| *** kylek3h has quit IRC | 00:10 | |
| *** thorst has joined #openstack-ansible | 00:20 | |
| *** yifei has joined #openstack-ansible | 00:27 | |
| *** esberglu has joined #openstack-ansible | 00:33 | |
| *** esberglu has quit IRC | 00:37 | |
| *** markvoelker has joined #openstack-ansible | 00:42 | |
| *** markvoelker has quit IRC | 01:15 | |
| *** thorst has quit IRC | 01:22 | |
| *** thorst has joined #openstack-ansible | 01:23 | |
| *** hw_wutianwei has joined #openstack-ansible | 01:27 | |
| *** sxc731 has joined #openstack-ansible | 01:34 | |
| *** sxc731 has left #openstack-ansible | 01:34 | |
| *** thorst has quit IRC | 01:35 | |
| *** sxc731 has joined #openstack-ansible | 01:59 | |
| *** esberglu has joined #openstack-ansible | 02:21 | |
| *** sxc731 has joined #openstack-ansible | 02:22 | |
| *** esberglu has quit IRC | 02:25 | |
| *** thorst has joined #openstack-ansible | 02:36 | |
| *** sxc731 has quit IRC | 02:59 | |
| *** markvoelker has joined #openstack-ansible | 03:12 | |
| *** esberglu has joined #openstack-ansible | 03:15 | |
| *** japestinho has joined #openstack-ansible | 03:15 | |
| *** thorst has quit IRC | 03:17 | |
| *** thorst has joined #openstack-ansible | 03:17 | |
| *** esberglu has quit IRC | 03:20 | |
| *** hamzy has quit IRC | 03:20 | |
| *** thorst has quit IRC | 03:24 | |
| *** gkadam has joined #openstack-ansible | 03:38 | |
| *** gkadam has quit IRC | 03:39 | |
| *** gkadam has joined #openstack-ansible | 03:39 | |
| *** markvoelker has quit IRC | 03:47 | |
| *** chhavi has joined #openstack-ansible | 04:05 | |
| *** esberglu has joined #openstack-ansible | 04:09 | |
| *** dmsimard is now known as dmsimard|off | 04:10 | |
| *** esberglu has quit IRC | 04:13 | |
| *** udesale has joined #openstack-ansible | 04:13 | |
| *** Dinesh_Bhor has joined #openstack-ansible | 04:14 | |
| *** thorst has joined #openstack-ansible | 04:25 | |
| *** hybridpollo has quit IRC | 04:28 | |
| *** thorst has quit IRC | 04:29 | |
| *** markvoelker has joined #openstack-ansible | 04:44 | |
| *** udesale has quit IRC | 04:48 | |
| *** udesale has joined #openstack-ansible | 04:49 | |
| *** ecelik has joined #openstack-ansible | 05:00 | |
| *** Taseer has joined #openstack-ansible | 05:01 | |
| *** esberglu has joined #openstack-ansible | 05:03 | |
| *** mwe1 has quit IRC | 05:06 | |
| *** esberglu has quit IRC | 05:08 | |
| *** mwe1 has joined #openstack-ansible | 05:08 | |
| *** markvoelker has quit IRC | 05:17 | |
| *** thorst has joined #openstack-ansible | 05:25 | |
| *** sxc731 has joined #openstack-ansible | 05:32 | |
| *** sxc731 has quit IRC | 05:34 | |
| *** sxc731 has joined #openstack-ansible | 05:39 | |
| *** jvidal has joined #openstack-ansible | 05:44 | |
| *** esberglu has joined #openstack-ansible | 05:58 | |
| *** esberglu has quit IRC | 06:02 | |
| *** Oku_OS-away is now known as Oku_OS | 06:08 | |
| *** sxc731 has quit IRC | 06:12 | |
| masber | good afternoon all, is openstack-ansible ready for pike? | 06:14 |
|---|---|---|
| *** markvoelker has joined #openstack-ansible | 06:14 | |
| *** armaan has joined #openstack-ansible | 06:28 | |
| *** markvoelker has quit IRC | 06:48 | |
| *** esberglu has joined #openstack-ansible | 06:52 | |
| *** armaan has quit IRC | 06:52 | |
| *** thorst has quit IRC | 06:52 | |
| *** armaan has joined #openstack-ansible | 06:52 | |
| *** sxc731 has joined #openstack-ansible | 06:54 | |
| *** ivveh has joined #openstack-ansible | 06:54 | |
| *** ivveh- has quit IRC | 06:56 | |
| *** esberglu has quit IRC | 06:56 | |
| *** sxc731 has left #openstack-ansible | 06:57 | |
| *** armaan has quit IRC | 07:00 | |
| *** armaan has joined #openstack-ansible | 07:01 | |
| *** flaviosr has joined #openstack-ansible | 07:02 | |
| *** pcaruana has joined #openstack-ansible | 07:07 | |
| *** mbuil has joined #openstack-ansible | 07:11 | |
| *** chhavi has quit IRC | 07:28 | |
| masber | what I mean is, is it possible to install openstack pike through openstack-ansible? | 07:42 |
| *** chyka has joined #openstack-ansible | 07:43 | |
| *** markvoelker has joined #openstack-ansible | 07:45 | |
| *** esberglu has joined #openstack-ansible | 07:46 | |
| *** chyka has quit IRC | 07:48 | |
| *** thorst has joined #openstack-ansible | 07:49 | |
| *** esberglu has quit IRC | 07:50 | |
