| *** yaya has quit IRC | 00:08 | |
| *** darrenc_afk is now known as darrenc | 00:11 | |
| *** jaypipes has quit IRC | 00:18 | |
| *** alop has quit IRC | 00:55 | |
| *** tlian has quit IRC | 01:06 | |
| *** sdake_ has quit IRC | 01:15 | |
| *** sdake has joined #openstack-ansible | 01:16 | |
| *** pellaeon has quit IRC | 01:35 | |
| *** pellaeon has joined #openstack-ansible | 01:35 | |
| *** JRobinson__ has joined #openstack-ansible | 01:36 | |
| *** galstrom_zzz is now known as galstrom | 01:42 | |
| openstackgerrit | David Stanek proposed stackforge/os-ansible-deployment: Uses tox for automating documentation builds https://review.openstack.org/207577 | 01:52 |
|---|---|---|
| *** galstrom is now known as galstrom_zzz | 02:35 | |
| *** sdake has quit IRC | 02:48 | |
| *** sdake has joined #openstack-ansible | 03:16 | |
| *** sdake_ has joined #openstack-ansible | 03:38 | |
| *** sdake has quit IRC | 03:41 | |
| *** JRobinson__ is now known as JRobinson__afk | 03:51 | |
| *** galstrom_zzz is now known as galstrom | 04:11 | |
| *** daneyon has quit IRC | 04:17 | |
| *** daneyon has joined #openstack-ansible | 04:17 | |
| *** galstrom is now known as galstrom_zzz | 04:52 | |
| *** JRobinson__afk is now known as JRobinson__ | 05:06 | |
| *** fawadkhaliq has joined #openstack-ansible | 05:14 | |
| *** fawadk has joined #openstack-ansible | 05:15 | |
| *** fawadkhaliq has quit IRC | 05:18 | |
| openstackgerrit | Steve Lewis proposed stackforge/os-ansible-deployment: Updated master to the latest SHAs - 03.Aug.2015 https://review.openstack.org/208801 | 05:22 |
| *** javeriak has joined #openstack-ansible | 06:04 | |
| *** JRobinson__ has quit IRC | 06:07 | |
| openstackgerrit | Steve Lewis proposed stackforge/os-ansible-deployment: Add support for swift proxy config: sorting_method https://review.openstack.org/208817 | 06:16 |
| *** daneyon has quit IRC | 06:30 | |
| *** javeriak has quit IRC | 06:38 | |
| *** javeriak has joined #openstack-ansible | 06:38 | |
| svg | Good morning all o/ | 06:53 |
| stevelle | o/ | 07:02 |
| *** gparaskevas has joined #openstack-ansible | 07:07 | |
| *** fawadkhaliq has joined #openstack-ansible | 07:12 | |
| *** javeriak_ has joined #openstack-ansible | 07:14 | |
| *** javeriak has quit IRC | 07:14 | |
| *** fawadk has quit IRC | 07:14 | |
| *** fawadk has joined #openstack-ansible | 07:15 | |
| *** javeriak has joined #openstack-ansible | 07:17 | |
| *** fawadkhaliq has quit IRC | 07:18 | |
| *** javeriak_ has quit IRC | 07:20 | |
| *** javeriak has quit IRC | 07:24 | |
| *** javeriak has joined #openstack-ansible | 07:24 | |
| *** javeriak_ has joined #openstack-ansible | 07:27 | |
| *** javeriak has quit IRC | 07:28 | |
| odyssey4me | o/ svg | 07:35 |
| odyssey4me | stevelle you still awake? | 07:35 |
| *** fawadk has quit IRC | 07:45 | |
| svg | to make a temporary test kilo versionm right now, with the ceph patch in, should I use kilo HEAD or 11.0.4? I'm inclined to take head, unless someone has better advice? | 07:52 |
| *** fawadkhaliq has joined #openstack-ansible | 07:57 | |
| mattt | svg: assuming you're using this for testing i'd go head, it may not apply cleanly to 11.0.4 | 07:58 |
| svg | thx | 07:59 |
| mattt | you're welcome, and good morning to you! | 07:59 |
| svg | same, same | 07:59 |
| * svg has a hard time waking up todat | 07:59 | |
| evrardjp | hello | 08:01 |
| evrardjp | svg, I'm going to do the same this afternoon | 08:02 |
| evrardjp | quick question: for kilo branch we are following upstream kilo stable branch right? | 08:03 |
| evrardjp | so there should be a merge soon with the new kilo release, right? | 08:03 |
| *** fawadkhaliq has quit IRC | 08:06 | |
| odyssey4me | evrardjp yes, and yes - once we tag 11.1.0 there'll be a follow-on merge to update the sha's from stable/kilo | 08:10 |
| odyssey4me | we'll then continue to tag a hotfix version every two weeks, after which we update the sha's every time | 08:11 |
| mattt | shas were last bumped 23/07, so quite recent | 08:12 |
| odyssey4me | bgmccollum blast, that explains the behaviour cc gparaskevas | 08:20 |
| *** javeriak_ has quit IRC | 08:25 | |
| *** javeriak has joined #openstack-ansible | 08:26 | |
| *** javeriak has quit IRC | 08:26 | |
| *** javeriak has joined #openstack-ansible | 08:34 | |
| *** jwitko has quit IRC | 08:35 | |
| *** jwitko has joined #openstack-ansible | 08:37 | |
| *** javeriak_ has joined #openstack-ansible | 08:38 | |
| *** javeriak has quit IRC | 08:38 | |
| gparaskevas | odyssey4me: Hey goodmmorning | 08:43 |
| gparaskevas | odyssey4me: what happend? | 08:43 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated master for new dev work - 27.07.2015 https://review.openstack.org/199126 | 08:54 |
| odyssey4me | gparaskevas bgmccollum provided f3eedback that using git review with HTTP passwords requires git-review 1.25 or newer | 09:03 |
| gparaskevas | odyssey4me: ok but what about using ssh? | 09:03 |
| odyssey4me | gparaskevas using ssh is built-in, but you need to find a way to get through your firewall on the required port | 09:03 |
| odyssey4me | use an ssh tunnel or something | 09:03 |
| *** javeriak has joined #openstack-ansible | 09:19 | |
| gparaskevas | odyssey4me: i Have a quick question regarding python beaver | 09:19 |
| odyssey4me | gparaskevas best is usually to ask, not ask to ask :) | 09:20 |
| gparaskevas | odyssey4me: i have tried to evaluate rpc-openstack but unfortunalty all many pip installs cannot be completed inclding pthon-beaver it self | 09:20 |
| gparaskevas | odyssey4me: i can see that pip is redirected to local repository which does not have all the packages. | 09:21 |
| *** javeriak_ has quit IRC | 09:22 | |
| gparaskevas | odyssey4me: why is that happening and how cna i overcome? | 09:22 |
| *** javeriak has quit IRC | 09:29 | |
| *** javeriak has joined #openstack-ansible | 09:29 | |
| odyssey4me | gparaskevas note that the version of rpc you're trying to evaluate has not yet released any documentation to describe how to install it yet - that's because RPC11 is not yet released | 09:31 |
| *** fawadkhaliq has joined #openstack-ansible | 09:31 | |
| gparaskevas | odyssey4me: i am using this: https://github.com/rcbops/rpc-openstack/tree/kilo BTW anyway I have created my own playbooks for ELK and Nagios that extend OSAD but i was using syslog and not beaver. Now i want use beaver and i cannot install it because of pip configuration | 09:33 |
| gparaskevas | odyssey4me: is there a way to overcome this? | 09:33 |
| odyssey4me | gparaskevas you have to rebuild the repo, but set the base that it searches from to be a base that includes both the rpc and osad playbooks | 09:34 |
| gparaskevas | odyssey4me: ok how can i do that i am not very famailiar with pip :/ | 09:34 |
| odyssey4me | you'll notice that rpc-openstack has this play to build the repo for the rpc packages: https://github.com/rcbops/rpc-openstack/blob/master/rpcd/playbooks/repo-build.yml | 09:34 |
| gparaskevas | yes | 09:35 |
| odyssey4me | you just need to cd to the rpc-openstack playbooks directory, then execute: openstack-ansible repo-build.yml | 09:35 |
| odyssey4me | once you've done that, then execute the beaver.yml playbook | 09:36 |
| gparaskevas | ok i have seen that le me try. thanks again! | 09:36 |
| odyssey4me | oh, sorry - after the repo-build you need to add the rpc package repo to the hosts and containers, so you have to execute the repo-pip-setup.yml playbook | 09:36 |
| odyssey4me | (it's been a while since I worked on it - sorry about the mixup) | 09:37 |
| gparaskevas | ok so both? | 09:37 |
| odyssey4me | repo-build, then repo-pip-setup, then beaver | 09:37 |
| gparaskevas | ok is there any interest in nagios for OSAD? I can see that you try to keep it clean and OS specific but since i like to contribute i cannot force my self not to ask | 09:38 |
| mattt | gparaskevas: sec, i saw a blog post on this | 09:43 |
| gparaskevas | mattt: this one? https://developer.rackspace.com/blog/openstack-osad-and-nagios-against-the-world/ | 09:44 |
| mattt | yep :) | 09:44 |
| gparaskevas | mattt: :P | 09:44 |
| mattt | i love nagios, so i'm all for this | 09:44 |
| mattt | but not had time to try it tbh :( | 09:44 |
| gparaskevas | mattt: haha i have seen it but i have not tried it i have created my own. i can share if you like | 09:44 |
| *** javeriak has quit IRC | 09:45 | |
| mattt | gparaskevas: yeah would love to see what yoiu have | 09:48 |
| gparaskevas | mattt: ok | 09:48 |
| odyssey4me | gparaskevas so the starting point would be to submit a spec to the project, proposing how you would integrate it and how it would affect the playbooks, roles, etc | 09:49 |
| gparaskevas | odyssey4me: ok let me see how to submit a spec | 09:49 |
| mattt | i'm not sure nagios config shoudl live in os-a-d personally | 09:52 |
| mattt | i thought you had a separate git repo :) | 09:53 |
| *** openstackgerrit_ has joined #openstack-ansible | 09:53 | |
| *** javeriak has joined #openstack-ansible | 09:54 | |
| mattt | ^^^ gparaskevas | 09:54 |
| gparaskevas | i have a separate repository like rpc-openstack | 09:54 |
| gparaskevas | so i guess there is no problem | 09:55 |
