openstack-ansible | 01:03 | |
| *** woodard has quit IRC | 01:13 | |
| *** phiche1 has joined #openstack-ansible | 01:14 | |
| *** Mudpuppy has joined #openstack-ansible | 01:14 | |
| *** phiche has quit IRC | 01:17 | |
| cloudnull | evenings | 01:36 |
|---|---|---|
| cloudnull | hows it ? | 01:36 |
| coolj | o/ | 01:36 |
| *** tlian2 has joined #openstack-ansible | 01:46 | |
| *** kjelly has joined #openstack-ansible | 01:47 | |
| *** tlian has quit IRC | 01:49 | |
| *** cemmason has joined #openstack-ansible | 01:51 | |
| openstackgerrit | Merged openstack/openstack-ansible: Reference correct group in haproxy_config.yml https://review.openstack.org/244637 | 02:04 |
| *** sacharya has joined #openstack-ansible | 02:05 | |
| *** kmARC has joined #openstack-ansible | 02:10 | |
| *** Bjoern_ has joined #openstack-ansible | 02:13 | |
| *** sdake has joined #openstack-ansible | 02:28 | |
| *** mss has joined #openstack-ansible | 02:32 | |
| *** fawadkhaliq has joined #openstack-ansible | 02:42 | |
| *** tlian2 has quit IRC | 02:50 | |
| cloudnull | what up coolj | 03:04 |
| coolj | java shennanigans trying to get ilo remote console working :\ | 03:18 |
| *** cemmason has quit IRC | 03:19 | |
| *** tlian has joined #openstack-ansible | 03:22 | |
| cloudnull | fight the java powers http://lists.openstack.org/pipermail/openstack-dev/2015-November/078997.html | 03:27 |
| cloudnull | operators, please tell dev their doing it wrong... :) | 03:28 |
| *** cemmason has joined #openstack-ansible | 03:30 | |
| cloudnull | the people vs zookeeper | 03:30 |
| gus | cloudnull: redis isn't a suitable solution here, in case clint's reply didn't make the case sufficiently. | 03:34 |
| cloudnull | gus: yea i can see why it wouldnt be however I think there are still better choices than zk . | 03:34 |
| gus | right. I think the choice is basically between zk, etcd, consul. | 03:34 |
| cloudnull | I'd be interested in helping out in creating the drivers for console, etcd, etc... | 03:35 |
| coolj | what is DLM in this context? | 03:35 |
| gus | the latter 2 in go, so we've got an "exotic" language in all 3 of those. | 03:35 |
| cloudnull | and maintaining them | 03:35 |
| *** sdake has quit IRC | 03:35 | |
| gus | coolj: distributed lock manager. | 03:35 |
| coolj | ah | 03:35 |
| gus | cloudnull: those other backend choices already exist in tooz aiui. | 03:35 |
| cloudnull | however zk brings in java -- and while "go" may be an encomberered language in terms of Google (Alphabet) controlling it I'd say google has less of a track record of being evil than Oracle does. | 03:37 |
| gus | my understanding of the tradeoffs (fwiw): zk - mature, featureful, works, but java; etcd - simple featureset which is both strength and weakness; consul - more features than etcd, but smallest community / less widely deployed of the 3. | 03:38 |
| cloudnull | gus: also Ive found numerious open issues in ZK that have been open for years which can result in the same failure domains as redis using redislock, so idk if the maturity is really a factor to be considered here. | 03:38 |
| gus | cloudnull: agreed re java vs go. My understanding is that zk works on openjdk, but a) there's a lot of confusion with little data around that on the internets and b) openjdk is still at the mercy of oracle. | 03:39 |
| cloudnull | its really about the right tool being used, and if its ZK thats fine , but i have a fear of the Java infection | 03:39 |
| cloudnull | also we cant get access to the proper dev kits in OpenJDK because their controlleb by Oracle (noted in earlier messages) | 03:40 |
| cloudnull | however ZK does work in OpenJDK and has been developed with OpenJDK in mind | 03:40 |
| cloudnull | but scale will be an issue for the likes RAX/IBM public cloud, and potentially RAX Private cloud | 03:41 |
| gus | cloudnull: I haven't looked through all the zk bugs in that search, but most of them seem to be code/thread races which could lead to hangs (or crashes) rather than corrupted data. | 03:42 |
| cloudnull | i;ve not used ZK at large scale but i have played with java apps w/ OpenJDK at moderate scale and it didnt end well. | 03:42 |
| cloudnull | gus thats true but they're == in that a hang will result in a missed message or a released lock which will cause problems (including corrupt data especially in terms of ironic as noted in the previous thread) | 03:43 |
| gus | I don't think its fair to compare redis and zk - zk comes from the beginning from a strong theoretical approach to distributed consensus, whereas redis (afaiui) was a fast in-memory datastructure that has tried to retro-fit persistence in later revisions. | 03:43 |
| *** Mudpuppy has quit IRC | 03:43 | |
| cloudnull | gus your right , i looked at redis as an alternitive because it has 0 deps . | 03:44 |
| cloudnull | and is far lighter than zk | 03:44 |
| *** Mudpuppy has joined #openstack-ansible | 03:44 | |
| cloudnull | which would be ideal for the devstack gate | 03:44 |
| gus | cloudnull: yeah, I think it's reasonable to have a non-robust single-host option for devstack/aio testing. | 03:45 |
| gus | May as well use memcache, assuming the tooz driver for that fakes enough of the API on top. | 03:45 |
| *** sdake has joined #openstack-ansible | 03:45 | |
| gus | One of the concerns about using tooz here is that the drivers don't necessarily expose the same semantics/featureset. | 03:46 |
| *** tlian has quit IRC | 03:46 | |
| *** sacharya has quit IRC | 03:46 | |
| gus | (I mean: this was one of the primary concerns against tooz mentioned in the dlm session at tokyo) | 03:47 |
| *** Mudpuppy has quit IRC | 03:48 | |
| cloudnull | i wasn't in those talks sadly . | 03:48 |
| cloudnull | in the thread those concers seem to be missing . | 03:48 |
| cloudnull | when i started on the thread my point originally was for us to use something that will allow us to get the tools into the developers hands without disrupting their workflow in a way that could resemble production. w/ redis + tooz (or any other abstration layer) we could just that without forcing devs to move onto other systems. | 03:49 |
| gus | "java" was definitely the primary concern against zk, so I think that issue was well represented. | 03:49 |
| *** Mudpuppy has joined #openstack-ansible | 03:49 | |
| cloudnull | yea for sure | 03:49 |
| cloudnull | so what is missing that people want in a DLM solution ? | 03:50 |
| gus | in redis? | 03:50 |
| cloudnull | no in general | 03:50 |
| *** Bjoern_ has quit IRC | 03:50 | |
| gus | oh, nothing. | 03:50 |
| cloudnull | it seems that people really want a service discovery | 03:50 |
| gus | oh you mean what do people hope to gain from using a DLM. | 03:50 |
| cloudnull | yes | 03:51 |
| cloudnull | it seems like theyre trying to shoe something with a solution looking for a problem to end up doing something different in the future. | 03:51 |
| gus | One is service discovery (ironic in particular), others are cheap/robust locking (ironic, cinder, and just about everyone else) | 03:51 |
| * cloudnull my opinion after only being apart of the ML thread | 03:51 | |
| *** Mudpuppy has quit IRC | 03:52 | |
| *** jhesketh has quit IRC | 03:52 | |
| gus | cloudnull: another way to look at this, is that every other cloud implementation I'm aware of (google's borg, kubernetes, aws, mesos, marathon) went with distributed key-value stores. | 03:53 |
| gus | openstack went with SQL. | 03:53 |
| *** jhesketh has joined #openstack-ansible | 03:53 | |
| cloudnull | yea theres something to be said there | 03:54 |
| gus | I happen to agree with everyone else ;) the zk/etcd/consul hammer lets you solve _so_ many problems in this space in a way that is easier to understand, (generally) more reliable, more scalable than SQL-based solutions. | 03:54 |
| cloudnull | no argument here . | 03:54 |
| *** cemmason has quit IRC | 03:55 | |
| cloudnull | just zk brings java, (etcd|console) brings go so I lean toward the later . | 03:55 |
| gus | So the fact that we even have to talk about it, and talk about it to the exclusion of actually making _any_ progress, is the silly bit - imo. | 03:55 |
| *** eil397 has joined #openstack-ansible | 03:56 | |
| gus | So: we can identify a bunch of immediate use cases where projects want to use a DLM. | 03:56 |