| *** chyka has joined #openstack-ansible | 07:53 | |
| *** pbandark has joined #openstack-ansible | 07:53 | |
| *** shardy has joined #openstack-ansible | 07:55 | |
| *** gkadam is now known as gkadam-afk | 07:55 | |
| *** chyka has quit IRC | 07:57 | |
| *** mrch has joined #openstack-ansible | 08:01 | |
| *** mpranjic has quit IRC | 08:06 | |
| *** ivveh has quit IRC | 08:13 | |
| *** ivveh has joined #openstack-ansible | 08:16 | |
| *** markvoelker has quit IRC | 08:18 | |
| *** armaan has quit IRC | 08:25 | |
| *** esracelik has joined #openstack-ansible | 08:25 | |
| *** ecelik has quit IRC | 08:26 | |
| *** armaan has joined #openstack-ansible | 08:26 | |
| *** armaan has quit IRC | 08:31 | |
| *** armaan has joined #openstack-ansible | 08:32 | |
| *** esracelik is now known as ecelik | 08:33 | |
| *** esberglu has joined #openstack-ansible | 08:40 | |
| hwoarang | masber: yes it is | 08:40 |
| *** gkadam-afk is now known as gkadam | 08:41 | |
| *** armaan has quit IRC | 08:44 | |
| *** udesale__ has joined #openstack-ansible | 08:44 | |
| *** esberglu has quit IRC | 08:44 | |
| *** armaan has joined #openstack-ansible | 08:44 | |
| *** udesale has quit IRC | 08:46 | |
| *** armaan has quit IRC | 08:50 | |
| *** armaan has joined #openstack-ansible | 08:50 | |
| *** udesale has joined #openstack-ansible | 08:52 | |
| *** udesale__ has quit IRC | 08:54 | |
| *** sxc731 has joined #openstack-ansible | 08:57 | |
| *** sxc731 has left #openstack-ansible | 08:57 | |
| *** thorst has quit IRC | 09:07 | |
| *** markvoelker has joined #openstack-ansible | 09:15 | |
| *** electrofelix has joined #openstack-ansible | 09:18 | |
| *** chyka has joined #openstack-ansible | 09:19 | |
| *** chyka has quit IRC | 09:23 | |
| *** pbandark1 has joined #openstack-ansible | 09:30 | |
| *** pbandark has quit IRC | 09:30 | |
| *** pbandark1 is now known as pbandark | 09:30 | |
| *** yifei has quit IRC | 09:34 | |
| *** esberglu has joined #openstack-ansible | 09:34 | |
| *** esberglu has quit IRC | 09:38 | |
| *** chhavi has joined #openstack-ansible | 09:43 | |
| *** markvoelker has quit IRC | 09:48 | |
| *** drifterza has joined #openstack-ansible | 09:49 | |
| *** ejb1123_ has joined #openstack-ansible | 09:59 | |
| *** thorst has joined #openstack-ansible | 10:04 | |
| *** hw_wutianwei has quit IRC | 10:18 | |
| evrardjp | good morning everyone | 10:20 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts master: Add missing space in rpm command https://review.openstack.org/506724 | 10:27 |
| *** mpranjic has joined #openstack-ansible | 10:28 | |
| *** esberglu has joined #openstack-ansible | 10:28 | |
| *** esberglu has quit IRC | 10:33 | |
| *** udesale has quit IRC | 10:37 | |
| Taseer | hwoarang: Should this host not be localhost ? https://github.com/opnfv/releng-xci/blob/master/xci/file/aio/configure-opnfvhost.yml#L2 | 10:44 |
| Taseer | Why do we want to remove folders when we always spawn a new VM each time ? | 10:44 |
| hwoarang | Taseer: wrong channel? :) | 10:45 |
| *** markvoelker has joined #openstack-ansible | 10:45 | |
| hwoarang | this is about the OPNFV osa so lets discuss in the opnfv channel | 10:45 |
| Taseer | Yes, sorry | 10:46 |
| ivve | anyone else noticing the rabbitmq playbook fail on downloading the .deb package? | 10:55 |
| ivve | fatal: [infra1_rabbit_mq_container-c142f90e]: FAILED! => {"attempts": 3, "changed": false, "dest": "/opt/rabbitmq-server.deb", "failed": true, "msg": "Connection failure: timed out", "state": "absent", "url": "http://www.rabbitmq.com/releases/rabbitmq-server/v3.6.9/rabbitmq-server_3.6.9-1_all.deb"} | 10:56 |
| *** armaan has quit IRC | 10:56 | |
| ivve | manual download works but takes ages | 10:56 |
| *** armaan has joined #openstack-ansible | 10:57 | |
| odyssey4me | ivve if you download it and put it into that folder, then it'll accept it as-is and continue | 10:57 |
| odyssey4me | in that folder, with that name | 10:58 |
| ivve | i know | 10:59 |
| ivve | (it also needs a rename i guess) | 10:59 |
| *** sxc731 has joined #openstack-ansible | 10:59 | |
| ivve | guess the problem is rabbitmq.com in this case | 11:00 |
| odyssey4me | ivve you could also set 'rabbitmq_install_method: external_repo' in your user_variables.yml to make it use the rabbitmq apt repo instead of doing the download | 11:00 |