| mattt | gparaskevas: it'd be interesting to hear what the community thinks, if they want a reference monitoring implementation inside osad then why not | 09:55 |
| mattt | but my gut feeling would be that most people would want it kept separate | 09:55 |
| gparaskevas | mattt: yes this is what i feel also. i any case they can be stored and maintained seperatly but they can see eachother in every installation :P | 09:56 |
| mattt | cool that sounds ideal to me :) | 09:57 |
| evrardjp | mattt: FYI, I don't want monitoring in my OSAD, but I don't mind having some monitoring practices in the doc... | 10:18 |
| evrardjp | monitoring is really a specific job, done differently in each enterprise | 10:19 |
| git-harry | Yeah, this project needs to be careful of feature creep | 10:20 |
| mattt | nagios does seem to be a popular choice tho, it would be good to have an external repo that plugs everything together as far as nagios is concerned that the community can hack on | 10:22 |
| evrardjp | using scripts for checking health is a kinda standard way of doing it | 10:24 |
| evrardjp | so I agree, nagios in an external repo would be a good idea... | 10:25 |
| evrardjp | Zabbix is also quite popular | 10:25 |
| mattt | can zabbix use nagios lugins? | 10:26 |
| mattt | *plugins | 10:26 |
| evrardjp | not really, but nagios can start commands and return response codes for giving status... Zabbix can do the same | 10:32 |
| evrardjp | so one script for status checking can be used for both | 10:32 |
| *** fawadk has joined #openstack-ansible | 10:35 | |
| odyssey4me | mattt yes, through a wrapper | 10:36 |
| odyssey4me | but yeah, as we grow I think the strategy of being more modular is going to become more important over time | 10:37 |
| *** fawadkhaliq has quit IRC | 10:37 | |
| odyssey4me | so it's going to be more and more important to have the roles be independant (able to run on their own to achieve the goal of installing/configuring the service), and the playbooks be more opinionated | 10:38 |
| *** fawadkhaliq has joined #openstack-ansible | 10:38 | |
| odyssey4me | gparaskevas a few principles would be useful to bear in mind when preparing your blueprint | 10:39 |
| odyssey4me | 1 - don't expect the role to be imported into the os-ansible-deployment repository... expect instead to make use of the facility in bootstrap-ansible to import any external roles you need and perhaps only expect to include some playbooks in-repo, or perhaps some docs describing how to make use of the external roles. | 10:40 |
| odyssey4me | heh, I shoved it all in one principle :p | 10:40 |
| gparaskevas | odyssey4me: yes ofcource. If the playbooks doesnt need to be merged with OSAD which is clear that is shouldnt then blueprint isnt still nessesery. I will submit my repository which works on a modular type like rpc-openstack if i may | 10:41 |
| *** fawadk has quit IRC | 10:43 | |
| odyssey4me | gparaskevas sure - what you may consider doing is adding some docs to osad to reference what you've done so that others can learn from it | 10:46 |
| stevelle | odyssey4me: awake for a bit, whats news? | 11:02 |
| evrardjp | stevelle: how many hours do you sleep a day/a week? It's like I've seen you a few hours before... | 11:03 |
| stevelle | evrardjp: I feel like not as many as I would like :) | 11:04 |
| stevelle | also realizing it's lunch hour in western europe right now so might not get an answer | 11:06 |
| odyssey4me | stevelle I'm working a two-pronged attack for the sha updates | 11:10 |
| odyssey4me | one test is working to get updated to at least liberty-2, the other to the latest sha's | 11:10 |
| *** javeriak has quit IRC | 11:11 | |
| stevelle | I did latest shas last night and had two successful runs before I submitted a patch... which has failed twice now | 11:11 |
| odyssey4me | the issue on the latest sha's appears to be in keystone one of the oslo_messaging imports fails | 11:11 |
| stevelle | I know at least one of my successful runs had a different sha for heat, but I thought keystone was the same. | 11:12 |
| odyssey4me | stevelle http://paste.openstack.org/show/u3mIrPjsUWANe8ZJ0ksy/ | 11:12 |
| odyssey4me | stevelle also note that in the previous sha update attempt on master, cloudnull_afk had some extra bits: https://review.openstack.org/#/c/199126/ | 11:13 |
| stevelle | well I am no longer even remotely hopeful for my patch | 11:15 |
| odyssey4me | stevelle no worries, we'll work through it :) you should try to get some more rest | 11:16 |
| stevelle | odyssey4me: I look forward to checking back in a few hours to see where things are | 11:19 |
| *** openstackgerrit has quit IRC | 11:31 | |
| *** openstackgerrit_ is now known as openstackgerrit | 11:31 | |
| *** openstackgerrit has quit IRC | 11:32 | |
| *** openstackgerrit_ has joined #openstack-ansible | 11:32 | |
| *** openstackgerrit_ is now known as openstackgerrit | 11:32 | |
| *** openstackgerrit_ has joined #openstack-ansible | 11:33 | |
| *** sdake_ has quit IRC | 11:47 | |
| *** fawadkhaliq has quit IRC | 11:48 | |
| *** jaypipes has joined #openstack-ansible | 11:59 | |
| *** woodard has joined #openstack-ansible | 12:05 | |
| *** openstackgerrit has quit IRC | 12:16 | |
| *** openstackgerrit has joined #openstack-ansible | 12:17 | |
| *** fawadkhaliq has joined #openstack-ansible | 12:23 | |
| *** bapalm_ has joined #openstack-ansible | 12:34 | |
| *** sdake has joined #openstack-ansible | 12:36 | |
| *** pilgrimstack has joined #openstack-ansible | 12:41 | |
| *** pilgrimstack1 has quit IRC | 12:42 | |
| mattt | how do you rebootstrap your galera cluster again? | 12:54 |
| *** tlian has joined #openstack-ansible | 12:54 | |
| mattt | n/m says in the error :P | 12:54 |
| git-harry | you're welcome | 12:55 |
| mattt | git-harry: this is why they pay you the big bucks | 12:55 |
| gparaskevas | i ould love to here that :P | 12:55 |
| gparaskevas | would* | 12:55 |
| gparaskevas | i am having issues my self with galera | 12:56 |
| gparaskevas | i usualy do that: mysqld --tc-heuristic-recover=ROLLBACK if nodes cannot come up... | 12:56 |
| *** jaypipes has quit IRC | 12:57 | |
| mattt | gparaskevas: openstack-ansible -e galera_ignore_cluster_state=true galera-install.yml | 12:58 |
| mattt | cannot confirm or deny that it's working tho :) | 12:58 |
| *** jaypipes has joined #openstack-ansible | 13:00 | |
| *** gparaskevas has quit IRC | 13:01 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Improve Keystone Apache configuration https://review.openstack.org/209047 | 13:02 |
| odyssey4me | stevelle so https://review.openstack.org/209047 fixes the keystone import errors, now to move onto the next issue :) | 13:02 |
| evrardjp | I'd like to have multiple backend drivers for my storage (NetApp, LVM and I'd like to test the ceph integration)... Is it possible to have multiple backends under cinder_backends, or should I create a cinder_backend per container, and multiply the number of containers on the same host? | 13:12 |
| *** TheIntern has joined #openstack-ansible | 13:14 | |
| odyssey4me | evrardjp it's already built-in: http://osad.readthedocs.org/en/latest/install-guide/configure-cinder.html | 13:18 |
| evrardjp | it's not what I meant | 13:19 |
| evrardjp | but I'll use this page to be understable | 13:20 |
| evrardjp | On the point 9) on that documentation | 13:20 |
| evrardjp | there is netapp: and then the netapp directives | 13:20 |
| evrardjp | is it possible to put, on the same level as netapp, lvm: ? | 13:21 |
| evrardjp | at the same time | 13:21 |
| *** KLevenstein has joined #openstack-ansible | 13:21 | |
| *** KLevenstein is now known as KLev-away | 13:21 | |
| evrardjp | I think everything will be confused, right? | 13:21 |
| mattt | evrardjp: you can use both | 13:22 |
| mattt | evrardjp: i've just done that on an API so i can use lvm and ceph | 13:22 |
| evrardjp | in the same container? | 13:22 |
| evrardjp | so they are gonna be in the same avaiability zone I guess | 13:22 |
| evrardjp | I'll try and see how it's gonna play out | 13:23 |
| mattt | i had a single cinder-volume running on the aio instance itself | 13:23 |
| mattt | i'm not sure from a cinder perspective what is ideal but the playbooks seem to support this at least | 13:24 |
| odyssey4me | evrardjp exactly :) | 13:24 |
| odyssey4me | evrardjp I'm not sure about the AZ, I'd have to dig into the template/playbook to verify that | 13:24 |
| evrardjp | odyssey4me: the exactly is for "let's try and see" I guess, not the availability zone? | 13:25 |
| evrardjp | right. | 13:25 |
| evrardjp | :D | 13:25 |
| evrardjp | I'll look later don't worry about that | 13:25 |
| odyssey4me | evrardjp exactly is for the nfs being at the same level as netapp | 13:26 |
| evrardjp | so it's definitely possible to have multiple backends on the same container... | 13:27 |
| evrardjp | thanks for the info | 13:27 |
| odyssey4me | evrardjp yup | 13:27 |
| odyssey4me | evrardjp lvm is the only special case which requires the volume service to be on the same server as the storage | 13:28 |
| odyssey4me | all other cinder backends can share a volume service (and therefore the same volume service container) | 13:29 |