| *** cemmason has joined #openstack-ansible | 03:56 | |
| gus | Can they be done with SQL too? Yes. But it's just more awkwardness piled on the existing awkwardness. | 03:56 |
| gus | Most of the discussion is fake because people don't want to acknowledge that they don't actually _need_ a DLM, they're just looking for a way to finally convince the unwashed masses to let them use what we should have always used. | 03:58 |
| cloudnull | so the real purpose is to expand our capability set by getting a supported key/value store + abstration layer. DLM is simply a gateway to that better life | 03:58 |
| gus | yep. | 03:58 |
| gus | (at least this is my take on it, and from hallway discussions with the various people behind it) | 03:58 |
| *** cemmason has quit IRC | 03:59 | |
| *** Mudpuppy has joined #openstack-ansible | 03:59 | |
| *** cemmason has joined #openstack-ansible | 03:59 | |
| gus | 1) it's going to be ok. 2) As soon as we're "allowed" to assume a DLM, it's going to clean up a bunch of existing SQL-based code. | 03:59 |
| cloudnull | so how far off is tooz and how do we not infect OpenStack w/ Oracle Java? | 04:02 |
| gus | (or for a number of projects, the existing code is _not_ sql-based and is just plain broken in the face of machine failure) | 04:02 |
| gus | tooz has been around and ready for a year or so - it's quite a thin wrapper around the basic primitives that these systems offer so is pretty easy. | 04:04 |
| gus | the reason the room went with tooz rather than picking one particular DLM is that the choice of "best" DLM wasn't clear. | 04:04 |
| gus | so the expectation is that after a few years, a "defacto" best will emerge, like what happened with sql (galera). | 04:05 |
| gus | .. we just don't know which one that will be at this point. | 04:05 |
| cloudnull | if we go with whats currently in tooz etcd/consul are not options at this point | 04:06 |
| cloudnull | https://github.com/openstack/tooz/tree/master/tooz/drivers | 04:06 |
| gus | so to avoid java: make sure one of the others works really well, is easy to deply, is the default (<- probably the major factor). | 04:06 |
| *** fawadkhaliq has quit IRC | 04:06 | |
| cloudnull | is the default looks like itll be ZK | 04:06 |
| gus | oh wow, I always assumed they were without even looking. | 04:06 |
| gus | huh - perhaps we should add those, then. | 04:07 |
| cloudnull | which goes back to why i argued for redis at this point :) | 04:07 |
| cloudnull | hahaha | 04:07 |
| cloudnull | but yes we should add them | 04:07 |
| gus | https://blueprints.launchpad.net/python-tooz/+spec/add-consul-driver | 04:07 |
| cloudnull | https://review.openstack.org/#/c/245362/ | 04:07 |
| prometheanfire | cloudnull: http://nfv.space/2015/11/17/ansible-module-python-api-development/ | 04:08 |
| cloudnull | looks like Josh was keen to review, will likely need a few more interested parties | 04:08 |
| cloudnull | prometheanfire: yes ? | 04:10 |
| prometheanfire | just thought you'd be intrested is all | 04:13 |
| cloudnull | ah . | 04:15 |
| cloudnull | tyvm | 04:15 |
| *** fawadkhaliq has joined #openstack-ansible | 04:27 | |
| *** eil397 has quit IRC | 04:35 | |
| *** markvoelker_ has quit IRC | 04:40 | |
| *** scarlisle has joined #openstack-ansible | 04:44 | |
| *** alkari has joined #openstack-ansible | 04:55 | |
| *** HimanshuGarg has quit IRC | 05:01 | |
| *** Mudpuppy has quit IRC | 05:02 | |
| *** Bjoern_ has joined #openstack-ansible | 05:50 | |
| *** Bjoern_ has quit IRC | 05:55 | |
| *** galstrom_zzz is now known as galstrom | 05:56 | |
| *** elgertam has joined #openstack-ansible | 06:17 | |
| *** markvoelker has joined #openstack-ansible | 06:40 | |
| *** markvoelker has quit IRC | 06:45 | |
| *** alkari has quit IRC | 06:47 | |
| *** phiche1 has quit IRC | 07:22 | |
| *** subscope has joined #openstack-ansible | 07:24 | |
| *** mss has quit IRC | 07:31 | |
| *** phiche has joined #openstack-ansible | 07:41 | |
| *** markvoelker has joined #openstack-ansible | 07:41 | |
| *** scarlisle has quit IRC | 07:42 | |
| *** markvoelker has quit IRC | 07:46 | |
| *** jmccrory has quit IRC | 07:49 | |
| *** jmccrory has joined #openstack-ansible | 07:54 | |
| *** slotti has joined #openstack-ansible | 07:54 | |
| *** galstrom is now known as galstrom_zzz | 07:55 | |
| *** javeriak has joined #openstack-ansible | 08:07 | |
| *** ntt has joined #openstack-ansible | 08:14 | |
| *** GheRivero has left #openstack-ansible | 08:15 | |
| ntt | Hi, I'm trying to use openstack-ansible and I'd like to know if it is possible to install openstack with ansible on centos physical machine with lxc for controller and network nodes (perhaps with a fork of the original repo) ? | 08:15 |
| *** mss has joined #openstack-ansible | 08:27 | |
| *** gparaskevas has joined #openstack-ansible | 08:28 | |
| *** javeriak_ has joined #openstack-ansible | 08:29 | |
| *** egonzalez has joined #openstack-ansible | 08:30 | |
| *** javeriak has quit IRC | 08:31 | |
| *** mpavone has joined #openstack-ansible | 08:33 | |
| *** karimb has joined #openstack-ansible | 08:35 | |
| *** fawadkhaliq has quit IRC | 08:41 | |
| *** fawadkhaliq has joined #openstack-ansible | 08:42 | |
| *** sacharya has joined #openstack-ansible | 08:49 | |
| *** sacharya has quit IRC | 08:53 | |
| *** fawadkhaliq has quit IRC | 08:57 | |
| odyssey4me | ntt we're looking for contributors to make that happen | 09:17 |
| odyssey4me | right now, it's not possible | 09:17 |
| ntt | odyssey4me: I'm trying https://github.com/pvaylov/openstack-ansible | 09:18 |
| evrardjp | but it shouldn't be difficult if you're using lxc and linuxbridges | 09:18 |
| ntt | this kind of deployment can be considered in an environment production? | 09:19 |
| ntt | lxc for controller and network, I mean | 09:19 |
| evrardjp | there are ppl using openstack-ansible in production | 09:20 |
| evrardjp | I, for example :) | 09:20 |
| *** egonzalez has quit IRC | 09:20 | |
| odyssey4me | ntt yes, Rackspace Private Cloud uses it in production extensively | 09:24 |
| odyssey4me | there are several other organisations using openstack-ansible in production | 09:24 |
| ntt | evrardjp: I'm searching for a deployment tool for openstack. Configure the enviroment without an automatic tool is a pain and it's difficult | 09:24 |
| ntt | ok.... good. But I need to use centos | 09:25 |
| odyssey4me | there is one that I know of that uses a (closed, unfortunately) fork of openstack-ansible on RHEL | 09:25 |
| evrardjp | odyssey4me: isn't it possible to bring them to the community? | 09:25 |
| odyssey4me | ntt if you're prepared to be part of a group that does the work to make that happen, then it can happen | 09:25 |
| odyssey4me | evrardjp I'm trying to coax someone to take the lead on that work, but no-one has bitten yet | 09:26 |
| evrardjp | :) | 09:26 |
| evrardjp | redhat guys? | 09:26 |
| ntt | I'm working for a small ISP and we are interested | 09:26 |
| ntt | I'm using openstack since the grizzly release and actually I'm using icehouse | 09:27 |
| ntt | I have big problems related to the upgrade | 09:27 |
| odyssey4me | ntt yeah, the upgrades will not be much fun | 09:27 |
| ntt | absolutely | 09:27 |
| odyssey4me | ntt if you can join our Thu community meeting and raise CentOS support on the agenda, then we can talk about it | 09:28 |
| odyssey4me | ntt https://wiki.openstack.org/wiki/Meetings/openstack-ansible#Agenda_for_next_meeting | 09:28 |
| ntt | But with a deployment tool, and a deep knowledge of the database structure of openstack, It can be more easy | 09:28 |
| odyssey4me | if you can add it to the open discussion | 09:28 |
| odyssey4me | that'd be great | 09:28 |
| ntt | I'm completely new to openstack meetings.... I'd like to learn :) | 09:29 |
| kmARC | hi all, I'm still strugling with setting up AIO (liberty) Currently I have a problem with neutron agents: | 09:33 |
| kmARC | failed: [aio1_neutron_server_container-2e866b98] => {"failed": true} | 09:33 |
| kmARC | msg: neutron fact collection failed: unable to find migration with revision 59cb5b6cf4d | 09:33 |
| kmARC | I already tried re-running openstack-ansible setup-everything.yml --limit aio1_neutron_server_container-$HASH | 09:34 |