| ivve | aye | 11:01 |
| *** sxc731 has left #openstack-ansible | 11:06 | |
| openstackgerrit | Daniel Marks proposed openstack/openstack-ansible stable/ocata: Fix Barbican's health checks https://review.openstack.org/507057 | 11:07 |
| robcresswell | Hey all, is there any documentation for adding CORS to an AIO? The AIO docs themselves were perfect (wanted to try a more reliable setup then devstack). I'm not very familiar with ansible so trying to figure out if there's a tidy way to do CORS on most of the services. | 11:13 |
| *** thorst has quit IRC | 11:14 | |
| robcresswell | I'm aware I could just do it manually, but just wanted to check if there was any existing "turn on CORS" button | 11:14 |
| asettle | robcresswell: I don't think so... | 11:14 |
| asettle | Not from memory, anyway | 11:14 |
| asettle | I can check | 11:14 |
| asettle | odyssey4me do you remember? | 11:14 |
| *** d3n14l has joined #openstack-ansible | 11:14 | |
| robcresswell | It's not super urgent, no rush to look :) | 11:17 |
| odyssey4me | robcresswell I don't think we have something universal - but recently there was a facility added to allow header modification... I don't really know how to activate CORS so I can't really advise :/ | 11:18 |
| *** markvoelker has quit IRC | 11:18 | |
| robcresswell | odyssey4me: Ah fair enough. There's some OSLO middleware that lets you do it with a couple of lines per config file | 11:19 |
| robcresswell | Just wondered if there was anything in the existing templates to do it | 11:19 |
| d3n14l | I backported a fix for barbican in stable/pike to stable/ocata - could someone have a look? https://review.openstack.org/#/c/507057/ | 11:20 |
| d3n14l | I would really appreciate to have that merged and available for cherry-picking in ocata. | 11:21 |
| *** esberglu has joined #openstack-ansible | 11:22 | |
| *** clayton has quit IRC | 11:25 | |
| *** esberglu has quit IRC | 11:27 | |
| *** armaan has quit IRC | 11:30 | |
| *** clayton has joined #openstack-ansible | 11:30 | |
| *** armaan has joined #openstack-ansible | 11:30 | |
| *** ivve has quit IRC | 11:34 | |
| evrardjp | robcresswell: mmm that's interesting, maybe there is something we could do with a global user_ override for config template for convenience | 11:36 |
| evrardjp | but I need to know a little more about it | 11:36 |
| evrardjp | d3n14l: sure | 11:37 |
| ejb1123_ | How would I set vmware to be my hypervisor? | 11:43 |
| *** deadnull has joined #openstack-ansible | 11:45 | |
| *** smatzek has joined #openstack-ansible | 11:45 | |
| evrardjp | ejb1123_: for what part? | 11:49 |
| openstackgerrit | Albert Mikaelyan proposed openstack/openstack-ansible-os_designate master: Use openstack proto variable in defaults https://review.openstack.org/504125 | 11:51 |
| evrardjp | it looks, according to this old documentation https://docs.openstack.org/ocata/config-reference/compute/hypervisor-vmware.html , that you'd need to have nsx support in our neutron role. The rest is probably straightforward, as changes in the config files can be templated with an user override | 11:51 |
| *** d3n14l1 has joined #openstack-ansible | 11:57 | |
| *** d3n14l has quit IRC | 11:59 | |
| *** thorst has joined #openstack-ansible | 12:01 | |
| *** d3n14l1 has left #openstack-ansible | 12:01 | |
| *** thorst_ has joined #openstack-ansible | 12:04 | |
| *** thorst has quit IRC | 12:05 | |
| *** d3n14l has joined #openstack-ansible | 12:05 | |
| *** sxc731 has joined #openstack-ansible | 12:05 | |
| robcresswell | evrardjp: https://gist.githubusercontent.com/robcresswell/81cd93456e8a2c0481cede622051346b/raw/5c55c7311b764fc09ef74b34a2ff5615f37b25be/local.conf was what I was using for devstack previously. | 12:06 |
| robcresswell | evrardjp: You can see how minimal the CORS stuff is for dev. | 12:07 |
| *** sxc731 has left #openstack-ansible | 12:07 | |
| evrardjp | yeah | 12:07 |
| *** Taseer has quit IRC | 12:07 | |
| evrardjp | what would be the impact of [[post-config|$KEYSTONE_CONF]] ? | 12:07 |
| evrardjp | what file does it change? | 12:08 |
| evrardjp | I guess I'm up to reverse engineer devstack. | 12:08 |
| evrardjp | wow seems easy | 12:10 |
| evrardjp | robcresswell: I am surprised, shouldn't there be something on the webserver side too ? | 12:13 |