| evrardjp | that makes sense odyssey4me | 13:29 |
| Sam-I-Am | whats with reviews taking so long? | 13:35 |
| Sam-I-Am | https://review.openstack.org/#/c/204597/ | 13:35 |
| Sam-I-Am | still waiting | 13:35 |
| mattt | Sam-I-Am: a good # of cores are out at the moment | 13:38 |
| *** Mudpuppy has joined #openstack-ansible | 13:39 | |
| Sam-I-Am | mattt: this is a problem | 13:39 |
| mattt | Sam-I-Am: why? | 13:40 |
| *** Mudpuppy has quit IRC | 13:40 | |
| Sam-I-Am | because we cant get stuff done? | 13:40 |
| mattt | sure we can | 13:40 |
| mattt | urgent stuff will go through as normal | 13:40 |
| *** Mudpuppy has joined #openstack-ansible | 13:41 | |
| mattt | Sam-I-Am: is that review blocking anything at the minute? | 13:41 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Re-order setup-openstack play to match the gate check order https://review.openstack.org/209071 | 13:44 |
| Sam-I-Am | its not blocking anything | 13:44 |
| Sam-I-Am | things just tend to get stale though | 13:44 |
| odyssey4me | Sam-I-Am it's a pretty big change into something most of the team are not good at (neutron), and most of the cores are away | 13:45 |
| odyssey4me | if you could get some hardware tests done by the guys who know neutron well then there'd be more confidence in core votes | 13:45 |
| odyssey4me | also note that the bug is only targeted for 11.1.1, so there's no rush - the attention is elsewhere at the moment | 13:45 |
| Sam-I-Am | odyssey4me: how does one go about getting hardware tests done? the multi-node gate has been hosed for a while. | 13:46 |
| odyssey4me | Sam-I-Am have a chat to Apsu and rackertom to see if they can spend some cycles on it - also d34dh0r53 regularly works in the hardware labs | 13:47 |
| *** javeriak has joined #openstack-ansible | 13:47 | |
| odyssey4me | I'm not sure if svg or evrardjp have some sort of test infrastructure and are willing to test it out too. I like the patch as it cleans things up - I just have no idea whether it causes any issues. | 13:48 |
| evrardjp | what should I test? | 13:48 |
| Sam-I-Am | evrardjp: https://review.openstack.org/#/c/204597/ | 13:49 |
| Sam-I-Am | Apsu rackertom d34dh0r53 y'all around? | 13:49 |
| evrardjp | I have a test infrastructure, but the resource missing is not hardware, it's time ;) | 13:49 |
| odyssey4me | evrardjp yep, we all suffer from the lack of that resource :) | 13:50 |
| evrardjp | I know :) | 13:50 |
| Sam-I-Am | sleep is optional | 13:50 |
| Sam-I-Am | you can donate the hardware to me :) | 13:50 |
| evrardjp | I'll cherry pick into kilo Sam-I-Am and restart my playbooks soon | 13:50 |
| odyssey4me | thanks evrardjp :) | 13:51 |
| evrardjp | put the words in a different order and you get what I meant :p | 13:51 |
| evrardjp | just finishing multiple backends testing first | 13:51 |
| rackertom | Sam-I-Am: m'yes | 13:53 |
| Sam-I-Am | evrardjp: thx | 13:56 |
| Sam-I-Am | rackertom: whatchoo think of the neutron patch? | 13:56 |
| *** sigmavirus24_awa is now known as sigmavirus24 | 14:00 | |
| rackertom | Sam-I-Am: I thought I +1 it already. Saw a new patch was added, so re-+1. | 14:00 |
| Sam-I-Am | rackertom: aight thanks. thats sort of the fun of rebasing :/ | 14:01 |
| rackertom | I really liked the grouping, and obviously conditionally including options that only need to be on a server is good stuff. | 14:02 |
| rackertom | Makes the rendered templates look more like what someone would expect if following the docs. | 14:03 |
| odyssey4me | rackertom the concern I have is whether we're introducing any regressions with it, or perhaps new bugs... so ideally I'd like to see someone provide feedback of more thorough testing | 14:04 |
| Sam-I-Am | its neutron, there's always new bugs :) | 14:05 |
| Sam-I-Am | i've tested the neutron bits on my own multi-node labs | 14:05 |
| Sam-I-Am | just not embededed in ansible | 14:05 |
| Sam-I-Am | however, the configs look the same | 14:05 |
| rackertom | odyssey4me: I absolutely agree. Though, again, client nodes should just be ignoring server only options, and moving options within a file wouldn't matter either, I totally think having a good lab setup for this would be a great help. | 14:06 |
| rackertom | Every bit of my attempts at lab work has ended in smouldering piles of rubble, so maybe when I get this sha thing sorted, I could try to find hosts to test this on. | 14:08 |
| *** TheIntern has quit IRC | 14:08 | |
| evrardjp | odyssey4me: I'm sorry but at the end I won't test the multiple backends directly... I want to test multiple availability zones so... one container per AZ it is... | 14:10 |
| *** spotz_zzz is now known as spotz | 14:11 | |
| *** javeriak has quit IRC | 14:12 | |
| Sam-I-Am | rackertom: whats up with your lab? | 14:12 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Updated master to the latest SHAs - 03.Aug.2015 https://review.openstack.org/208801 | 14:13 |
| Apsu | Morning. | 14:18 |
| Sam-I-Am | moo | 14:18 |
| Apsu | odyssey4me: Sam-I-Am: Sup? | 14:18 |
| rackertom | Sam-I-Am: my physical labs never fully work. | 14:19 |
| *** phalmos has joined #openstack-ansible | 14:20 | |
| Sam-I-Am | Apsu: looks for love on https://review.openstack.org/#/c/204597/ | 14:20 |
| Sam-I-Am | er, looking | 14:20 |
| Sam-I-Am | ideally, somewhere to multi-node test it | 14:20 |
| Apsu | Sam-I-Am: Ok, so why haven't you run it through our jenkins lab? | 14:21 |
| Sam-I-Am | Apsu: no acces | 14:22 |
| Sam-I-Am | remember, i'm in docs :) | 14:22 |
| Apsu | I remember that's what you tell everyone | 14:22 |
| * Apsu stares at review you Just linked | 14:23 | |
| *** TheIntern has joined #openstack-ansible | 14:24 | |
| Apsu | Sam-I-Am: Also, I thought those jenkins multi-node jobs were nightly and such | 14:25 |
| Sam-I-Am | i dont know what happened to the 'wherenow' ci bits | 14:25 |
| Apsu | Has the beautiful Jenkins I built been asploded?! | 14:25 |
| Sam-I-Am | i think you were part of that business | 14:25 |
| Sam-I-Am | i havent seen a wherenow appendage on reviews in a while | 14:25 |
| Sam-I-Am | maybe the hardware flooded or caught fire? | 14:25 |
| Apsu | Both/and. | 14:26 |
| palendae | Apsu: You'd have to ask hughsaunders, but I'm pretty sure the multi-node jobs were removed from per-commit checking | 14:26 |
| palendae | Because they weren't voting and were failing a lot | 14:26 |
| Apsu | lol | 14:26 |
| Apsu | Super useful, then. k | 14:26 |
| palendae | Yep | 14:27 |
| odyssey4me | Apsu the multi-node labs are dead at this time, and have been for a few months | 14:27 |
| odyssey4me | hughsaunders did a rage quit on jenkins :p | 14:27 |
| odyssey4me | well, actually at stupid hardware issues | 14:28 |
| mattt | dammit hughsaunders | 14:28 |
| Apsu | That's a paddlin' | 14:29 |
| Sam-I-Am | Apsu: sounds like you should fix it :) | 14:29 |
| *** phalmos has quit IRC | 14:30 | |
| Sam-I-Am | how did this become invalid? | 14:31 |
| Sam-I-Am | https://bugs.launchpad.net/openstack-ansible/+bug/1475436 | 14:31 |
| openstack | Launchpad bug 1475436 in openstack-ansible juno "VLAN range issue in ml2_conf.ini" [Wishlist,Triaged] | 14:31 |
| Sam-I-Am | seems like it needs more discussion | 14:31 |
| Sam-I-Am | particularly along the lines of 'is this workaround hackish, or do we need to document this?' | 14:31 |
| *** TheIntern has quit IRC | 14:32 | |
| Apsu | Sam-I-Am: It's invalid for Kilo+ because the mechanism for parsing the OSAD config and generating ml2_conf.ini has changed, and the bug doesn't apply there anymore. | 14:35 |
| Apsu | It's still valid for Juno, and there's a workaround that should probably be documented. | 14:36 |
| *** TheIntern has joined #openstack-ansible | 14:37 | |
| Sam-I-Am | Apsu: so is what approd0 says... the way? | 14:40 |
| *** phalmos has joined #openstack-ansible | 14:40 | |
| Apsu | It should work, because the Juno template doesn't loop over the ranges variable to split it by commas | 14:40 |
| Apsu | Meaning we're basically doing an injection vulnerability on our config | 14:41 |
| Sam-I-Am | thats why i thought it seemed hackish | 14:41 |
| Apsu | So pretty much whatever's in there will get added as-is, so if you make sure it has the right name for each range pair, you can add as many as you want | 14:41 |
| evrardjp | svg mattt you there? I'd like to configure ceph, cherry-picked from master into kilo... for the user_secrets I've added cinder_ceph_client uuid... should I put the token I receive after client creation with the == ? | 14:47 |
| evrardjp | and for the user should I put "client." before my client name, like it's shown in ceph auth list? | 14:47 |
| Sam-I-Am | Apsu: i sort of think we should make this a little more... bettah | 14:49 |
| svg | evrardjp: the uuid will be generated randomly just like other secrets | 14:49 |
| Sam-I-Am | rather than documenting a hack that happens to work | 14:49 |
| Sam-I-Am | also, injection | 14:49 |
| Sam-I-Am | although i guess we dont validate a bunch of values for things | 14:49 |
| Apsu | Sam-I-Am: Me too. But I was told "this won't be getting backported as a fix". Or something similar that sounded like "no". | 14:50 |