| kmARC | also tried openstack-ansible lxc-containers-destroy --limit neutron_all | 09:34 |
| kmARC | the task it dies at: TASK: [os_neutron | Get neutron migrations facts] | 09:35 |
| *** mpavone has quit IRC | 09:37 | |
| openstackgerrit | Jesse Pretorius proposed openstack/openstack-ansible: Remove unused neutron variable definition https://review.openstack.org/246822 | 09:37 |
| odyssey4me | kmARC we have a patch in review that may fix that - if you can test and provide feedback in the review, that'd be great: https://review.openstack.org/240560 | 09:38 |
| ntt | odyssey4me: how can I add myself to the list of regular attendees? | 09:39 |
| odyssey4me | ntt on the top right of the page you'll be able to login, once you've done that you'll be able to edit the wiki page and add yourself | 09:40 |
| odyssey4me | I've added Multi-platform support discussion to the agenda :) | 09:40 |
| ntt | Should I use an ubuntu one account? | 09:41 |
| odyssey4me | ntt yes, you must as far as I know | 09:42 |
| kmARC | odyssey4me: will look | 09:42 |
| ntt | ok | 09:42 |
| *** markvoelker has joined #openstack-ansible | 09:42 | |
| odyssey4me | ntt you'll probably need to work through http://docs.openstack.org/infra/manual/developers.html | 09:43 |
| ntt | ok, added myself to the attendees | 09:44 |
| odyssey4me | hughsaunders with regards to patterns for multi-OS enablement: https://github.com/odyssey4me/ansible-openvas/blob/master/tasks/install_common.yml and https://github.com/odyssey4me/ansible-openvas/blob/master/tasks/install_library.yml#L18-L30 easily handles different package installers and package lists | 09:44 |
| kmARC | odyssey4me: so that means even if I run the one-liner curl script with replacing "master" to "liberty" will still install packages from master? | 09:45 |
| kmARC | (anyways, seems the patch is working, now the neutron container are set up properly) | 09:45 |
| odyssey4me | kmARC did you download and modify the script? | 09:46 |
| kmARC | you mean the patch? | 09:46 |
| *** neilus1 has quit IRC | 09:46 | |
| odyssey4me | kmARC no, I mean the one-liner AIO script | 09:46 |
| *** markvoelker has quit IRC | 09:46 | |
| kmARC | nope, I just modified this line: curl https://raw.githubusercontent.com/openstack/openstack-ansible/master/scripts/run-aio-build.sh | sudo bash | 09:47 |
| kmARC | by replacing "master" to "liberty" | 09:47 |
| kmARC | and then I see that in /opt/openstack-ansible I am on the master branch | 09:47 |
| odyssey4me | kmARC ah yes - https://github.com/openstack/openstack-ansible/blob/liberty/scripts/run-aio-build.sh#L23 needs to be changed to refer to the liberty branch | 09:48 |
| odyssey4me | kmARC care to put in a patch for that? | 09:48 |
| kmARC | odyssey4me: I can try :-) | 09:50 |
| evrardjp | odyssey4me: about the multi-OS enablement changes, maybe wait for tomorrow? ;) | 10:06 |
| evrardjp | I recall a promise of someone that is due tomorrow.... probably related to ansible2 | 10:07 |
| odyssey4me | evrardjp haha, oh really :p | 10:12 |
| odyssey4me | evrardjp so yes, we'll do an initial discussion tomorrow - then a broader workshop the week after tomorrow | 10:13 |
| odyssey4me | tomorrow we're doing a workshop on improving the install guide: https://etherpad.openstack.org/p/oa-install-docs | 10:13 |
| evrardjp | ansible fest is tomorrow evening, maybe we'll see a new release the day after tomorrow :D | 10:13 |
| evrardjp | odyssey4me: I'd happy to help if if fits my agenda. do you have an idea of the timeframe? | 10:14 |
| odyssey4me | evrardjp that would depend on the time contributors have - I'd like to see it done in the Mitaka cycle | 10:15 |
| evrardjp | maybe mentionning it in this chan's topic would be good | 10:16 |
| evrardjp | "next documentation hackathon on <date>" | 10:16 |
| evrardjp | on the wiki ofc :) | 10:17 |
| evrardjp | maybe prepare more than one, to bootstrap the process, and then reduce the frequency when the doc team estimates the quality has been reached? | 10:18 |
| evrardjp | (but still continue to avoid documentation drift) | 10:18 |
| odyssey4me | yeah, that needs to be part of the discussion - how do we make it good and keep it good | 10:19 |
| openstackgerrit | Mark Korondi proposed openstack/openstack-ansible: Use 'liberty' branch for packages when using one-liner aio build script https://review.openstack.org/246840 | 10:20 |
| odyssey4me | evrardjp if you could chime in with your intended support, that'd be great: http://lists.openstack.org/pipermail/openstack-dev/2015-November/079732.html | 10:20 |
| odyssey4me | ntt same for you :) | 10:20 |
| kmARC | odyssey4me: https://review.openstack.org/246840 | 10:21 |
| evrardjp | there s still work to do on independant roles, for example explaining how we'll do versioning of os_ roles | 10:23 |
| odyssey4me | kmARC perfect, thank you :) | 10:23 |
| evrardjp | I don't mind helping, I just don't have a centos lab available :p | 10:24 |
| odyssey4me | evrardjp yeah, but until the roles are all split out we'll probably not figure that bit out | 10:24 |
| odyssey4me | I'm testing out the format of using half the community meeting for a focus workshop to discuss a topic. If it works then we'll continue it. | 10:25 |
| odyssey4me | And we can use that platform to discuss stuff like that | 10:26 |
| evrardjp | this is going to be tough for the one holding the meeting | 10:26 |
| evrardjp | (the chair I mean) | 10:26 |
| odyssey4me | But it's important to also use the ML too in order to catch people who don't attend the meeting, so I'm opting to send out an ML post announcing the discussion and providing the etherpad link. | 10:27 |
| evrardjp | half the time to do the rest, and discussion to hold afloat during 30 minutes | 10:27 |
| odyssey4me | I'll do the same for the docs discussion now. | 10:27 |
| evrardjp | odyssey4me: I think it's good to use the ML | 10:27 |
| odyssey4me | yeah, it does cut down the time available - but it does also bring focus | 10:28 |
| evrardjp | at least to call people to have a discussion | 10:28 |
| evrardjp | (during meetings or here in this channel) | 10:28 |
| odyssey4me | essentially the primary things we need to discuss in the meetings relates to design work, and release decisions | 10:28 |
| evrardjp | the hardest is to manage time and scope on this | 10:28 |
| evrardjp | right | 10:28 |
| kmARC | odyssey4me: I am not familiar with the openstack-ansible review workflow. Should I give in additional infos to get this patch through? | 10:30 |
| evrardjp | do you mind if I adapt the wiki to adjust what you said and give me your opinion? | 10:30 |
| odyssey4me | kmARC in this particular case the patch is simple and the purpose obvious, so there's nothing more that you need to do except to wait for another core to vote it through | 10:30 |
| kmARC | ok thanks | 10:31 |
| kmARC | anyways I ran into the next problem. failed: [aio1] => {"failed": true} | 10:31 |
| kmARC | msg: Failed to reload sysctl: fs.inotify.max_user_watches = 36864 | 10:31 |
| odyssey4me | evrardjp I have adjusted the agenda to reflect that, but if you think it could be made clearer then go for it | 10:31 |
| odyssey4me | kmARC you are using Ubuntu 14.04 right? | 10:32 |
| kmARC | yes | 10:32 |
| kmARC | what I see, I already have these kind of settings with sysctl | 10:32 |
| odyssey4me | kmARC and does your image have the ext4 file system as the base? | 10:32 |
| kmARC | yes | 10:33 |
| odyssey4me | odd, I wonder why Ansible would be failing on those tasks in your case. | 10:33 |
| evrardjp | odyssey4me: you can check, it's just a simple edit, but this brings clarity on how you (or any chair, because you planned to rotate this role) will manage the schedule: review/release/design[=next steps]/open-discussion | 10:33 |
| odyssey4me | kmARC what's the ansible version being used? | 10:33 |
| ntt | Is there a way to remove network interfaces using openstack-ansible? (eg: scripts)? | 10:34 |
| kmARC | whatever the aio script installed :) | 10:34 |
| kmARC | 1.9.4 | 10:34 |
| kmARC | so but the errr is this: | 10:34 |
| kmARC | sysctl: cannot stat /proc/sys/net/bridge/bridge-nf-call-ip6tables: No such file or directory | 10:34 |
| kmARC | sysctl: cannot stat /proc/sys/net/bridge/bridge-nf-call-iptables: No such file or directory | 10:34 |