| evrardjp | or has devstack nothing like this? | 12:14 |
| logan- | i had a heck of a time getting cors going with glance for newton | 12:14 |
| logan- | http://paste.openstack.org/show/621839/ | 12:14 |
| evrardjp | when I am looking at that, I see it shouldn't be that hard to do exactly the same as devstack for keystone with an override https://github.com/openstack/openstack-ansible-os_keystone/blob/master/tasks/keystone_post_install.yml#L46 | 12:14 |
| evrardjp | logan-: oh let me see | 12:14 |
| evrardjp | robcresswell: it looks like it could be simple as defining your overrides like logan did, but with your content of devstack | 12:15 |
| *** markvoelker has joined #openstack-ansible | 12:16 | |
| logan- | yep | 12:16 |
| *** esberglu has joined #openstack-ansible | 12:16 | |
| evrardjp | I am double checking variable names | 12:16 |
| robcresswell | evrardjp: So that middleware does all the config for you afaik; it just pulls a couple of values from $SERVICE.conf | 12:17 |
| evrardjp | yes | 12:18 |
| robcresswell | evrardjp: https://github.com/openstack/keystone/blob/master/etc/keystone.conf.sample#L587 | 12:18 |
| evrardjp | well | 12:18 |
| evrardjp | so keystone_keystone_conf_overrides: ... nova_nova_conf_overrides: ... swift_swift_conf_overrides: | 12:18 |
| evrardjp | and to avoid repeating yourself, you could do a cors_override: .... and then use that into your keystone_keystone_conf_overrides ... | 12:18 |
| robcresswell | logan-: for UI work at least, you don't really need many of the headers, I've found. | 12:19 |
| robcresswell | Mostly just the domain and possibly method changes. | 12:20 |
| robcresswell | Thats just my use case, so far. | 12:20 |
| logan- | yeah robcresswell makes sense | 12:20 |
| *** MasterofJOKers_ has joined #openstack-ansible | 12:20 | |
| logan- | i mean as long as headers like 'origin' aren't missing from service defaults you should be in good shape | 12:20 |
| *** esberglu has quit IRC | 12:21 | |
| *** MasterofJOKers has quit IRC | 12:22 | |
| *** MasterofJOKers_ is now known as MasterofJOKers | 12:22 | |
| *** markvoelker has quit IRC | 12:23 | |
| *** markvoelker has joined #openstack-ansible | 12:24 | |
| odyssey4me | I would imagine that activating CORS is something many people want to do - so it makes sense for us to make it easier for people | 12:24 |
| odyssey4me | if it adds a tax to the environment, then perhaps it should be opt-in | 12:24 |
| odyssey4me | heh... | 12:25 |
| * odyssey4me just landed a nova patch | 12:25 | |
| odyssey4me | https://review.openstack.org/506188 | 12:25 |
| evrardjp | woot | 12:25 |
| evrardjp | robcresswell: want me to draft the user config file for you? | 12:26 |
| evrardjp | because it's exactly the same principle, just a different format :p | 12:27 |
| d3n14l | I just tried to create an encrypted volume in ocata (15.1.8), is it correct that encrypted volumes on a ceph backend are not yet supported?! | 12:27 |
| *** weezS has joined #openstack-ansible | 12:27 | |
| evrardjp | robcresswell: I am not sure your paste worked for ironic :p | 12:31 |
| robcresswell | evrardjp: Oh, yeah, thats just sitting there | 12:32 |
| evrardjp | haha | 12:33 |
| robcresswell | It was originally an ironic setup that I modified, and forgot to remove the ironic cors conf. | 12:33 |
| evrardjp | "hold my beer" | 12:33 |
| robcresswell | haha | 12:33 |
| robcresswell | It was only ever intended as a personal reference item, I just realised it was useful for the discussion :) | 12:33 |
| evrardjp | let me just finish that quick, I am not sure what to do for allow_headers | 12:34 |
| *** drifterza has quit IRC | 12:34 | |
| *** weezS has quit IRC | 12:35 | |
| *** gkadam has quit IRC | 12:36 | |
| evrardjp | robcresswell: https://gist.github.com/evrardjp/5f1ae8d1afdf7b23d3224bc55388adab | 12:38 |
| evrardjp | drop that into your user variables, then do the deploy of the openstack bits, to have everything configured. | 12:39 |
| evrardjp | I will put that into docs somewhere. | 12:39 |
| *** kenperkins has joined #openstack-ansible | 12:45 | |
| *** udesale has joined #openstack-ansible | 12:48 | |
| admin0 | have most done ocata -> pike ? | 12:49 |
| admin0 | any known issues or bugs or nuisances one should be aware of ? | 12:50 |