| Apsu | The fix is pretty easy. | 14:50 |
| svg | and just give a user name, no client. prefix | 14:50 |
| Apsu | Just add a loop in the template | 14:50 |
| Sam-I-Am | maybe not backported, but for master | 14:50 |
| evrardjp | svg: I am not using the token generation | 14:50 |
| Sam-I-Am | who closed this bug? | 14:50 |
| Apsu | Sam-I-Am: Master/Kilo aren't affected, that's what I was saying | 14:50 |
| svg | Also evrardjp I found a small bug in user_variables #comments abou ceph_conf | 14:50 |
| Apsu | We switched from a Jinja template to a Python module | 14:50 |
| svg | evrardjp: what tokens are you talking about? | 14:51 |
| *** galstrom_zzz is now known as galstrom | 14:51 | |
| evrardjp | ok I get the issue, we are not speaking about the same stuff | 14:51 |
| svg | you were talking about ceph client uuid? | 14:51 |
| Sam-I-Am | Apsu: yeah... but does approd0's workaround apply to juno or newer versions? | 14:51 |
| Apsu | Juno. | 14:51 |
| Sam-I-Am | Apsu: in other words, we really dont document how to 'officially' accomplish multiple vlan ranges in any version | 14:52 |
| svg | the only ceph uuid is nova_ceph_client_uuid | 14:52 |
| evrardjp | svg: Yes, but I realize that cinder_ceph_client_uuid is an openstack uuid, nothing really interesting for ceph right? | 14:52 |
| evrardjp | just to summarize | 14:52 |
| evrardjp | I currently have my ceph cluster, with a pool for cinder | 14:52 |
| svg | no it's a uuid for libvirt config | 14:52 |
| evrardjp | I've created a user for it | 14:53 |
| evrardjp | it's mentionned as client.openstack in ceph auth list | 14:53 |
| evrardjp | so to use ceph in openstack I edited the openstack_user_config.yml | 14:53 |
| Apsu | Sam-I-Am: Hm, apparently we don't. I thought it was present in comment form in the example user_config. | 14:53 |
| Sam-I-Am | Apsu: no, its not... and its not obvious either. | 14:54 |
| Apsu | Sam-I-Am: The 'right' way is a:b,c:d[,...] | 14:54 |
| evrardjp | svg: in that file I have to write down a rbd_user and rbd_secret_uuid | 14:54 |
| Apsu | Sam-I-Am: The example Does say it corresponds to the ml2_conf.ini plugin options, which uses comma separation. But that's not super clear | 14:55 |
| evrardjp | in the openstack_user_config.yml.example, the default value for rbd_secret_uuid is "cinder_ceph_client_uuid" | 14:55 |
| Sam-I-Am | Apsu: yeah, i see that | 14:55 |
| evrardjp | so I thought this was the token used for auth | 14:55 |
| evrardjp | svg: but it isn't right? | 14:55 |
| openstackgerrit | Matt Thompson proposed stackforge/os-ansible-deployment: Bump tempest sha https://review.openstack.org/209104 | 14:56 |
| svg | no | 14:56 |
| Sam-I-Am | Apsu: debating re-opening that bug as a docs thing or just creating a new bug | 14:56 |
| svg | the _uuid var needs to go in user_secrets and will be generated by the inventory | 14:57 |
| Apsu | Sam-I-Am: Go for it, imo | 14:57 |
| svg | by that script | 14:57 |
| evrardjp | ok | 14:57 |
| Sam-I-Am | Apsu: option A or B ? | 14:57 |
| evrardjp | svg: so I have to get the token at another time I guess | 14:57 |
| evrardjp | ok | 14:57 |
| Apsu | Sam-I-Am: Both/and. Close your eyes and point. Or, re-open as a docs thing. Since it will have appropriate context already. | 14:58 |
| Sam-I-Am | right | 14:58 |
| svg | evrardjp: I don't follow | 14:58 |
| svg | what token are you talking about? | 14:59 |
| svg | toke != uuid | 14:59 |
| *** javeriak has joined #openstack-ansible | 14:59 | |
| svg | cinder_ceph_client_uuid: => user_secrets.yml | 14:59 |
| svg | and the name is default set: inventory/group_vars/hosts.yml:cinder_ceph_client: cinder | 15:00 |
| svg | only overrule if you need another name | 15:00 |
| evrardjp | svg: I got that the uuid is for libvirt | 15:03 |
| evrardjp | for the ceph auth you need a valid token for your client | 15:03 |
| evrardjp | you get it from ceph auth get-or-create, IIRC | 15:03 |
| svg | you mean a client 'secret' | 15:04 |
| *** yaya has joined #openstack-ansible | 15:04 | |
| svg | the ansible scripts will manage that | 15:04 |
| svg | you just need to mke sure you have ssh root access to your monitor hosts | 15:04 |
| evrardjp | yes, token - key - secret, whatever it's called | 15:04 |
| evrardjp | that's an issue | 15:05 |
| svg | Well, given those terms are oveabused in openstack all the place, it gets confusing :) | 15:05 |
| evrardjp | yeah sorry | 15:05 |
| evrardjp | :) | 15:05 |
| evrardjp | so I don't really have to care about that | 15:05 |
| svg | no | 15:05 |
| evrardjp | that token* | 15:05 |
| evrardjp | that whatever* | 15:06 |
| svg | see roles/ceph_client/tasks/ceph_auth.yml | 15:06 |
| *** javeriak has quit IRC | 15:06 | |
| *** javeriak has joined #openstack-ansible | 15:06 | |
| evrardjp | yeah I remember that | 15:06 |
| evrardjp | but I was afraid that rbd_secret_uuid in openstack_* | 15:07 |
| evrardjp | was used for auth | 15:07 |
| evrardjp | it's just for libvirt I guess | 15:07 |
| svg | yes | 15:07 |
| svg | Anyone around here that understands the inventory/dynamic_inventory.py sccript? | 15:07 |
| mattt | define understand | 15:08 |
| mattt | svg: i've made a small change to it, but i've never looked at it in detail | 15:09 |
| mattt | got a specific question? | 15:09 |
| svg | Well, it barfs at me from a fresh config :) | 15:11 |
| svg | File "inventory/dynamic_inventory.py", line 596, in _add_additional_networks | 15:11 |
| svg | container['ansible_ssh_host'] = networks[old_address]['address'] | 15:11 |
| svg | KeyError: 'address' | 15:11 |
| svg | trying to understand what that old_address stuff might be | 15:11 |
| odyssey4me | sigmavirus24 so https://review.openstack.org/209047 failed to build, and https://review.openstack.org/208801 will likely also fail without that config - should I merge the patches or do you think it's worth trying to keep them seperate and just to figure out why it's failing and see if that can be resolved first? | 15:11 |
| svg | given it's a fres config and not an upgrade, I could see how there wouldnt be old addresses | 15:11 |
| sigmavirus24 | odyssey4me: IMO merge them so they'll pass | 15:12 |
| *** daneyon has joined #openstack-ansible | 15:12 | |
| odyssey4me | it looks like it's a dep failure: stderr: Could not find a version that satisfies the requirement pbr<2.0,>=1.3 (from oslo.middleware) (from versions: 0.11.0)\nNo matching distribution found for pbr<2.0,>=1.3 (from oslo.middleware)\n', u'failed': True, 'attempts': 1, u'cmd': u'/usr/local/bin/pip install oslo.middleware'} | 15:12 |
| sigmavirus24 | odyssey4me: it seems reasonable to me that with a SHA bump (on master (aka unstable)) we might need to update configuration etc. | 15:12 |
| *** daneyon has quit IRC | 15:13 | |
| odyssey4me | sigmavirus24 yeah, quite possible - I'm going to do a quick rebuild of a test AIO to see if it's something that can work without merging | 15:14 |
| *** phalmos has quit IRC | 15:14 | |
| odyssey4me | that dep failure is odd | 15:14 |
| mattt | svg: having a look, nothing jumps out of me immediately tho | 15:16 |
| *** javeriak_ has joined #openstack-ansible | 15:16 | |
| sigmavirus24 | odyssey4me: so that's because we're pulling in an old version of requirements.txt | 15:17 |
| sigmavirus24 | And I think we use the lower bound on things when building now | 15:17 |
| sigmavirus24 | maybe | 15:17 |
| sigmavirus24 | because the old lower bound in requirements.txt was >=0.11 | 15:17 |
| *** javeriak has quit IRC | 15:17 | |
| sigmavirus24 | and the oslo.middleware that we're pulling in isn't pinned as far as I can tell | 15:17 |
| sigmavirus24 | so that has a different set of requirements from global-requirements and that is causing the conflict | 15:18 |
| sigmavirus24 | Also odyssey4me requirements synchronize across projects via reviews | 15:18 |
| svg | on a side node mattt, I found a small mistake in the ceph patch in etc/openstack_deploy/user_variables.yml: it doesnt document the new way of doing the ceph.conf config | 15:18 |
| svg | this should fix it: http://sprunge.us/OYVd | 15:18 |
| sigmavirus24 | so if that >=0.11,<2.0 requirement didn't sync to every project that's prior to the SHAs currently on master, yaprt might be packaging 0.11 because that's the only version that matches everything | 15:18 |
| sigmavirus24 | because it might be >=0.11,<1.0 on some package and 0.11.0 is the only version that matches of pbr | 15:19 |
| *** daneyon has joined #openstack-ansible | 15:19 | |
| mattt | svg: derp, i forgot to update that | 15:20 |
| mattt | svg: are you going to review that or shall i? | 15:20 |
| svg | I got it ready, was just wondering if I need to do more than just a review (need to file a bug?) | 15:20 |
| evrardjp | svg: your link was PRECISELY what I was asking now | 15:21 |
| evrardjp | :p | 15:21 |
| evrardjp | what I was planning to ask* | 15:22 |
| *** phalmos has joined #openstack-ansible | 15:22 | |
| mattt | evrardjp: i herped the derp :( | 15:22 |
| svg | evrardjp: I mentioned it earlier to you :) | 15:22 |
| evrardjp | indeed, but with the pastebin it's easier :D | 15:22 |
| mattt | svg: wonder if you can target it at the existing bug | 15:23 |