| kmARC | sysctl: cannot stat /proc/sys/net/bridge/bridge-nf-call-arptables: No such file or directory | 10:34 |
| odyssey4me | kmARC ok - just verifying | 10:34 |
| kmARC | ls /proc/sys/net/ | 10:35 |
| kmARC | core dccp ipv4 ipv6 netfilter nf_conntrack_max unix | 10:35 |
| odyssey4me | kmARC ah, so it seems that packages are missing which are expected... we are perhaps assuming too much | 10:35 |
| kmARC | yeah, installing arptables, etc. | 10:36 |
| *** fawadkhaliq has joined #openstack-ansible | 10:36 | |
| odyssey4me | kmARC whatever you discover is missing should be added to https://github.com/openstack/openstack-ansible/blob/liberty/scripts/bootstrap-aio.sh#L101-L115 :) | 10:37 |
| kmARC | okay will see | 10:37 |
| kmARC | however iptables are installed | 10:37 |
| evrardjp | bridge-utils? | 10:38 |
| kmARC | yepp | 10:38 |
| kmARC | otherwise bridges would've not be created | 10:38 |
| evrardjp | just to make sure :p | 10:38 |
| evrardjp | do you have everything set in your /etc/modules? | 10:39 |
| kmARC | looks like | 10:39 |
| evrardjp | like 8021q, ip_tables, nf_conntrack... | 10:39 |
| kmARC | 8021q is there, ip_tables, ip6_tables too | 10:39 |
| kmARC | lsmod also lists them | 10:40 |
| odyssey4me | ntt there is a way to delete any file on many hosts with ansible :) | 10:41 |
| evrardjp | many files too | 10:41 |
| evrardjp | don't run ansible -m shell -a "rm -rf /" all for example | 10:42 |
| odyssey4me | hahaha, probably a bad idea - unless of course you're keen on vastly simplifying your life | 10:43 |
| odyssey4me | ntt there's a good explanation of how to do that sort of thing in http://docs.ansible.com/ansible/intro_getting_started.html and http://docs.ansible.com/ansible/intro_adhoc.html | 10:43 |
| ntt | Sorry.... I don't think I expressed myself properly. I meant: I tried to launch openstack-ansible aio, and now I want to cleanup all (containers, firewall rules and network interfaces.....) | 10:46 |
| openstackgerrit | Merged openstack/openstack-ansible: Dropping logrotate.conf for all Openstack aggregate logs https://review.openstack.org/245920 | 10:46 |
| ntt | I found a script : teardown.sh | 10:46 |
| odyssey4me | ntt: http://docs.openstack.org/developer/openstack-ansible/developer-docs/quickstart-aio.html#rebuilding-the-aio | 10:46 |
| ntt | but It seems that I cannot delete network interfaces | 10:47 |
| odyssey4me | ntt the network interfaces is just a file in /etc/network/interfaces.d | 10:47 |
| *** sacharya has joined #openstack-ansible | 10:54 | |
| odyssey4me | evrardjp I've announced the documentation workshop too now: http://lists.openstack.org/pipermail/openstack-dev/2015-November/079734.html | 10:57 |
| evrardjp | I think it's time for me to subscribe this list | 10:58 |
| *** sacharya has quit IRC | 10:58 | |
| odyssey4me | evrardjp it's quite busy, so be sure to use filters | 11:00 |
| odyssey4me | I've also posted to the openstack-operators list, which is less busy and has a different audience. | 11:01 |
| odyssey4me | ie http://lists.openstack.org/pipermail/openstack-operators/2015-November/008847.html and http://lists.openstack.org/pipermail/openstack-operators/2015-November/008849.html | 11:01 |
| *** HimanshuGarg has joined #openstack-ansible | 11:04 | |
| prometheanfire | odyssey4me: https://bugs.launchpad.net/openstack-ansible/+bug/1476770 should be fixed with any new tag | 11:08 |
| openstack | Launchpad bug 1476770 in python-glanceclient "_translate_from_glance fails with "AttributeError: id" in grenade" [High,In progress] - Assigned to Flavio Percoco (flaper87) | 11:08 |
| prometheanfire | odyssey4me: the upper-constraints was bumped to 1.1.0 for glanceclient, which seems to fix it (did for me) | 11:08 |
| odyssey4me | prometheanfire ah, nice - I guess that sigmavirus24_awa's patch made it through then | 11:11 |
| prometheanfire | I guess | 11:11 |
| odyssey4me | evrardjp do you think you can make the mid cycle ops meetup for EU: https://etherpad.openstack.org/p/MAN-ops-meetup and http://europeanopenstackoperatorsmeetup.eventbrite.com/ | 11:17 |
| *** aswadr has joined #openstack-ansible | 11:32 | |
| ntt | How can I exclude some openstack components from the installation? (I would like to exclude swift and ceilometer) | 11:33 |
| odyssey4me | ntt excluding swift is simply a matter of not including any swift_hosts in openstack_user_config | 11:37 |
| odyssey4me | excluding ceilometer is similar, don't include any metering-compute_hosts and metering-infra_hosts | 11:38 |
| odyssey4me | ceilometer enablement in the services is disabled by default in the base user_variables file provided: https://github.com/openstack/openstack-ansible/blob/master/etc/openstack_deploy/user_variables.yml#L21-L25 | 11:39 |
| *** subscope has quit IRC | 11:41 | |
| *** subscope has joined #openstack-ansible | 11:42 | |
| *** markvoelker has joined #openstack-ansible | 11:43 | |
| ntt | odyssey4me: so, 172.29.236.0/24 should be a sort of management network using the etc/openstack_deploy/openstack_user_config.yml example file? | 11:43 |
| *** markvoelker has quit IRC | 11:47 | |
| *** jhesketh has quit IRC | 11:55 | |
| kmARC | odyssey4me: what now? https://review.openstack.org/#/c/246840/ | 11:56 |
| kmARC | I guess this problem comes from something other | 11:56 |
| *** jhesketh has joined #openstack-ansible | 12:00 | |
| *** egonzalez has joined #openstack-ansible | 12:03 | |
| odyssey4me | kmARC yes, an unrelated failure - a recheck is in progress | 12:04 |
| odyssey4me | ntt yes, the container/management network is the primary network which is used for containers to communicate with each other and for the ansible deployment host to talk to the containers/hosts | 12:05 |
| ntt | actually I have a lot of errors: failed: [aio1_nova_scheduler_container-e4ac81e9 -> aio1] => {"error": "Failed to start container [ aio1_nova_scheduler_container-e4ac81e9 ]", "failed": true, "lxc_container": {"init_pid": -1, "interfaces": [], "ips": [], "state": "stopped"}, "rc": 1} | 12:07 |
| ntt | and in /var/log/messages : Cannot find device "vethH6PSO3" | 12:08 |
| ntt | I'm trying to relaunch an aio installation after destroying containers | 12:08 |
| kmARC | odyssey4me: actually I ran into the same issue on my local playground. Now I am going to redeploy it from ö disabling everything which is not needed for a basic setup | 12:09 |
| kmARC | (and of course now with manual steps, on libertyÖ | 12:09 |
| *** aswadr has quit IRC | 12:15 | |
| *** neilus has joined #openstack-ansible | 12:20 | |
| *** subscope has quit IRC | 12:20 | |
| *** subscope has joined #openstack-ansible | 12:21 | |
| *** subscope has quit IRC | 12:21 | |
| *** egonzalez has quit IRC | 12:22 | |
| *** alejandrito has joined #openstack-ansible | 12:23 | |
| *** hassaan has joined #openstack-ansible | 12:29 | |
| *** ziggypup has joined #openstack-ansible | 12:30 | |
| *** ziggypup has left #openstack-ansible | 12:30 | |
| hassaan | hi | 12:37 |
| hassaan | I am getting some errors in nova's logs when trying to create a snapshot of a running instance. The snapshot never shows on the horizon/cli and in glance db the snapshots created have the "Deleted" status. Nova's logs are not very helpful. They state: AttributeError: id | 12:39 |
| hassaan | can someone please comment or give some insight. Is this a bug that needs to be filed? | 12:40 |
| *** hassaan has quit IRC | 12:41 | |
| *** hassaan has joined #openstack-ansible | 12:41 | |
| *** hassaan is now known as hassp | 12:42 | |
| *** fawadkhaliq has quit IRC | 12:42 | |
| *** hassp is now known as hassaan | 12:42 | |
| hassaan | i am using the kilo and liberty branches. faced the same error on both | 12:43 |
| *** hassaan has quit IRC | 12:43 | |
| *** markvoelker has joined #openstack-ansible | 12:43 | |
| *** markvoelker has quit IRC | 12:48 | |
| odyssey4me | hassaan you're hitting https://bugs.launchpad.net/openstack-ansible/+bug/1476770 | 12:48 |
| openstack | Launchpad bug 1476770 in python-glanceclient "_translate_from_glance fails with "AttributeError: id" in grenade" [High,In progress] - Assigned to Flavio Percoco (flaper87) | 12:48 |
| odyssey4me | now that the issue has been resolved upstream, we can bump SHA's and release a fresh tag to resolve it | 12:49 |