| evrardjp | see gates. (no pun to hardware drive manufacturer intended) | 12:52 |
| evrardjp | robcresswell: updated the gist with a fancier version if you want to be more modular with your overrides. | 12:56 |
| admin0 | evrardjp, quick review of: https://review.openstack.org/#/c/506325/ and https://review.openstack.org/#/c/506353/ | 12:56 |
| ejb1123_ | evrardjp: I want to use vmware for compute and storage. | 12:59 |
| ejb1123_ | I just want to replace vmware integrated openstack | 12:59 |
| evrardjp | ejb1123_: ok, so no plan to run OSA with kvm inside vmware :p | 12:59 |
| evrardjp | ejb1123_: well I am not sure who did that already | 13:00 |
| *** udesale has quit IRC | 13:00 | |
| evrardjp | but it's technically easier to do if you already have one available: you know what to configure in your nova, glance, and neutron configuration file | 13:00 |
| evrardjp | ejb1123_: the work to be done in openstack-ansible would simply support nsx, because I am not sure everything works without it | 13:00 |
| evrardjp | ejb1123_: tell me if I am wrong | 13:01 |
| mhayden | happy monday | 13:03 |
| ejb1123_ | Idk evrardjp I'm just getting started with deploying openstack | 13:05 |
| evrardjp | :) | 13:05 |
| evrardjp | ejb1123_: it's technically possible, but it will not work out of the box, because nobody worked on this yet. | 13:06 |
| ejb1123_ | Well great, I'm going to have a fun time. | 13:06 |
| openstackgerrit | Logan V proposed openstack/openstack-ansible master: Better CI detection in bootstrap-hosts https://review.openstack.org/507092 | 13:06 |
| evrardjp | mhayden: happy monday sir :) | 13:07 |
| evrardjp | ejb1123_: the good thing is that, for what's existing it's gonna be easy here. And what's not included can easily be done. That's the advantage | 13:08 |
| ejb1123_ | Why can't I just put my own nova.conf sections in ansible and have ansible inject them into the config? | 13:08 |
| odyssey4me | ejb1123_ that's what the config overrides are | 13:09 |
| *** woodard_ has quit IRC | 13:09 | |
| ejb1123_ | Where is config overrides documented? | 13:09 |
| *** woodard has joined #openstack-ansible | 13:09 | |
| odyssey4me | ejb1123_ https://docs.openstack.org/project-deploy-guide/openstack-ansible/pike/app-advanced-config-override.html | 13:09 |
| evrardjp | ejb1123_: yeah that's what will be easy | 13:09 |
| evrardjp | and like I said above, you should follow the appropriate documentation , which should help you know which override you can do | 13:10 |
| evrardjp | however, I am not sure on how to install nsx for neutron, which is a different thing, maybe it's easy, maybe it's not | 13:10 |
| *** esberglu has joined #openstack-ansible | 13:11 | |
| robcresswell | evrardjp: Ooh, fancy. Thanks. | 13:11 |
| ejb1123_ | Okay, thx. So if I want to keep my config version controlled should I just commit the etc folder to my own repo? | 13:12 |
| evrardjp | robcresswell: haha. Well, that's how I'd do it. But logan- just told me you may hit a few issues, with the https://github.com/openstack/openstack-ansible-os_horizon/blob/stable/newton/defaults/main.yml#L83 | 13:12 |
| odyssey4me | yup | 13:12 |
| robcresswell | evrardjp: This isn't for horizon work ;) | 13:13 |
| robcresswell | But good to know | 13:13 |
| evrardjp | oh yes, true YOU'VE CHANGED. | 13:13 |
| evrardjp | :p | 13:14 |
| robcresswell | haha | 13:14 |
| evrardjp | "no it's not you, it's me... " | 13:14 |
| evrardjp | I will stop with those terrible jokes now. | 13:14 |
| *** drifterza has joined #openstack-ansible | 13:15 | |
| evrardjp | so more seriously, I'd be happy to have your opinion on what's working and what's not | 13:15 |
| evrardjp | I think logan- might be interested by your findings, and with your horizon expertise it could be even better. :D | 13:16 |
| logan- | ^ yep any tweaks needed to make specific services work would be good to know | 13:16 |
| robcresswell | afaik, the only issue is the "direct" image upload | 13:17 |
| robcresswell | Since it goes direct to service rather than proxying the server (which itself is problematic for large images) | 13:17 |
| *** d3n14l has left #openstack-ansible | 13:18 | |
| evrardjp | robcresswell: what's your use case you are testing right now needing CORS then, if horizon is not really linked to your use case? | 13:19 |