| evrardjp | I grepped so I found the ceph_mons thing | 15:23 |
| evrardjp | but still it's better in the doc | 15:23 |
| *** javeriak_ has quit IRC | 15:23 | |
| evrardjp | ceph_mons should be written like the old mon_hosts? or within brackets? | 15:25 |
| mattt | svg: i don't think a bug is necessary for that due to it changing a commented out section in user vars, however others may disagree | 15:25 |
| mattt | s/changing/removing/ | 15:25 |
| evrardjp | ['ip1','ip2','ip3'] ? | 15:25 |
| mattt | evrardjp: ceph_mons is a list | 15:25 |
| odyssey4me | sigmavirus24 odd, we have an issue here - with the updates sha's the pbr version fail is still there | 15:25 |
| odyssey4me | No matching distribution found for pbr<1.0,>=0.5.21 (from python-seamicroclient) | 15:26 |
| evrardjp | if the doc is incorrect, it's a bug, cause we consider the documentation as code :p | 15:26 |
| mattt | evrardjp: https://github.com/mattt416/rpc-openstack/blob/ceph_local/rpcd/etc/openstack_deploy/user_variables.yml#L38-L45 | 15:26 |
| * mattt goes and sprays himself down with bleach | 15:26 | |
| odyssey4me | sigmavirus24 the repo server has pbr 1.4.0 | 15:26 |
| d34dh0r53 | evrardjp: svg: something I put together yesterday https://gist.github.com/d34dh0r53/015a00d2100089fd4a0e, comments are welcome :) | 15:27 |
| odyssey4me | so I move to remove python-seamicroclient from the requirements | 15:27 |
| odyssey4me | let me test if that works | 15:27 |
| sigmavirus24 | odyssey4me: so uh | 15:27 |
| sigmavirus24 | There should be a way to tell tempest things to bugger off | 15:28 |
| mattt | evrardjp: i like to create bugs for everything, but i'm a bit pedantic and others disagree on the need | 15:28 |
| openstackgerrit | Serge van Ginderachter proposed stackforge/os-ansible-deployment: Update the documented ceph user variables https://review.openstack.org/209130 | 15:28 |
| sigmavirus24 | mattt: bugs for everything works for me because it usually describes why a patch is being worked on | 15:28 |
| sigmavirus24 | that way if someone else has to pick it up they have context | 15:29 |
| sigmavirus24 | /cc evrardjp | 15:29 |
| odyssey4me | sigmavirus24 this is the issue with the repo trying to satisfy all requirements and only have one version of all the requirements in it - we end up getting stuck in this nastiness | 15:29 |
| evrardjp | mattt: :D I agree, bugs are cool stuff. but you know I'm a hipster, I'm working with openstack and all... | 15:29 |
| evrardjp | :p | 15:29 |
| evrardjp | d34dh0r53: I guess the gist is already outdated | 15:30 |
| evrardjp | because the user_variables has changed | 15:30 |
| evrardjp | (the ceph.conf part) | 15:31 |
| svg | yes | 15:31 |
| d34dh0r53 | lol, when? | 15:31 |
| mattt | lolz | 15:31 |
| d34dh0r53 | I made this yesterday :/ | 15:31 |
| mattt | odyssey4me: *silence* | 15:31 |
| svg | d34dh0r53: nova_ceph_client_uuid: missing is a bug? | 15:32 |
| d34dh0r53 | svg: I'm not sure, possibly an illustration of my unfamiliarity with ceph | 15:32 |
| evrardjp | it proves one thing: ceph brings a lot of interest. | 15:33 |
| svg | the default is to use the cinder value | 15:33 |
| svg | nova_ceph_client_uuid: '{{ cinder_ceph_client_uuid | default() }}' | 15:33 |
| svg | as nova needs access to both cinder and nova ceph backends | 15:33 |
| svg | one user to rule them both | 15:33 |
| d34dh0r53 | so when I do that the virsh secret-define fails with a duplicate key | 15:34 |
| d34dh0r53 | ahh | 15:34 |
| d34dh0r53 | makes sense | 15:34 |
| *** yaya has quit IRC | 15:34 | |
| mattt | d34dh0r53: yeah if you set the uuid in your secrets file you won't see that | 15:34 |
| evrardjp | svg: for the ceph_mons I can use an itemized list, right? I didn't get the "it's a list" remark from mattt, as list have numerous ways to be written :p | 15:36 |
| *** javeriak has joined #openstack-ansible | 15:37 | |
| svg | well, both [1, 2, 3] and - 1\n- 2\m- 3 are valid yaml :) | 15:37 |
| evrardjp | if it's an inline list it's ['1','2','3'] i guess | 15:37 |
| mattt | evrardjp: i thought they were the same thing | 15:37 |
| evrardjp | ok that's what I thought | 15:37 |
| d34dh0r53 | I used - ip.ad.dr.ess\n in mine and it worked fine (needs to be indented under ceph_mons:) | 15:38 |
| evrardjp | mattt I'm just taking cautious steps here.... Don't want to really redo all that stuff because of a typo | 15:38 |
| evrardjp | sorry for being overly verbose | 15:38 |
| evrardjp | d34dh0r53: that is what I thought | 15:39 |
| mattt | evrardjp: :thumbsup: | 15:39 |
| svg | evrardjp: mattt integers vs strings if you quote them | 15:39 |
| evrardjp | svg: true | 15:40 |
| *** yaya has joined #openstack-ansible | 15:40 | |
| *** javeriak_ has joined #openstack-ansible | 15:42 | |
| *** KLev-away is now known as KLevenstein | 15:42 | |
| evrardjp | it's really different to have deployed host to connect on the ceph storage and the deploying host connecting to the ceph storage on my side... so it changes the model and also how the ceph.conf is fetched... | 15:43 |
| *** javeriak has quit IRC | 15:44 | |
| d34dh0r53 | reviews plz https://review.openstack.org/#/c/209130/ | 15:45 |
| odyssey4me | sigmavirus24 as I dig down the rabbit hole it seems that the dearth of clients we insist on including in the utility container is the culprit | 15:48 |
| odyssey4me | I'm starting to think that a cull is in order | 15:48 |
| evrardjp | d34dh0r53: done | 15:48 |
| d34dh0r53 | tys | 15:48 |
| odyssey4me | why should we include clients for projects which we aren't setting up the back-ends for | 15:48 |
| sigmavirus24 | odyssey4me: I thought tempest was put into virtualenvs | 15:48 |
| *** alop has joined #openstack-ansible | 15:48 | |
| sigmavirus24 | why isn't that in the virtualenv with the rest of tempest? | 15:48 |
| odyssey4me | sigmavirus24 it's not tempest, it's the utility container's pip packages | 15:49 |
| sigmavirus24 | Oh | 15:49 |
| sigmavirus24 | Why is that client in there then? | 15:49 |
| * sigmavirus24 is lost =P | 15:49 | |
| sigmavirus24 | Sam-I-Am: can you look at 209130? | 15:49 |
| odyssey4me | yeah, exactly | 15:49 |
| Sam-I-Am | sigmavirus24: not sme, but ok from a docs thing | 15:50 |
| Sam-I-Am | an example would be useful | 15:51 |
| odyssey4me | yeah, I think that perhaps it'd be good to put together a doc to add to the repo and attached to the install guide | 15:52 |
| odyssey4me | rather than trying to be too complete in user_variables | 15:52 |
| palendae | odyssey4me: That makes sense... sigmavirus24 and I talked about making a schema validator for user_variables and/or user_config, and generating the example files from that | 15:53 |
| palendae | In that scenario, we couldn't have docs in the example file | 15:53 |
| sigmavirus24 | palendae: so we could generate docs in the example file with that | 15:54 |
| sigmavirus24 | If we have documentation in the schema | 15:54 |
| sigmavirus24 | but yeah | 15:54 |
| sigmavirus24 | I have N things to work on right now | 15:54 |
| *** phalmos has quit IRC | 15:54 | |
| Sam-I-Am | config file generation is a Good Thing | 15:54 |
| sigmavirus24 | And K conversations happening in different channels | 15:54 |
| Sam-I-Am | dont do a neutron :) | 15:54 |
| odyssey4me | palendae sigmavirus24 I love that idea - rather have docstrings or something in the code and generate example configs (and docs) from them | 15:54 |
| palendae | sigmavirus24: Oh, no, it's just a mention of, "We have this idea" | 15:55 |
| *** yaya has quit IRC | 15:55 | |
| palendae | Not to work on now | 15:55 |
| palendae | *Maybe* at the work hackathon in a few weeks | 15:55 |
| evrardjp | Sam-I-Am: for the patch you asked me to test, does it involve a change of neutron config for better testing? | 15:55 |
| odyssey4me | palendae I think for dev documentation that'd be perfect | 15:56 |
| evrardjp | I've cherry-picked it, and I was planning to test directly | 15:56 |
| palendae | odyssey4me: There's also the problem of people generating invalid configs and filing bugs | 15:56 |
| palendae | Because they forgot a variable | 15:56 |
| palendae | So I'd like to help remedy that | 15:56 |
| sigmavirus24 | yep | 15:56 |
| odyssey4me | for more long-hand documentation made for first-timers we can then narrow down the docs we carry to only cover the simple and common use-cases | 15:56 |
| sigmavirus24 | That's why i wrote the project in the first place | 15:56 |
| palendae | sigmavirus24: Right | 15:56 |
| sigmavirus24 | "I forgot a thing and forgot its format" | 15:57 |
| palendae | But reducing the number of places we have to add a new option would be good | 15:57 |
| odyssey4me | sigmavirus24 palendae yep, and then there're the format and name changes between versions too :/ | 15:57 |
| palendae | Not sure if we'd handle that | 15:57 |
| Sam-I-Am | evrardjp: it doesnt change how neutron works | 15:57 |
| palendae | minimum viable product :) | 15:57 |
| Sam-I-Am | evrardjp: it just removes unnecessary config items from certain places, and clarifies some others | 15:58 |
| palendae | And things like infra-repo_hosts vs openstack_repo_url being defined in 2 different files would need to be taken into account | 15:58 |