| odyssey4me | this will be discussed tomorrow | 12:49 |
| *** mpavone has joined #openstack-ansible | 12:57 | |
| evrardjp | odyssey4me: I'll check for the mid-cycle meetup | 13:00 |
| mancdaz | question: I want to include some playbook only if there is a certain entry in the host_vars (added through container_vars in the user_config) | 13:02 |
| mancdaz | specifically, I want to only include it if the cinder backend is set to lvm | 13:03 |
| *** markvoelker has joined #openstack-ansible | 13:09 | |
| *** Reddy has joined #openstack-ansible | 13:12 | |
| *** sacharya has joined #openstack-ansible | 13:19 | |
| odyssey4me | mancdaz it may make more sense to allow the playbook to always run, but make it do nothing if the backend is not lvm | 13:20 |
| evrardjp | mancdaz: is it a specific role? | 13:20 |
| evrardjp | if it's a specific role, you can have run it conditionally | 13:20 |
| evrardjp | -have | 13:20 |
| mancdaz | odyssey4me evrardjp it's a task within a role | 13:20 |
| mancdaz | I think I found what I need | 13:21 |
| evrardjp | then only when: is needed | 13:21 |
| mancdaz | since the cinder backends get added to the hostvars | 13:21 |
| mancdaz | I can run it 'when' lvm in hostvars['cinder_backends'] | 13:21 |
| mancdaz | s/run/include/ | 13:21 |
| mancdaz | so include it in a playbook, but using that conditional | 13:22 |
| mancdaz | odyssey4me kind of like we already do here: https://github.com/rcbops/rpc-openstack/blob/master/rpcd/playbooks/roles/rpc_maas/tasks/main.yml | 13:23 |
| *** sacharya has quit IRC | 13:24 | |
| evrardjp | this is something we should avoid :p | 13:24 |
| evrardjp | IMO | 13:24 |
| odyssey4me | mancdaz yep, it may be prudent to also include the length check to validate that the group has any hosts at all | 13:24 |
| evrardjp | but sometimes, we don't have choice | 13:24 |
| mancdaz | evrardjp you mean conditional includes? | 13:24 |
| evrardjp | no the conditional rules are important | 13:24 |
| evrardjp | but the coupling of the inventory in the roles should be (still IMO) avoided as much as possible | 13:25 |
| evrardjp | I'd rather do additional tasks in the playbook instead of adding inventory stuff in the role | 13:26 |
| evrardjp | on this: https://github.com/openstack/openstack-ansible/blob/master/playbooks/haproxy-install.yml you can see you can run roles conditionally | 13:27 |
| evrardjp | (L24) | 13:27 |
| mancdaz | evrardjp ah ok you mean do the logic in the playbooks and keep the roles cleaner | 13:27 |
| evrardjp | it's not what you're looking for, but it will help in a later phase, when decoupling inventory and roles | 13:27 |
| evrardjp | mancdaz: yup | 13:27 |
| evrardjp | this way roles can be applied everywhere | 13:28 |
| evrardjp | with different inventories, or whatever | 13:28 |
| mancdaz | evrardjp yes, I think that is the general direction everyone wants to go in | 13:28 |
| evrardjp | it is | 13:28 |
| evrardjp | :) | 13:28 |
| evrardjp | it could bring a larger community | 13:28 |
| evrardjp | which is good :) | 13:28 |
| mancdaz | always! | 13:29 |
| evrardjp | if you're looking for a filter based on the size of a group, check this: https://github.com/openstack/openstack-ansible/blob/324597e29592aed8fc8270e8c2cd1411ddf79e2c/playbooks/roles/repo_server/tasks/main.yml#L24 | 13:29 |
| *** woodard has joined #openstack-ansible | 13:34 | |
| *** hassaan has joined #openstack-ansible | 13:39 | |
| *** tlian has joined #openstack-ansible | 13:45 | |
| kmARC | odyssey4me: failure again. TBH I can't make it work either on my local playground. Did this thing ever work with liberty? | 13:47 |
| odyssey4me | kmARC we use it for gate checks on every commit, so yes :) | 13:49 |
| kmARC | hm | 13:50 |
| kmARC | It always has the same problem | 13:50 |
| kmARC | one of the galera containers fails to come up | 13:50 |
| odyssey4me | kmARC what RAM/CPU specs does your test server have? | 13:51 |
| kmARC | 16GB RAM, 80GB Disk, 12Cores | 13:51 |
| odyssey4me | and does the container fail to start, or do all the containers start but galera fails to successfully initiate a cluster? | 13:51 |
| kmARC | only one galera container has the problem (cannot connect to mysql) | 13:52 |
| kmARC | after removing the galera containers | 13:52 |
| kmARC | and running again, it works | 13:52 |
| kmARC | but then I got other errors. With swift, then the sysctl errors | 13:53 |
| *** karimb has quit IRC | 13:53 | |
| kmARC | do you think the specs of the machine is not enough for AIO? | 13:53 |
| kmARC | I `watch df`, it stays below sixty-something GBs | 13:53 |
| *** karimb has joined #openstack-ansible | 13:54 | |
| odyssey4me | no, those specs are fine | 13:54 |
| odyssey4me | that is odd | 13:54 |
| odyssey4me | is it a physical server, or a VM, or an instance? | 13:54 |
| kmARC | It's a vmware vm (vt-x passthru enabled) | 13:54 |
| kmARC | (I would've gone mad if I couldn't respawn this machine to start again from scratch :-) ) | 13:55 |
| odyssey4me | kmARC yep, several of us use cloud instances for testing | 13:56 |
| odyssey4me | the ubuntu image you're using as a base - is it some sort of build from the server cd or a pre-built image? | 13:57 |
| kmARC | nope, I installed ubuntu manually. Did a apt-get update&& dist-upgrade prior to start playing with OSA | 13:58 |
| odyssey4me | kmARC ok - so here's a fairly minimal starting package list from a base image in the CI infrastructure: http://logs.openstack.org/04/245704/2/check/gate-openstack-ansible-commit-nv/869b046/logs/instance-info/host_packages_info_1447751648.log | 13:58 |
| odyssey4me | perhaps something is in there that you're missing and is pertinent | 13:58 |
| pabelanger | odyssey4me: cloudnull: if you have some time today: https://review.openstack.org/#/c/246633/ just a reorg of JJB jobs | 13:58 |
| odyssey4me | note that not everything in there is absolutely required | 13:58 |
| odyssey4me | pabelanger morning :) | 13:59 |
| kmARC | okay, let me check | 13:59 |
| kmARC | (btw x11-common? ) | 14:00 |
| pabelanger | odyssey4me: morning | 14:00 |
| pabelanger | odyssey4me: I posted a question last night about idempotent and currently how this team is doing it, if at all. Do you have any scripts in place to help with that? | 14:01 |
| kmARC | odyssey4me: E: Unable to locate package hiera | 14:01 |
| kmARC | E: Unable to locate package puppetlabs-release | 14:01 |
| odyssey4me | pabelanger not yet - unfortunately we shell/command out quite a bit so idempotence isn't always possible | 14:01 |
| odyssey4me | pabelanger we'd like to get there, but it'll likely have to wait until after ansible 2.0 releases as the results will be more predictable | 14:02 |
| pabelanger | odyssey4me: okay, that is what I thought. I've started writing something in python to help, going to get some eyes on it from -infra but will also link here | 14:02 |
| odyssey4me | cool! | 14:02 |
| odyssey4me | kmARC those aren't likely required | 14:03 |
| evrardjp | pabelanger: cool indeed | 14:03 |
| kmARC | yeah, I filtered now puppet related packages out | 14:03 |
| evrardjp | this should be listed in best practices somewhere, because it's gonna be useful with independant roles repositories | 14:04 |
| evrardjp | I'm manually parsing the ansible output for changed in my roles :p | 14:04 |
| *** KLevenstein has joined #openstack-ansible | 14:05 | |
| pabelanger | odyssey4me: evrardjp: Ya, right now I am importing ansible.playbook.PlayBook and manually invoking playbook runs twice, first time I check for failures, next time I check if anything changed. Error out on any failure. It works well so far | 14:06 |
| evrardjp | unreadable sentence, I'll rephrase: I am currently parsing ansible output for changed different than 0, which is not really ideal. It would be cool to share this recipe for roles that will be included through ansible-requirements.yml | 14:06 |
| kmARC | odyssey4me: are you sure about this package list? Even if I add --no-install-(recommends|suggests) it wants to install stuff like emacs, java, etc. | 14:06 |
| pabelanger | I actually think we could move this to an external program, like ansible-lint in the future | 14:06 |
| evrardjp | I don't know ansible-lint, isn't it only a linter? or it's also something that can be useful for idempotency checks? | 14:08 |