| robcresswell | evrardjp: Super secret special coolness | 13:20 |
| openstackgerrit | Sashi Dahal proposed openstack/openstack-ansible master: working example of QNAP cinder integration https://review.openstack.org/506353 | 13:20 |
| robcresswell | (Basically just investigating UI options for composable clouds) | 13:20 |
| *** yifei has joined #openstack-ansible | 13:20 | |
| robcresswell | For standalone services. | 13:20 |
| robcresswell | Having CORS enabled makes it easier to do UI testing, instead of messing with proxy servers | 13:21 |
| evrardjp | ok, so still web UI based | 13:21 |
| evrardjp | yeah | 13:21 |
| evrardjp | understood | 13:21 |
| evrardjp | well | 13:21 |
| robcresswell | Yeah, thats the only reason you'd need CORS afaik. | 13:21 |
| evrardjp | you got the basic understanding of that | 13:21 |
| evrardjp | I mean | 13:21 |
| evrardjp | with the paste, you've got what needs to be done for each of the service, I trust you'd do the right thing for your use case ;) | 13:22 |
| *** yifei has quit IRC | 13:25 | |
| *** gouthamr has joined #openstack-ansible | 13:26 | |
| *** goldenfri has quit IRC | 13:27 | |
| *** lbragstad has joined #openstack-ansible | 13:28 | |
| *** gouthamr has quit IRC | 13:29 | |
| *** gouthamr has joined #openstack-ansible | 13:30 | |
| *** smatzek has quit IRC | 13:38 | |
| *** dave-mccowan has joined #openstack-ansible | 13:41 | |
| *** chhavi has quit IRC | 13:46 | |
| openstackgerrit | Markos Chandras (hwoarang) proposed openstack/openstack-ansible-lxc_hosts master: SUSE: Fix repo setup on container cache https://review.openstack.org/507106 | 13:48 |
| *** gmonteiro has joined #openstack-ansible | 13:49 | |
| *** jheroux has joined #openstack-ansible | 13:49 | |
| gmonteiro | good morning everyone | 13:49 |
| gmonteiro | Someone had a problem during deploy in pike in the deploy user images on glance stage with the error message 'Cloud default was not found'. ? | 13:50 |
| *** ecelik has quit IRC | 13:56 | |
| mgariepy | logan-, are you running cinder-volume in a container ? | 13:57 |
| admin0 | i am getting (1040, u'Too many connections')"], my.cnf has a computed cap of 1600 .which I already reached .. is there a way to override this from variables ? | 13:57 |
| mgariepy | admin0, /etc/openstack_deploy/user_variables.yml:galera_max_connections: 4800 | 13:57 |
| admin0 | aah .. i see it | 13:59 |
| *** smatzek has joined #openstack-ansible | 13:59 | |
| *** MasterofJOKers has quit IRC | 14:01 | |
| evrardjp | gmonteiro: let me check on our gates | 14:02 |
| evrardjp | mhayden: did you just see what admin0 and mgariepy has defined? ;) | 14:02 |
| *** MasterofJOKers has joined #openstack-ansible | 14:03 | |
| admin0 | huh .. what ? | 14:03 |
| admin0 | :) | 14:03 |
| mgariepy | evrardjp, ? | 14:04 |
| *** ivve has joined #openstack-ansible | 14:04 | |
| evrardjp | admin0: mhayden you have both defined galera_max_connections: 4800 ? | 14:05 |
| evrardjp | mgariepy: * | 14:05 |
| mgariepy | I had some issue a while back | 14:05 |
| admin0 | in my case 1600 was not enough, so i hiked it up to 3200 for now | 14:05 |
| mhayden | galera should be able to handle that many | 14:05 |
| mgariepy | galera-front-1 FRONTEND state:OPEN current_session:1498 | 14:06 |
| mgariepy | this is my current connection count. | 14:06 |
| admin0 | my controller here is 48 core, 256gb of ram each | 14:07 |
| armaan | logan-: Hello, I am looking at this commit https://github.com/openstack/openstack-ansible-os_keystone/commit/1c469f978bf7940acd474ff2c7da67f08bad741b | 14:08 |
| armaan | logan-: I suppose this is breaking minor version upgrade in Newton if this variable is set "haproxy_ssl_all_vips: true". | 14:09 |
| mgariepy | mines are 24, 2400 was computed in the template but at some point it wasn't enough. | 14:10 |
| mgariepy | any rabbitmq expert around ? | 14:10 |
| mgariepy | I have a controller upgraded to 16.04 and rabbit on it failed twice.. | 14:10 |
| admin0 | to take OSA to the next level, we need: integrated ELK {elasticsearch, logstash, kibana}, integrated dashboards{grafana/influx with collectd stats} and some sort of integrated monitoring+alerting | 14:11 |
| *** jwitko has joined #openstack-ansible | 14:11 | |
| admin0 | anyone has those already done that they can share etc | 14:11 |