| odyssey4me | evrardjp it should work with your existing config | 15:58 |
| *** woodard has quit IRC | 15:58 | |
| palendae | You need one or the other | 15:58 |
| sigmavirus24 | palendae: we can graft that on later | 15:58 |
| odyssey4me | if it doesn't, then the patch needs to be fixed :) | 15:58 |
| palendae | Sure | 15:58 |
| sigmavirus24 | like "This variable name has been renamed to blah" | 15:58 |
| *** woodard has joined #openstack-ansible | 15:58 | |
| palendae | as long as we don't use oslo.config's deprecation methodology | 15:58 |
| openstackgerrit | Serge van Ginderachter proposed stackforge/os-ansible-deployment: Update the documented ceph user variables https://review.openstack.org/209130 | 15:58 |
| palendae | Cause that was impenetrable | 15:58 |
| sigmavirus24 | palendae: we won't be able to | 15:58 |
| palendae | Good | 15:59 |
| odyssey4me | oh and sigmavirus24 you were right - it was all the tempest clients, but I think they still get taken into account for all the requirements.... anyway, testing another try and making this work | 15:59 |
| evrardjp | ok relaunching all my playbooks now | 15:59 |
| Sam-I-Am | bug triage meeting? | 16:01 |
| Sam-I-Am | or something | 16:01 |
| *** phalmos has joined #openstack-ansible | 16:02 | |
| woodard | bug meeting? | 16:02 |
| odyssey4me | alrighty everyone - we have a bug triage meeting now | 16:02 |
| Sam-I-Am | yes, we do | 16:03 |
| odyssey4me | cloudnull, mattt, andymccr, d34dh0r53, hughsaunders, b3rnard0, palendae, Sam-I-Am, odyssey4me, serverascode, rromans, mancdaz, dolphm, _shaps_, BjoernT, claco, echiu, dstanek, jwagner, ayoung bug triage meeting in this channel :) | 16:03 |
| odyssey4me | let's start from the top: https://bugs.launchpad.net/openstack-ansible/+bug/1479484 | 16:03 |
| openstack | Launchpad bug 1479484 in openstack-ansible trunk "server_id missing in my.cnf" [Medium,New] | 16:03 |
| odyssey4me | phalmos you around? | 16:04 |
| *** yaya has joined #openstack-ansible | 16:04 | |
| phalmos | Yes sir | 16:04 |
| odyssey4me | ok, we discussed this briefly on thu and left it as 'new' to ensure we could spend a little more time on it today | 16:04 |
| odyssey4me | this should be a pretty simple template change in the galera_server role - I guess the only complexity is to ensure the numbers are unique for each node | 16:05 |
| evrardjp | svg: isn't there a space missing in your new example? ;) | 16:05 |
| svg | Where? | 16:05 |
| evrardjp | end of day I didn't see if there was one space or two... | 16:05 |
| evrardjp | https://review.openstack.org/#/c/209130/2/etc/openstack_deploy/user_variables.yml | 16:06 |
| Sam-I-Am | phalmos: what does this change accomplish? | 16:06 |
| evrardjp | ceph_mons | 16:06 |
| odyssey4me | phalmos which section/file should the server_id setting be found? | 16:06 |
| svg | yes, though one space is still valid yaml :) | 16:06 |
| evrardjp | oh | 16:06 |
| phalmos | It allows a directed replication if you need to restore | 16:06 |
| evrardjp | really ? | 16:06 |
| svg | of course | 16:06 |
| evrardjp | good to know | 16:07 |
| Sam-I-Am | phalmos: can you add those details to the bug? | 16:07 |
| odyssey4me | evrardjp yes, it just looks better if it's two :p | 16:07 |
| phalmos | yes. | 16:07 |
| Sam-I-Am | and where the option goes, pls | 16:07 |
| Sam-I-Am | makes it easier on the brain later | 16:07 |
| odyssey4me | Sam-I-Am +1 | 16:07 |
| Sam-I-Am | 'why is this what what?' heh | 16:07 |
| phalmos | will do | 16:07 |
| openstackgerrit | Serge van Ginderachter proposed stackforge/os-ansible-deployment: Update the documented ceph user variables https://review.openstack.org/209130 | 16:07 |
| *** erikmwilson has joined #openstack-ansible | 16:07 | |
| evrardjp | :D | 16:08 |
| odyssey4me | once that's done we can mark it as triaged and we'll hunt down someone to put a patch in for it | 16:08 |
| Sam-I-Am | phalmos: or you can patch it :) | 16:08 |
| odyssey4me | that said, if anyone here wants to pick it up, feel free to do so :) | 16:08 |
| Sam-I-Am | looks fairly straightforward | 16:08 |
| odyssey4me | next: https://bugs.launchpad.net/openstack-ansible/+bug/1399684 | 16:08 |
| openstack | Launchpad bug 1399684 in openstack-ansible "tenant_network_types should be configurable" [Low,New] | 16:09 |
| Sam-I-Am | odyssey4me: i just commented on this bug | 16:09 |
| odyssey4me | I see so. | 16:09 |
| odyssey4me | Sam-I-Am does your recent review change this at all? | 16:09 |
| Sam-I-Am | if it actually works that way (and i think it does), do we simply want to say 'order the networks in the order of priority' or actually code up some way to specifically set it | 16:10 |
| Sam-I-Am | no, it doesnt | 16:10 |
| Sam-I-Am | adding an option for default project network type probably isn't difficult, but it needs to validate that the value is also in one of the provider network stanzas | 16:10 |
| odyssey4me | so it's most likely that unless we introduce some sort of ordering flag that this will never be possible to order | 16:11 |
| Sam-I-Am | dont want to have default network = vlan and there's no vlan provider network | 16:11 |
| Sam-I-Am | odyssey4me: right now i think it just uses the order of your provider networks | 16:11 |
| odyssey4me | I like cloudnull_afk's thinking - just allow it to be overridden, then deployers can specify an order if they want | 16:11 |
| Sam-I-Am | so it isnt indeterminate | 16:11 |
| odyssey4me | Sam-I-Am yeah, it'll likely be a list return which will be different every time most likely | 16:12 |
| Sam-I-Am | the only value that really matters is the first value | 16:12 |
| Sam-I-Am | so thats the only one we need to consider | 16:12 |
| odyssey4me | so this needs some extra work to triage | 16:12 |
| odyssey4me | we can confirm it, and it's already marked as low priority | 16:12 |
| Sam-I-Am | if you create a network as a non-admin project, it uses the first type in the list | 16:12 |
| Sam-I-Am | the only way to use any of the other types is to either a) be admin or b) allow it in policy | 16:13 |
| odyssey4me | next: https://bugs.launchpad.net/openstack-ansible/+bug/1412762 | 16:13 |
| openstack | Launchpad bug 1412762 in openstack-ansible juno "Disable unneeded heat resource-types" [Low,Confirmed] | 16:13 |
| odyssey4me | mattt are you around? | 16:13 |
| odyssey4me | Sam-I-Am on the previous bug, if you want to spend some time triaging that then go for it | 16:14 |
| Sam-I-Am | mmkay | 16:14 |
| odyssey4me | ok, we'll carry this bug until next week | 16:14 |
| mattt | odyssey4me: on a call bu tyes | 16:14 |
| mattt | *but yes | 16:14 |
| svg | mattt: or someone else: trying to understand the dynamic inventory script, and why it barfs here: http://paste.ubuntu.com/12001011/ | 16:15 |
| svg | but I'm off heading home, ttyl guys | 16:15 |
| odyssey4me | mattt I resurrected the above bug (disable unneeded heat resource types for discussion with regards to whether it needs to be done or whether we should can it | 16:15 |
| odyssey4me | miguelgrinberg is of the opinion that it's not really worth the bother | 16:15 |
| mattt | odyssey4me: it's a nice to have for sure | 16:15 |
| mattt | odyssey4me: but low prioprity | 16:16 |
| odyssey4me | I am paraphrasing quite a bit there :) | 16:16 |
| mattt | *priority | 16:16 |
| odyssey4me | ok mattt I'm going to classify it as a wishlist item then | 16:16 |
| mattt | odyssey4me: that sounds fair | 16:16 |
| miguelgrinberg | mattt: maybe it makes sense for rpc-openstack, but not for osad | 16:16 |
| odyssey4me | alright, wishlist - but no target | 16:17 |
| odyssey4me | next: https://bugs.launchpad.net/openstack-ansible/+bug/1479473 | 16:17 |
| openstack | Launchpad bug 1479473 in openstack-ansible trunk "Need log_slave_update enabled" [Wishlist,New] | 16:17 |
| odyssey4me | phalmos as with the other one, if you're able to add examples of what you're looking for into the bug then we should easily get a patch in pretty quickly | 16:18 |
| odyssey4me | this is a low hanging fruit wishlist item which helps with data restores when they're needed | 16:18 |
| phalmos | yup | 16:18 |
| odyssey4me | next: https://bugs.launchpad.net/openstack-ansible/+bug/1475436 | 16:19 |
| openstack | Launchpad bug 1475436 in openstack-ansible juno "VLAN range issue in ml2_conf.ini" [Wishlist,Triaged] | 16:19 |
| Sam-I-Am | yeah | 16:19 |
| Apsu | Fun one | 16:19 |
| Sam-I-Am | i re-opened this | 16:19 |
| *** woodard has quit IRC | 16:19 | |
| Sam-I-Am | mostly because we dont document how to accomplish this | 16:19 |
| Sam-I-Am | probably needs a patch for juno, and another patch for master for backport to kilo | 16:20 |
| odyssey4me | Sam-I-Am a doc patch then? | 16:20 |
| *** woodard has joined #openstack-ansible | 16:20 | |
| Sam-I-Am | odyssey4me: yeah, i think this is all docs... perhaps some testing. | 16:20 |
| Apsu | Yep | 16:21 |
| Sam-I-Am | seems like appprod0 posted the workaround for juno | 16:21 |
| Apsu | s/workaround/injection attack/ | 16:21 |
| Sam-I-Am | haha | 16:21 |
| Apsu | It's a smart workaround, makes me laugh. | 16:21 |
| Sam-I-Am | one of these days we need to validate stuff a bit better :) | 16:21 |