| *** bapalm has joined #openstack-ansible | 14:08 | |
| pabelanger | evrardjp: no, it is a linter. I was just saying we could have a 2nd program to help with idempotent runs. | 14:11 |
| *** XiaBing has joined #openstack-ansible | 14:11 | |
| evrardjp | ok | 14:11 |
| pabelanger | over adding the file into tools/ folder | 14:11 |
| *** XiaBing has quit IRC | 14:12 | |
| odyssey4me | pabelanger afaik the usual way to test idempotency is simply to execute the playbook again, then monitor for anything that doesn't have only an 'OK' result | 14:12 |
| odyssey4me | eg: https://github.com/odyssey4me/ansible-openvas/blob/master/.travis.yml#L38-L41 | 14:13 |
| pabelanger | odyssey4me: Right, that's all I am doing. However, it gets tricky using tox.ini and bash to parse results | 14:13 |
| odyssey4me | ah yes, fair point | 14:14 |
| *** XiaBing has joined #openstack-ansible | 14:15 | |
| *** cemmason has quit IRC | 14:17 | |
| *** mgoddard__ has joined #openstack-ansible | 14:18 | |
| *** cemmason has joined #openstack-ansible | 14:18 | |
| *** hassaan_ has joined #openstack-ansible | 14:20 | |
| *** mgoddard_ has quit IRC | 14:20 | |
| kmARC | odyssey4me: Does kernel version matter? Your package list has 3.13, on my system it's 3.19 | 14:22 |
| *** hassaan has quit IRC | 14:23 | |
| odyssey4me | kmARC as long as it's 3.13.0-34 or later then it's fine | 14:24 |
| kmARC | yeah I thought so. Anyways, now I installed all the packages you have (excluding puppet, including emacs) | 14:25 |
| kmARC | maybe it will magically cure it | 14:25 |
| *** mgoddard__ has quit IRC | 14:28 | |
| *** mgoddard has joined #openstack-ansible | 14:29 | |
| kmARC | so ntp wasn't installed. And the bootstrap scripts didn't install it either. Can that be a problem? | 14:29 |
| *** openstackgerrit has quit IRC | 14:31 | |
| *** openstackgerrit has joined #openstack-ansible | 14:32 | |
| *** XiaBing has quit IRC | 14:36 | |
| kmARC | Nope. Still fails at galera_server | 14:38 |
| kmARC | :-\ | 14:38 |
| *** neilus has quit IRC | 14:39 | |
| *** bapalm has quit IRC | 14:43 | |
| *** bapalm has joined #openstack-ansible | 14:43 | |
| *** fawadkhaliq has joined #openstack-ansible | 14:44 | |
| cloudnull | ntt: did you get your question answered ? | 14:51 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:52 | |
| ntt | cloudnull: no | 14:52 |
| cloudnull | ntt: so ... CentOS support is not available at this time, we are presently only using Ubuntu. | 14:52 |
| cloudnull | However if youre keen on seeing CentOS supported we'd love to make it go | 14:53 |
| evrardjp | kmARC: first, for your safe, you should use the same ntp server across your hosts, that's gonna make your life easier | 14:53 |
| ntt | thanks. All my infrastructure is centos based, and I cannot use ubuntu | 14:53 |
| cloudnull | I did a fork for proof of concept https://github.com/cloudnull/os-ansible-deployment/tree/master-rhel to do RHEL/CentOS but its not been merged back into the main project as of yet | 14:54 |
| evrardjp | kmARC: then, could you check http://docs.openstack.org/developer/openstack-ansible/install-guide/configure-configurationintegrity.html ? | 14:54 |
| ntt | cloudnull: however I'm using https://github.com/pvaylov/openstack-ansible and I think the problem is that I don't know enough ansible | 14:55 |
| kmARC | evrardjp: I know, I use it everywhere. This is a one-host AIO build. Anyways, I think one of the bootstrap scripts should install ntp too | 14:55 |
| odyssey4me | cloudnull FYI https://etherpad.openstack.org/p/openstack-ansible-multi-os-support | 14:55 |
| evrardjp | kmARC: so your galera installation playbook fails? does it mention why? | 14:56 |
| ntt | cloudnull: thank you. Can I give a try from your repo? | 14:56 |
| cloudnull | that forks a bit old at this point however it "should" still work | 14:56 |
| kmARC | evradjp: | 14:56 |
| cloudnull | ntt: are you using "https://github.com/pvaylov/openstack-ansible/commits/kilo-centos" the "kilo-centos" branch ? | 14:56 |
| ntt | yes | 14:56 |
| cloudnull | cool | 14:57 |
| kmARC | failed: [aio1_galera_container-3b233cd0] => {"changed": true, "cmd": ["/usr/bin/mysqladmin", "--defaults-file=/etc/mysql/debian.cnf", "ping"] | 14:57 |
| kmARC | , "delta": "0:00:00.002966", "end": "2015-11-18 15:53:41.836941", "rc": 1, "start": "2015-11-18 15:53:41.833975", "stdout_lines": [], "warnin | 14:57 |
| kmARC | gs": []} | 14:57 |
| kmARC | ah sorry, will paste to pastebin in a sec | 14:57 |
| ntt | which version of openstack are you using in master-rhel? | 14:58 |
| *** mss has quit IRC | 14:58 | |
| kmARC | evradjp: http://paste.openstack.org/show/479262/ | 14:58 |
| cloudnull | ntt: it was targetting Liberty at that time | 14:59 |
| kmARC | this is my 5th try, it always fails. Now if i remove the galera containers and set them up again, it will work | 14:59 |
| kmARC | The interesting part is, only one galera container fails. Always | 14:59 |
| ntt | cloudnull & odyssey4me : I think that liberty is great, but using ansible with juno or at least kilo release would be great | 15:00 |
| kmARC | evrardjp: of course the syntax check is OK. | 15:01 |
| cloudnull | yea. I'd like to get multi-OS supported however I dont think it will get done for Kilo at this point. | 15:01 |
| cloudnull | however it'd be great to get pvaylov to commit some of the CentOS changes into our tree so the broader community can use them | 15:02 |
| ntt | The problem is related to upgrade. and in particular to the upgrade of the database structure. People that are using icehouse or juno should upgrade manually up to liberty :( | 15:02 |
| ntt | cloudnull: The service command supports only basic LSB actions (start, stop, restart, try-restart, reload, force-reload, status). For other actions, please try to use systemctl. | 15:04 |
| ntt | I'm trying your master-rhel branch | 15:05 |
| kmARC | evrardjp, odyssey4me: I am going crazy about this one-host installer. I am even willing to set-up a new Ubuntu host if needed, but I can't see why it's not working. | 15:05 |
| kmARC | Also, jenkins doesn't let my patch in because it fails on other point | 15:06 |
| cloudnull | for our project we've maintained a single release upgrade, so Icehouse > Juno > kilo | 15:06 |
| cloudnull | we're working on the Kilo > liberty upgrade | 15:06 |
| evrardjp | kmARC: I'll check your pastebin | 15:07 |
| cloudnull | ntt: was that stdout from a command within the branch? | 15:07 |
| ntt | yes | 15:08 |
| *** spotz_zzz is now known as spotz | 15:08 | |
| *** subscope has joined #openstack-ansible | 15:08 | |
| ntt | this error happens when I launch bootstrap-aio.sh | 15:08 |
| evrardjp | kmARC: if it's just a new host, I wouldn't spend too much time trying to fix it | 15:09 |
| cloudnull | ah | 15:09 |
| evrardjp | I'd do the teardown.sh and then re start all the playbooks. do you mind sending me your user_config? | 15:09 |
| kmARC | it's the default what build-aio.sh creates | 15:10 |
| kmARC | I didn't modify anything to it | 15:10 |
| evrardjp | kmARC: just to make sure everything is alright there, it's not because the syntax is correct that everything should be fine | 15:10 |
| evrardjp | oh | 15:10 |
| kmARC | basically, I must say, that the one line installer simply doesn't work :( | 15:10 |
| kmARC | I am still in the evaluation phase whether we should use this to deploy our new cluster, but I coldn't even make it work on a playground | 15:11 |
| evrardjp | are you running latest stable tag, or master? | 15:11 |
| evrardjp | (just as starter) | 15:11 |
| cloudnull | ntt: yea that last time I looked at that was 4mo ago and was done only as a POC so I'd imagine there are some crufty bits | 15:11 |
| kmARC | I tryes with master -> no luck. Now I'm trying with liberty | 15:12 |
| kmARC | still no luck | 15:12 |
| ntt | Ok .... I wait a solution for centos. In the meanwhile I will try with an ubuntu vm | 15:13 |
| cloudnull | We know it works on Ubuntu at this point. | 15:14 |
| evrardjp | kmARC: if you're retrying with liberty, don't forget to first cleanup everything | 15:15 |
| kmARC | evrardjp: I didn't clean up, I actually respawned my virgin ubuntu vm from a template (I am on vmware) | 15:15 |
| ntt | cloudnull ... ok, but I need to learn openstack-ansible and ubuntu is useful for this | 15:16 |