| mgariepy | generating quite a lot of loggin from the other 2. | 14:12 |
| mgariepy | admin0, something like : https://github.com/openstack/openstack-ansible-ops/tree/master/cluster_metrics | 14:12 |
| mgariepy | grafana, influx telegraf | 14:13 |
| evrardjp | admin0: rackspace. It's a product even :p and yes we've put stuff in ops. | 14:13 |
| admin0 | ok .. 1/3 met .. i will try this .. what about ELK and monitoring/alerting :) | 14:13 |
| evrardjp | admin0: 3/3 for rackspace :p | 14:14 |
| evrardjp | we do have ELK :p | 14:14 |
| evrardjp | collectd is not used, but another thing is | 14:15 |
| evrardjp | but that's another topic. | 14:17 |
| evrardjp | admin0: FYI there is more and more community things starting around ops healthchecks | 14:17 |
| admin0 | well, the thing is .. OSA is getting good and good ( perhaps the best ) in terms of flexibility and options and the speed of things .. but just having it setup is just one part of the wheel .setup is kind of one time thing .. but operations is daily and ongoing thing .. so if there are tools that can be added .. it would be great | 14:18 |
| cloudnull | mornings | 14:18 |
| admin0 | mornings | 14:19 |
| openstackgerrit | Merged openstack/openstack-ansible stable/ocata: Fix Barbican's health checks https://review.openstack.org/507057 | 14:21 |
| *** vakuznet has joined #openstack-ansible | 14:24 | |
| *** gmonteiro has quit IRC | 14:24 | |
| admin0 | evrardjp, where can i find those healthchecks ? | 14:27 |
| admin0 | can those tools be officially requested in the form of feature-request/bug report ? | 14:27 |
| evrardjp | let's just split this conversation in what you can find right now, and where. | 14:28 |
| admin0 | ok | 14:29 |
| admin0 | so dashboard i will find in the ops repo .. i will try and report/fix bugs there | 14:29 |
| admin0 | elk might be the 3rd priority .. as log host is already there and grep -r works | 14:29 |
| evrardjp | For now, you have monitorstack that's a project started by mhayden and a few others to upstream tools for monitoring openstack. There are other tools like this in the world (nagios scripts, etc...) | 14:29 |
| evrardjp | One of those other tools is part of a rackspace product | 14:30 |
| openstackgerrit | Jesse Pretorius (odyssey4me) proposed openstack/openstack-ansible-specs master: Spec: Python Build/Install Process Simplification https://review.openstack.org/501296 | 14:30 |
| evrardjp | monitorstack is just a starting project | 14:30 |
| evrardjp | that's the first item | 14:30 |
| evrardjp | for monitorstack it's under the openstack umbrella, for rackspace toolkit, you need to talk to someone else, it's not up to this channel :p | 14:30 |
| *** gmonteiro has joined #openstack-ansible | 14:31 | |
| evrardjp | for a logging stack rackspace is taking the approach to upstream recently, but I am not aware of the current status of the upstreaming work. | 14:32 |
| evrardjp | if you are looking for a solution, as I said, rackspace is providing one. | 14:32 |
| evrardjp | and last item is Grafana | 14:32 |
| evrardjp | if you want influx + telegraph, cluster metrics would do the trick | 14:32 |
| evrardjp | iirc | 14:33 |
| admin0 | rackspace(NL) was not very keen to support (additional) bits on an already installed OSA cluster :) | 14:33 |
| admin0 | and we could not go all-in :) | 14:33 |
| admin0 | i will (try) to be happy with clustermetics and monitorstack :) | 14:33 |
| evrardjp | you'll have to write code for it | 14:34 |
| evrardjp | it's just started | 14:34 |
| *** woodard has quit IRC | 14:36 | |
| evrardjp | for monitorstack, you can ping mhayden, and for cluster metrics, you can ping cloudnull | 14:38 |
| cloudnull | o/ | 14:38 |
| admin0 | evrardjp, thanks :) will definately ping them | 14:42 |
| evrardjp | hey cloudnull! | 14:42 |
| evrardjp | I thought you had a few days off | 14:42 |
| evrardjp | already back to work? | 14:42 |
| admin0 | anyone uses rally on OSA .. is it best to install on the deploy node itself or one of the util containers ? | 14:42 |
| cloudnull | evrardjp: yup | 14:43 |
| cloudnull | first day back | 14:43 |
| evrardjp | you didn't come by Belgium :p | 14:43 |
| cloudnull | no. sorry | 14:43 |
| evrardjp | next time! | 14:43 |
| cloudnull | I did have a stella though | 14:43 |
| cloudnull | so we're all good | 14:43 |