| Sam-I-Am | but thats a side project | 16:21 |
| Apsu | Validation is for sissies. Be liberal in what you accept! | 16:21 |
| Sam-I-Am | wasnt most of this written in texas? | 16:22 |
| Sam-I-Am | there aint no liberal | 16:22 |
| Apsu | SA is a microcosm unto itself. | 16:22 |
| Sam-I-Am | micromexico? | 16:23 |
| Apsu | lulz | 16:23 |
| Apsu | Ok, odyssey4me can move on now :P | 16:23 |
| Sam-I-Am | yep | 16:23 |
| Sam-I-Am | NEXT! | 16:23 |
| odyssey4me | sorry - had someone come to my desk :p | 16:23 |
| odyssey4me | Sam-I-Am is this a doc patch then? | 16:24 |
| Sam-I-Am | odyssey4me: think xo | 16:24 |
| Sam-I-Am | so | 16:24 |
| odyssey4me | ok, is this something you will be picking up? | 16:24 |
| Sam-I-Am | yeah | 16:25 |
| odyssey4me | ok, I'll assign to you - shall I target it to a milestone or leave it for now? | 16:25 |
| Sam-I-Am | whatever is best there | 16:25 |
| Sam-I-Am | not sure what the target would be | 16:26 |
| odyssey4me | https://bugs.launchpad.net/openstack-ansible/+bug/1479127 | 16:26 |
| openstack | Launchpad bug 1479127 in openstack-ansible "nf_conntrack table fills on swift nodes" [Undecided,New] | 16:26 |
| *** bapalm_ has quit IRC | 16:27 | |
| Sam-I-Am | seems like a valid issue | 16:27 |
| Sam-I-Am | but !sme | 16:27 |
| odyssey4me | I've definitely seen this in production before | 16:27 |
| odyssey4me | Apsu rackertom any comment on this? | 16:27 |
| Apsu | Sure, like we've been doing for ~15 years or so, if you know you'll have lots of connection initiation, you set the time-wait reuse sysctl | 16:28 |
| Apsu | Jordan linked a related bug that had that exact fix | 16:28 |
| odyssey4me | this is marked as related and has patches for it https://bugs.launchpad.net/openstack-ansible/+bug/1441363 | 16:28 |
| openstack | Launchpad bug 1441363 in openstack-ansible trunk "nf_conntrack schould be unloaded on swift object server" [Medium,Fix released] - Assigned to Andy McCrae (andrew-mccrae) | 16:28 |
| Apsu | Yep | 16:29 |
| *** openstackgerrit_ has quit IRC | 16:29 | |
| odyssey4me | so do we still have a problem? was the previous patch not good enough? | 16:29 |
| Apsu | Raising the nf_conntrack_max is basically never correct unless you expect to maintain a huge number of active connections | 16:29 |
| odyssey4me | or is this a situation where someone just wasn't running with that patch? | 16:29 |
| Apsu | If your issue is connection creation speed, you set tw_reuse | 16:29 |
| odyssey4me | it seems to me like we need ot know what version they were running to determine whether this had the tw_reuse option set | 16:30 |
| odyssey4me | marking as incomplete | 16:30 |
| Apsu | Well not only that, but we need to understand why they're running out of conntrack entries. | 16:31 |
| Apsu | Are these TW connections (without that sysctl)? Or are there > nf_conntrack_max active connections | 16:31 |
| Apsu | A simple ss -s to show a summary would be helpful | 16:31 |
| odyssey4me | Apsu can you pose some relevant question in the bug please? | 16:31 |
| Apsu | Sure | 16:32 |
| odyssey4me | thanks Apsu | 16:33 |
| odyssey4me | next: https://bugs.launchpad.net/openstack-ansible/+bug/1479360 | 16:33 |
| openstack | Launchpad bug 1479360 in openstack-ansible "MaaS: RabbitMQ's maas_user password not set or used in 10.1.10" [Undecided,New] | 16:33 |
| odyssey4me | palendae you around? | 16:33 |
| palendae | Yep | 16:33 |
| palendae | I'd imagine that would be a 10.1.12 thing | 16:33 |
| odyssey4me | is this a bug that's yet to be fixed? | 16:34 |
| palendae | I think so | 16:34 |
| odyssey4me | ok, cool - moving on: https://bugs.launchpad.net/openstack-ansible/+bug/1479870 | 16:34 |
| openstack | Launchpad bug 1479870 in openstack-ansible "teardown.sh does not cleanup lvm.conf" [Undecided,New] | 16:34 |
| palendae | d34dh0r53: ^ | 16:35 |
| palendae | rackertom: I think you looked at ^^ too | 16:35 |
| odyssey4me | heh, that is plausible - considering that teardown.sh is a dev/test tool I think we can wishlist this and move on? | 16:36 |
| palendae | I'd say so | 16:36 |
| odyssey4me | https://bugs.launchpad.net/openstack-ansible/+bug/1480914 | 16:36 |
| openstack | Launchpad bug 1480914 in openstack-ansible " Add Ceph/RBD support to playbooks" [Undecided,New] | 16:36 |
| palendae | I think that's in progress | 16:36 |
| odyssey4me | ah, that's a docimpact bug | 16:36 |
| palendae | Oh | 16:37 |
| odyssey4me | I'll assign to docs and move on | 16:37 |
| odyssey4me | next: https://bugs.launchpad.net/openstack-ansible/+bug/1480987 | 16:37 |
| openstack | Launchpad bug 1480987 in openstack-ansible ".my.cnf should not contain LB VIP" [Undecided,New] | 16:37 |
| phalmos | me again | 16:38 |
| odyssey4me | hmm, ok I disagree - except on the galera containers themselves | 16:38 |
| phalmos | Yes, in the containers is where things get confusing | 16:38 |
| phalmos | containers' client should only hit th elocal instance | 16:39 |
| phalmos | not the VIP | 16:39 |
| odyssey4me | phalmos is this juno specific, as I recall in kilo/master we already set the host to 127.0.0.1 in the galera containers | 16:39 |
| phalmos | This was Juno. I was unaware of the change in kilo/master | 16:39 |
| odyssey4me | ok cool - will mark this for juno only | 16:39 |
| *** yaya has quit IRC | 16:40 | |
| palendae | phalmos: Assuming you'd like this for 10.1.12 then? | 16:40 |
| palendae | I'll get cards made on RAX's side for these so we have visibility | 16:40 |
| phalmos | I guess so | 16:40 |
| phalmos | we should keep it consistent with Kilo. That way it isn't that much of a departure | 16:41 |
| phalmos | for support | 16:41 |
| palendae | Fair | 16:41 |
| odyssey4me | phalmos so in kilo only the galera containers are set to look at 127.0.0.1 | 16:41 |
| phalmos | ok, makes sense | 16:42 |
| odyssey4me | we have a review in to make sure that everything else points to the vip as it currently also points to 127.0.0.1 which is useless :p | 16:42 |
| odyssey4me | fyi https://review.openstack.org/207939 | 16:42 |
| odyssey4me | https://bugs.launchpad.net/openstack-ansible/+bug/1481380 | 16:43 |
| openstack | Launchpad bug 1481380 in openstack-ansible trunk "Bump tempest sha" [Undecided,New] | 16:43 |
| odyssey4me | mattt this is likely to be covered by https://review.openstack.org/208801 but it's fine that you have an independant review in which may be useful for backporting to kilo | 16:43 |
| odyssey4me | all done, thanks all - are there any other bugs that anyone wants to raise for discussion? | 16:45 |
| Sam-I-Am | not here | 16:46 |
| *** TheIntern has quit IRC | 16:48 | |
| odyssey4me | alright, then I guess that about covers it :) | 16:50 |
| odyssey4me | thank you all for your time | 16:51 |
| *** fawadkhaliq has quit IRC | 16:52 | |
| rackertom | palendae: Sorry, went to lunch. Yes, I looked at that problem once upon a time. There wa a .bak file being created and if you ran the setup too many times the .bak became the modified version. | 16:52 |
| rackertom | So, a simple way would be if *.bak does not exist, copy *.conf to *.orig. | 16:53 |
| rackertom | Sorta like dpkg does | 16:53 |
| evrardjp | Sam-I-Am: I'm off for today, and I'm not available tomorrow... So for your testing... it's following its course, but at the moment only the setup-infrastructure is re-running .... It's gonna take a while before you get your results. Sorry for that | 16:57 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Improve Keystone Apache configuration https://review.openstack.org/209047 | 17:00 |
| *** yaya has joined #openstack-ansible | 17:00 | |
| odyssey4me | sigmavirus24 stevelle rackertom ^ might do the trick - busy running two tests to find out | 17:00 |
| *** phalmos has quit IRC | 17:01 | |
| rackertom | odyssey4me: I totally agree with Apsu comment on there. Having 300k connections actively tracked by swift seems insane to me. | 17:01 |
| odyssey4me | not sure if tempest actually needs those for some reason | 17:01 |
| Sam-I-Am | evrardjp: np, thanks for testing it | 17:01 |
| sigmavirus24 | I was about to dart out for a ride over lunch | 17:01 |
| Sam-I-Am | sigmavirus24: so ride? | 17:01 |
| Apsu | rackertom: nod | 17:01 |
| sigmavirus24 | I'll test it when I get back dolphm | 17:01 |
| sigmavirus24 | er odyssey4me | 17:01 |
| Apsu | rackertom: Something's screwy. Most likely it's just a bunch of TIME-WAIT NAT entries | 17:02 |
| Apsu | The sockstat summary should tell us immediately | 17:02 |
| * dolphm got excited | 17:02 | |
| * dolphm is dissapoint | 17:02 | |
| rackertom | Yeah, my guess is the monitoring tool is behaving badly and opening/closing connections quickly whereas maybe a persistent connection might be better. | 17:03 |
| *** yaya has quit IRC | 17:03 | |
| Apsu | Yep. Or at least reusing TWs, which is 99.9999% safe these days | 17:04 |
| Apsu | Recycling is still iffy. I always get nervous when folks are like "just set this sysctl!". | 17:04 |
| Apsu | But I get more nervous when they start upping sysctl maxes. | 17:04 |
| Sam-I-Am | heh | 17:04 |
| Apsu | Because that's prooooobably not the solution to the actual issue | 17:04 |
| Sam-I-Am | Apsu: why do we want real solutions? | 17:04 |