| cloudnull | ntt: we can pursue more CentOS support but we need more people to want to dev / maintain the distro support | 15:16 |
| *** cloudtrainme has joined #openstack-ansible | 15:16 | |
| *** alkari has joined #openstack-ansible | 15:16 | |
| kmARC | evrardjp: more to this: https://review.openstack.org/#/c/246840/ <- It doesn't run either on gerrit. | 15:16 |
| kmARC | Because of an unrelated issue | 15:16 |
| *** d34dh0r53 has quit IRC | 15:17 | |
| ntt | cloudnull: I don't have experience with ansible. I know openstack and if I can be helpful I'm here. Furthermore, I want to learn.... | 15:17 |
| *** d34dh0r53 has joined #openstack-ansible | 15:17 | |
| cloudnull | sweet! well if you have questions please reach out we're here :) | 15:18 |
| ntt | cloudnull: i'm trying to "debug" openstack ansible and I'm trying to reuse roles just for testing purpose | 15:18 |
| ntt | I have some doubts related to host definition... I don't understand where in the code it realizes host /groups definition | 15:19 |
| *** sdake has quit IRC | 15:20 | |
| cloudnull | ansible categorizes the hosts, which determined in inventory | 15:20 |
| *** sdake has joined #openstack-ansible | 15:20 | |
| cloudnull | **which is determined in inventory | 15:20 |
| ntt | in the inventory folder i see a dynamic_inventory.py .... | 15:21 |
| ntt | where this script is used and how? | 15:21 |
| cloudnull | we have the configs in /etc/openstack_deploy | 15:21 |
| evrardjp | kmARC: I'll check the docs you're following | 15:22 |
| cloudnull | `dynamic_inventory.py` sources bits from that dir to generate a json file | 15:22 |
| evrardjp | kmARC: you're following this right? http://docs.openstack.org/developer/openstack-ansible/developer-docs/quickstart-aio.html | 15:23 |
| evrardjp | where you've checkout liberty instead of kilo | 15:23 |
| *** sacharya has joined #openstack-ansible | 15:23 | |
| evrardjp | s/you've checkout/you've done a checkout of/ | 15:24 |
| kmARC | evrardjp: exactly, with liberty | 15:24 |
| cloudnull | ntt: we're controlling ansible through http://docs.ansible.com/ansible/intro_dynamic_inventory.html | 15:26 |
| ntt | ok. thanks. I need to study and I'm sorry if I'm asking stupid questions but I want to understand as fast as possible | 15:27 |
| *** jwagner_away is now known as jwagner | 15:27 | |
| cloudnull | not stupid questions at all | 15:28 |
| *** sacharya has quit IRC | 15:28 | |
| cloudnull | fire questions away as you find them | 15:28 |
| *** Mudpuppy has joined #openstack-ansible | 15:31 | |
| kmARC | evrardjp: have you ever tried this think with the liberty branch and Ubuntu 14.04.3? | 15:31 |
| mhayden | mornin' | 15:31 |
| evrardjp | kmARC: I'm not running AIO | 15:31 |
| evrardjp | but I could | 15:31 |
| kmARC | You think it's only the AIO what is problematic? | 15:32 |
| kmARC | should I go with three vms and try to set it up on a multi-node? | 15:32 |
| ntt | ok guys... thank you all. I have to go | 15:33 |
| *** ntt has quit IRC | 15:34 | |
| *** alkari has quit IRC | 15:35 | |
| cloudnull | have a good one ntt | 15:44 |
| *** mgoddard has quit IRC | 15:59 | |
| *** tiagogomes__ has joined #openstack-ansible | 16:01 | |
| *** mattt1 has quit IRC | 16:02 | |
| *** mattt1 has joined #openstack-ansible | 16:02 | |
| *** tiagogomes_ has quit IRC | 16:03 | |
| *** mattt1 is now known as mattt | 16:05 | |
| *** mss has joined #openstack-ansible | 16:09 | |
| *** david8hu has quit IRC | 16:16 | |
| *** mpavone has quit IRC | 16:20 | |
| *** fawadkhaliq has quit IRC | 16:20 | |
| *** oneswig has joined #openstack-ansible | 16:29 | |
| *** mgoddard has joined #openstack-ansible | 16:31 | |
| *** gparaskevas has quit IRC | 16:37 | |
| *** javeriak_ has quit IRC | 16:38 | |
| *** slotti has quit IRC | 16:40 | |
| *** vdo has quit IRC | 16:41 | |
| *** mgoddard_ has joined #openstack-ansible | 16:45 | |
| *** sacharya has joined #openstack-ansible | 16:45 | |
| *** mgoddard has quit IRC | 16:48 | |
| *** KLevenstein is now known as KLev-away | 16:48 | |
| *** fawadkhaliq has joined #openstack-ansible | 16:54 | |
| *** fawadkhaliq has quit IRC | 16:55 | |
| *** fawadkhaliq has joined #openstack-ansible | 16:55 | |
| *** alkari has joined #openstack-ansible | 16:57 | |
| *** phiche has quit IRC | 16:57 | |
| *** HimanshuGarg has quit IRC | 17:06 | |
| evrardjp | I'm off! | 17:10 |
| *** alkari has quit IRC | 17:11 | |
| cloudnull | have a good one evrardjp | 17:15 |
| *** HimanshuGarg has joined #openstack-ansible | 17:19 | |
| *** tiagogomes__ has quit IRC | 17:24 | |
| *** tiagogomes__ has joined #openstack-ansible | 17:25 | |
| *** tiagogomes__ has quit IRC | 17:32 | |
| *** oneswig has quit IRC | 17:33 | |
| *** phiche has joined #openstack-ansible | 17:34 | |
| *** bryan_att has joined #openstack-ansible | 17:34 | |
| bryan_att | hi, any ansible experts out there than can help with a support issue for using ansible to install an openstack service? | 17:35 |
| *** mgoddard has joined #openstack-ansible | 17:38 | |
| *** mgoddard_ has quit IRC | 17:38 | |
| *** fignew has joined #openstack-ansible | 17:41 | |
| *** jwagner is now known as jwagner_away | 17:43 | |
| cloudnull | bryan_att: using openstack-ansible? | 17:49 |
| *** tiagogomes has joined #openstack-ansible | 17:52 | |
| *** tiagogomes has quit IRC | 17:58 | |
| *** tiagogomes has joined #openstack-ansible | 17:59 | |
| *** javeriak has joined #openstack-ansible | 18:01 | |
| *** tiagogomes has quit IRC | 18:03 | |
| *** javeriak_ has joined #openstack-ansible | 18:07 | |
| *** sdake has quit IRC | 18:08 | |
| *** spotz is now known as spotz_zzz | 18:08 | |
| mgariepy | i'm having a small issue with ceph + kilo, when I try to start an instance ( boot from image create new volume) it fails, Block Device Mapping is Invalid: failed to get image <id> | 18:09 |
| mgariepy | any idea ? :S | 18:09 |
| bryan_att | cloudnull: I'm using ansible, not sure if that is different from openstack-ansible | 18:09 |
| mgariepy | i can create a volume from an image then boot from it fine. | 18:09 |
| bryan_att | cloudnull: I'm trying to install Congress using a playbook that vmware provided to me | 18:10 |
| *** javeria__ has joined #openstack-ansible | 18:10 | |
| *** javeriak has quit IRC | 18:11 | |
| *** HimanshuGarg has quit IRC | 18:11 | |
| *** javeriak_ has quit IRC | 18:12 | |
| bryan_att | cloudnull: the specific problem i am having is how to skip a step that has already been completed, when I re-run the playbook to get past errors, but the thing that was already done (creating an endpoint) fails the playbook... for now I have it commented out but need to know how to script a "skip if already done" step | 18:12 |
| *** sdake has joined #openstack-ansible | 18:13 | |
| *** oneswig has joined #openstack-ansible | 18:14 | |
| bgmccollum | bryan_att: openstack-ansible is a project composed of roles and playbooks for deploying openstack using ansible...but we can help with general openstack / ansible things | 18:14 |
| bryan_att | cloudnull: i'm using ansible-playbook 1.9.4, and a playbook that installs congress as noted. any idea how to skip already completed steps in a playbook? I am a newbie to ansible. | 18:15 |
| bgmccollum | bryan_att: sounds like that particular tasks isn't idempotent...what does the task look like? does the task have any tags? you can tell ansible to skip a particular tag...or tell it to start at a particular task (past the one you're having trouble with)... | 18:15 |
| bryan_att | i'll paste it | 18:15 |
| bryan_att | https://www.irccloud.com/pastebin/N6lzoa3X/ | 18:16 |
| bryan_att | bgmccollum: (commented out currently so I can get past it) | 18:16 |
| *** hassaan_ has quit IRC | 18:16 | |
| bgmccollum | so, the task isn't tagged which wont help you... | 18:17 |
| bgmccollum | you can do something like -- ansible-playbook playbook.yml --start-at-task="Name of the task where you want to start" | 18:17 |
| bgmccollum | so, the name of the task after the endpoint create task | 18:18 |
| bryan_att | bgmccollum: that's very useful to know, thanks. is the "tagged" feature something that would allow me to skip already complete steps (non-idempotent)? | 18:18 |
| *** oneswig has quit IRC | 18:18 | |