| Sam-I-Am | just throw more X as it | 17:05 |
| Apsu | Oh good point, I forgot we weren't talking about real engineering. | 17:05 |
| Apsu | More is better. | 17:05 |
| *** daneyon has quit IRC | 17:05 | |
| *** daneyon has joined #openstack-ansible | 17:05 | |
| *** abitha has joined #openstack-ansible | 17:06 | |
| *** javeriak has joined #openstack-ansible | 17:06 | |
| *** daneyon has quit IRC | 17:07 | |
| *** daneyon has joined #openstack-ansible | 17:07 | |
| *** javeriak_ has quit IRC | 17:08 | |
| d34dh0r53 | core reviews please https://review.openstack.org/#/c/209130/ | 17:11 |
| rromans | http://docs-internal.rackspace.com/rpc/api/v11/bk-rpc-swift/content/ch-object-storage-overview.html | 17:21 |
| rromans | sorry, wrong window | 17:22 |
| *** daneyon_ has joined #openstack-ansible | 17:27 | |
| *** daneyon has quit IRC | 17:30 | |
| *** javeriak has quit IRC | 17:36 | |
| *** erikmwilson has left #openstack-ansible | 17:45 | |
| *** javeriak has joined #openstack-ansible | 17:58 | |
| *** javeriak has quit IRC | 18:03 | |
| *** javeriak has joined #openstack-ansible | 18:04 | |
| *** TheIntern has joined #openstack-ansible | 18:05 | |
| Sam-I-Am | stupid gate failures | 18:10 |
| *** phalmos has joined #openstack-ansible | 18:13 | |
| sigmavirus24 | Sam-I-Am: are they stupid? | 18:13 |
| Sam-I-Am | sigmavirus24: failing for no useful reason, sure | 18:14 |
| sigmavirus24 | that's not stupid, that's predictable | 18:14 |
| Sam-I-Am | lulz | 18:16 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Improve Keystone Apache configuration https://review.openstack.org/209047 | 18:19 |
| odyssey4me | sigmavirus24 it looks like openstack-ci is under pressure again and that our gating on master won't be functional until tomorrow (if then) | 18:21 |
| odyssey4me | we'll need to keep testing manually until we get a pass on a patch | 18:21 |
| odyssey4me | the above patch is still an attempt to keep the unblock relatively clean, but we're starting to need multi-purpose patches to unblock anyway | 18:22 |
| odyssey4me | I don't want to merge them until I get a clean build with them both added, but a fail with just one | 18:22 |
| odyssey4me | so far I have a clean build, but not a successful tempest test | 18:23 |
| sigmavirus24 | interesting | 18:27 |
| *** bapalm_ has joined #openstack-ansible | 18:41 | |
| odyssey4me | lbragstad great feedback - did those calculations come from some sort of background tests or something? | 18:42 |
| lbragstad | odyssey4me: those are from when dolphm and I poked at keystone during performance tests, not sure if the *actual* values of those calculations are what you need but that's how we went about the math | 18:43 |
| lbragstad | those numbers are what seemed to give us the best results | 18:44 |
| *** phalmos has quit IRC | 18:50 | |
| *** jasondotstar has joined #openstack-ansible | 19:01 | |
| odyssey4me | lbragstad excellent, I'll update the defaults | 19:02 |
| *** phalmos has joined #openstack-ansible | 19:05 | |
| *** alextricity has quit IRC | 19:17 | |
| *** TheIntern has quit IRC | 19:18 | |
| *** TheIntern has joined #openstack-ansible | 19:20 | |
| *** phalmos has quit IRC | 19:24 | |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Improve Keystone Apache configuration https://review.openstack.org/209047 | 19:32 |
| *** phalmos has joined #openstack-ansible | 19:32 | |
| odyssey4me | sigmavirus24 are you around? | 19:38 |
| *** britthouser has joined #openstack-ansible | 19:39 | |
| *** jasondotstar is now known as jasondotstar|afk | 19:39 | |
| *** britthou_ has joined #openstack-ansible | 19:40 | |
| sigmavirus24 | yep | 19:42 |
| sigmavirus24 | odyssey4me: | 19:43 |
| *** britthouser has quit IRC | 19:43 | |
| odyssey4me | sigmavirus24 so I've narrowed down the tempest test failures (at least some) to swift rejecting auth | 19:43 |
| sigmavirus24 | hm | 19:43 |
| odyssey4me | I've also confirmed this using the openstack client to simply try to create a container in swift from the utility lxc container | 19:44 |
| sigmavirus24 | any explanation in swift logs? | 19:44 |
| odyssey4me | sigmavirus24 just the usual stfu: proxy-server: Invalid user token - deferring reject downstream | 19:44 |
| sigmavirus24 | odd | 19:45 |
| sigmavirus24 | I'm still building that patch but I guess I'll have to restart =/ | 19:46 |
| odyssey4me | which patch? | 19:47 |
| sigmavirus24 | yours | 19:47 |
| sigmavirus24 | the keystone apache config one | 19:47 |
| sigmavirus24 | 209047 | 19:47 |
| odyssey4me | I've done the sequence of that + the sha updates, and am using the previous patch set of the keystone/apache one. | 19:48 |
| odyssey4me | interesting, it looks like keystone approves the token, but swift denies it | 19:48 |
| odyssey4me | odd | 19:48 |
| sigmavirus24 | could it be an issue with roles? | 19:50 |
| sigmavirus24 | as in yes this token is valid comes from keystone and then swift's default policy is different in the new SHA so the tempest user isn't authorized to do whatever? | 19:50 |
| odyssey4me | that's a possibility | 19:51 |
| odyssey4me | although I'm using the admin user for the testing right now | 19:52 |
| odyssey4me | sigmavirus24 here's the client debug log: http://paste.openstack.org/show/tQSr8mOjePbwpH3OlD8J/ | 19:54 |
| sigmavirus24 | odyssey4me: those are some mighty odd logs | 19:58 |
| odyssey4me | sigmavirus24 welcome to the openstack client logs :) | 19:58 |
| sigmavirus24 | It looks as though openstackclient tries twice before failing | 19:58 |
| odyssey4me | well, if it's ok I'll hand this over to you - see what you can do to dig up the next issue and hopefully resolve it | 19:59 |
| odyssey4me | otherwise I'll pick it up in the morning | 19:59 |
| *** javeriak_ has joined #openstack-ansible | 19:59 | |
| sigmavirus24 | yeah I'll try to dual wield tasks | 20:01 |
| *** javeriak has quit IRC | 20:02 | |
| odyssey4me | sigmavirus24 you know the priorities, so don't kill yourself over it | 20:03 |
| sigmavirus24 | yeah | 20:04 |
| sigmavirus24 | but killing myself is half the fun | 20:04 |
| odyssey4me | stevelle you around? if you have a gap to dig into it, the patches I've submitted get us to a successful build but swift appears to be rejecting auth and it needs digging into | 20:04 |
| odyssey4me | sigmavirus24 stevelle was looking into this last night, and it's blocking his work, so if he has cycles then he can take over :) | 20:05 |
| odyssey4me | you can just shake pom poms from the sidelines :p | 20:05 |
| stevelle | If I get a gap I'll repro odyssey4me | 20:06 |
| sigmavirus24 | erry day I'm building AIOs | 20:06 |
| *** yaya has joined #openstack-ansible | 20:06 | |
| odyssey4me | alright, have a good day - I'll pick up in the morning | 20:07 |
| odyssey4me | pop me an email if you find something more, or update the review with comments or whatever | 20:07 |
| openstackgerrit | Jesse Pretorius proposed stackforge/os-ansible-deployment: Reduce neutron configuration https://review.openstack.org/204597 | 20:10 |
| *** yaya has quit IRC | 20:16 | |
| *** javeriak_ has quit IRC | 20:27 | |
| *** woodard has quit IRC | 20:50 | |
| *** yaya has joined #openstack-ansible | 20:51 | |
| *** jmckind has joined #openstack-ansible | 20:56 | |
| mattt | odyssey4me: oops, didn't see that review ... i can close mine off if the end-to-end bump will go through relatively quickly | 21:04 |
| mattt | (i'll review it tomorrow to help that along) | 21:04 |
| *** bapalm_ has quit IRC | 21:07 | |
| *** jasondotstar|afk is now known as jasondotstar | 21:07 | |
| *** abitha has quit IRC | 21:10 | |
| *** abitha has joined #openstack-ansible | 21:11 | |
| mattt | svg: sorry, had to run off to airport earlier, can you check if your openstack_user_config.yml has ip_from_q names under the networks in provider_networks match the names in cidr_networks? | 21:16 |
| *** dmsimard has quit IRC | 21:21 | |
| *** spotz is now known as spotz_zzz | 21:24 | |
| *** jmckind has quit IRC | 21:24 | |
| *** jmckind has joined #openstack-ansible | 21:26 | |
| *** dmsimard has joined #openstack-ansible | 21:40 | |
| *** yaya has quit IRC | 21:43 | |
| *** yaya has joined #openstack-ansible | 21:43 | |
| *** phalmos has quit IRC | 21:51 | |
| *** yaya has quit IRC | 21:53 | |
| *** JRobinson__ has joined #openstack-ansible | 22:00 | |
| *** jmckind has quit IRC | 22:02 | |
| *** TheIntern has quit IRC | 22:05 | |
| *** sigmavirus24 is now known as sigmavirus24_awa | 22:25 | |
| *** KLevenstein has quit IRC | 22:30 | |
| *** jaypipes has quit IRC | 22:35 | |
| *** jasondotstar has quit IRC | 22:36 | |
| *** galstrom is now known as galstrom_zzz | 22:40 | |
| *** sdake has quit IRC | 23:00 | |
| *** jasondotstar has joined #openstack-ansible | 23:20 | |
| *** jmckind has joined #openstack-ansible | 23:22 | |
| *** jmckind has quit IRC | 23:23 | |
| *** jmckind has joined #openstack-ansible | 23:24 | |
| *** sdake has joined #openstack-ansible | 23:45 | |
| *** sdake_ has joined #openstack-ansible | 23:48 | |
| *** sdake has quit IRC | 23:52 | |
| *** galstrom_zzz is now known as galstrom | 23:55 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!