| bgmccollum | bryan_att: its a way of "tagging" tasks, and you can tell ansible to only run task of a particular tag, or skip tasks of a particular tag. | 18:19 |
| bgmccollum | generally: things run via the shell or command modules requires the command / script being run to bake idempotency into itself...ansible how now idea what the command / scripts does, and can't skip over things already done... | 18:20 |
| bryan_att | bgmccollum: ok, thanks. that may help me skip the steps a little more gracefully. | 18:20 |
| bgmccollum | bryan_att: id file a bug with vmware that subsequent runs of the playbooks fails on that task | 18:20 |
| bryan_att | bgmccollum: I may have to push a shell script that includes a bit more logic i guess to check to see if the endpoint is already created. | 18:21 |
| *** fawadkhaliq has quit IRC | 18:21 | |
| bryan_att | bgmccollum: I will - but this is not an officially published playbook, at least not by the congress projest | 18:21 |
| bgmccollum | when i must use shell / command modules, i run a subsequent task that drops a local fact (variable) that the task was done....then i can conditionally skip the tasks in the event the playbook was run again. | 18:22 |
| *** spotz_zzz is now known as spotz | 18:22 | |
| bgmccollum | much easier and generic enough than crafting smarter scripts | 18:23 |
| logan2 | mgariepy: are you still having problems with the BDM? | 18:24 |
| *** HimanshuGarg has joined #openstack-ansible | 18:26 | |
| bryan_att | bgmccollum: thanks, that seems a good way to go. I'll look into tagging and variable setting. | 18:26 |
| bgmccollum | bryan_att: here is an example using a local fact -- https://github.com/cloud-training/rpc-heat-ansible/blob/master/playbooks/rpc-11.0-playbook.yml#L392-L400 | 18:26 |
| bgmccollum | the next time the playbook is run, the local facts are read...and that task is skipped because the fact is defined | 18:27 |
| mgariepy | logan2, only when i try to boot from image create new vol. | 18:27 |
| *** KLev-away is now known as KLevenstein | 18:28 | |
| *** sdake has quit IRC | 18:28 | |
| *** HimanshuGarg has quit IRC | 18:30 | |
| bgmccollum | bryan_att: just noticed that task has -- when: init == "True" -- seems like you should be able to set init to False, in which that task should be skipped. grep around in your variables for init | 18:30 |
| *** spotz is now known as spotz_zzz | 18:32 | |
| bryan_att | bgmccollum: thanks, i wasn't sure what that was for but makes sense | 18:33 |
| *** cemmason has quit IRC | 18:33 | |
| *** cemmason has joined #openstack-ansible | 18:34 | |
| openstackgerrit | Merged openstack/openstack-ansible: Use 'liberty' branch for packages when using one-liner aio build script https://review.openstack.org/246840 | 18:38 |
| *** oneswig has joined #openstack-ansible | 18:40 | |
| cloudnull | does anyone know the person who owns https://github.com/pvaylov/openstack-ansible/tree/kilo-centos ? | 18:42 |
| *** HimanshuGarg has joined #openstack-ansible | 18:44 | |
| stevelle | kmARC: congrats, your fix is merged | 18:45 |
| *** cloudtrainme has quit IRC | 18:47 | |
| bgmccollum | cloudnull: its a fork of https://github.com/achekunov/openstack-ansible, which has references to fuel... | 18:49 |
| cloudnull | yea i see https://github.com/achekunov/openstack-ansible/commit/4eae91cb076d435193125785d8850e00bef2b232 and that there's a fuel repo in the user account but no contact or update details | 18:50 |
| bgmccollum | cloudnull : Alexey Chekunov <achekunov@mirantis.com> | 18:50 |
| bgmccollum | git log | 18:51 |
| *** spotz_zzz is now known as spotz | 18:51 | |
| bgmccollum | in the kilo-my branch | 18:51 |
| cloudnull | well thats why you get paid the big bucks | 18:51 |
| cloudnull | :) | 18:52 |
| bgmccollum | :) | 18:52 |
| *** oneswig has quit IRC | 18:53 | |
| *** harlowja has quit IRC | 19:05 | |
| *** Reddy has quit IRC | 19:06 | |
| *** Bjoern_ has joined #openstack-ansible | 19:07 | |
| *** harlowja has joined #openstack-ansible | 19:08 | |
| *** javeriak has joined #openstack-ansible | 19:12 | |
| *** javeria__ has quit IRC | 19:14 | |
| meteorfox | hi, quick question, I want to set the cinder volume_clear variable to none, so it doesn't secure delete the volumes. do I just add 'volume_clear: none' to each lvm block in the openstack_user_config.yml, or is there a better way? | 19:14 |
| *** javeriak_ has joined #openstack-ansible | 19:15 | |
| meteorfox | cloudnull: ^^ | 19:16 |
| *** javeriak has quit IRC | 19:17 | |
| *** cemmason has quit IRC | 19:20 | |
| *** cemmason has joined #openstack-ansible | 19:20 | |
| *** cloudtrainme has joined #openstack-ansible | 19:21 | |
| *** bapalm has quit IRC | 19:23 | |
| *** javeriak has joined #openstack-ansible | 19:25 | |
| *** bapalm has joined #openstack-ansible | 19:26 | |
| *** jwagner_away is now known as jwagner | 19:27 | |
| *** javeriak_ has quit IRC | 19:28 | |
| *** oneswig has joined #openstack-ansible | 19:39 | |
| *** oneswig has quit IRC | 19:43 | |
| *** subscope has quit IRC | 20:06 | |
| *** javeriak_ has joined #openstack-ansible | 20:06 | |
| *** subscope has joined #openstack-ansible | 20:06 | |
| *** mss_ has joined #openstack-ansible | 20:06 | |
| *** javeriak has quit IRC | 20:08 | |
| *** mss has quit IRC | 20:09 | |
| *** subscope has quit IRC | 20:09 | |
| *** subscope has joined #openstack-ansible | 20:10 | |
| *** KLevenstein has quit IRC | 20:10 | |
| *** subscope has quit IRC | 20:17 | |
| *** javeriak has joined #openstack-ansible | 20:18 | |
| *** KLevenstein has joined #openstack-ansible | 20:21 | |
| *** javeriak_ has quit IRC | 20:21 | |
| *** harlowja has quit IRC | 20:36 | |
| *** oneswig has joined #openstack-ansible | 20:39 | |
| *** harlowja has joined #openstack-ansible | 20:41 | |
| *** oneswig has quit IRC | 20:42 | |
| *** oneswig has joined #openstack-ansible | 20:44 | |
| *** oneswig has quit IRC | 20:48 | |
| *** javeriak_ has joined #openstack-ansible | 20:50 | |
| *** javeriak has quit IRC | 20:52 | |
| *** rebase has joined #openstack-ansible | 20:53 | |
| openstackgerrit | Merged openstack/openstack-ansible: fix typo in dynamic_inventory.py https://review.openstack.org/246395 | 20:54 |
| *** openstack has joined #openstack-ansible | 21:06 | |
| *** sdake has joined #openstack-ansible | 21:08 | |
| openstackgerrit | Javeria Khan proposed openstack/openstack-ansible: Update neutron service handler to restart only enabled services https://review.openstack.org/245825 | 21:23 |
| *** oneswig has joined #openstack-ansible | 21:29 | |
| *** woodard has quit IRC | 21:32 | |
| *** oneswig has quit IRC | 21:34 | |
| *** oneswig has joined #openstack-ansible | 21:36 | |
| *** javeriak_ has quit IRC | 21:42 | |
| *** elo has quit IRC | 21:44 | |
| *** HimanshuGarg has quit IRC | 21:49 | |
| *** aslaen has joined #openstack-ansible | 21:51 | |
| *** jwagner is now known as jwagner_away | 21:51 | |
| *** HimanshuGarg has joined #openstack-ansible | 22:01 | |
| *** aslaen has quit IRC | 22:02 | |
| *** KLevenstein has quit IRC | 22:05 | |
| *** KLevenstein has joined #openstack-ansible | 22:06 | |
| *** oneswig has quit IRC | 22:10 | |
| *** Mudpuppy has quit IRC | 22:14 | |
| *** jaybeers has joined #openstack-ansible | 22:17 | |
| *** Bjoern_ is now known as Bjoern_zZzZzZzZ | 22:29 | |
| bgmccollum | after a juno -> kilo upgrade, there is a stale nova service entry for the old nova_spice_console_container. is it worth adding a post upgrade cleanup task? | 22:33 |
| *** sacharya has quit IRC | 22:38 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 22:53 | |
| *** oneswig has joined #openstack-ansible | 22:54 | |
| *** oneswig has quit IRC | 22:58 | |
| *** sdake_ has joined #openstack-ansible | 23:17 | |
| *** sdake has quit IRC | 23:18 | |
| *** cloudtrainme has quit IRC | 23:19 | |
| *** Bjoern_zZzZzZzZ is now known as Bjoern_ | 23:19 | |
| *** KLevenstein has quit IRC | 23:29 | |
| *** karimb_ has joined #openstack-ansible | 23:35 | |
| *** karimb has quit IRC | 23:37 | |
| *** sdake_ has quit IRC | 23:40 | |
| *** sdake has joined #openstack-ansible | 23:40 | |
| *** karimb_ has quit IRC | 23:43 | |
| *** spotz is now known as spotz_zzz | 23:46 | |
| *** Bjoern_ has quit IRC | 23:51 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!