| evrardjp | lol | 14:43 |
| evrardjp | or not | 14:43 |
| evrardjp | ;) | 14:43 |
| evrardjp | hope you enjoyed the trip :D | 14:46 |
| cloudnull | I did | 14:51 |
| cloudnull | it was fun | 14:51 |
| cloudnull | The final country count was: Germany, Netherlands, Czech, Poland, Austria. | 14:52 |
| cloudnull | anyone around that might want to give this a review/push https://review.openstack.org/#/c/499886/ | 14:53 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts stable/pike: Add reno for bridge-nf-call change https://review.openstack.org/506364 | 14:57 |
| admin0 | is it possible to modify API endpoints post-deployment ? | 15:07 |
| *** sxc731 has joined #openstack-ansible | 15:08 | |
| admin0 | or if I change it, it will add the new one and i have to manually remove the old ones ? | 15:08 |
| *** sxc731 has left #openstack-ansible | 15:09 | |
| logan- | mgariepy: yes @ cinder-volume in container. | 15:20 |
| mgariepy | logan-, do you have ntp running in that container ? | 15:21 |
| logan- | no | 15:21 |
| logan- | ntp cannot run in containers | 15:21 |
| *** woodard has joined #openstack-ansible | 15:21 | |
| mgariepy | was pulled as a recommend on the ceph pkg on my container. | 15:22 |
| mgariepy | :) | 15:23 |
| mgariepy | not saying it was doing something useful, but was running in the container for some reason. | 15:23 |
| logan- | yeah | 15:23 |
| logan- | it'll get pulled in, and ntpd might even start, but it won't be able to set any time | 15:24 |
| admin0 | is there a way to provide automated reverse dns to ips ? | 15:24 |
| admin0 | hmm. ignore | 15:24 |
| logan- | armaan: hmm yeah quite possibly | 15:27 |
| cloudnull | ohai logan-, did you have a chance to look further into the slowness issues? | 15:28 |
| admin0 | my nova host-keys dunno how they get bad .. migration fails with connection-refused .. isn't there a quick way to re-generate and add the nova keys without running the full nova playbook | 15:28 |
| cloudnull | any luck with a profiler? | 15:28 |
| logan- | o/ cloudnull wecome back | 15:28 |
| cloudnull | :D | 15:28 |
| logan- | have not worked on a profiler, but I worked on https://review.openstack.org/#/c/504804/ after we talked and it seems to have cut 15-20 mins out of the gate times | 15:30 |
| cloudnull | ah very nice! | 15:31 |
| *** Oku_OS is now known as Oku_OS-away | 15:33 | |
| *** pcaruana has quit IRC | 15:33 | |
| *** pcaruana has joined #openstack-ansible | 15:34 | |
| admin0 | i find this in one of the nova tasks playbook: when: | 15:34 |
| admin0 | - nova_recreate_keys | bool -- how do i pass this book to ansible run ? | 15:34 |
| admin0 | i think its not recreating and thus my keys are not fixed | 15:35 |
| logan- | yeah it worked surprisingly well.. just a stab in the dark though. i wonder what we could learn from some ansible profiling | 15:36 |
| *** fellypefca has joined #openstack-ansible | 15:36 | |
| *** chyka has joined #openstack-ansible | 15:36 | |
| logan- | admin0: you can use '-e' to pass extra vars to ansible | 15:37 |
| logan- | admin0: so -e nova_recreate_keys=yes | 15:37 |
| admin0 | :) | 15:37 |
| admin0 | thanks logan- | 15:37 |
| *** galstrom_zzz is now known as galstrom | 15:57 | |
| Tahvok | error: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x764429E6: import read failed(2). | 15:57 |
| Tahvok | Is it known? | 15:57 |
| Tahvok | The requested URL returned error: 503 Service Temporarily Unavailable | 15:57 |
| odyssey4me | well, it looks like today is zuul v3 implementation day evrardjp cloudnull logan- jmccrory | 15:57 |
| logan- | oh right | 15:58 |
| Tahvok | So it's expected for things to not pass? | 15:58 |
| odyssey4me | apparently we may need to implement 'required projects' like so http://git.openstack.org/cgit/openstack-infra/openstack-zuul-jobs/tree/zuul.yaml#n327 for our jobs so that z-c works as it did before, because in zuul v3 it's just a shim | 15:58 |
| admin0 | hmm.. still fails .. to test, i should be able to ssh as nova from any of the compute node to any of the other compute nodes ? | 15:58 |
| odyssey4me | it's very plausible that the way we're using it will still work, I think - it'll just be faster if we do that for the job | 15:59 |
| evrardjp | odyssey4me: ok that's important | 16:00 |
| cloudnull | ok. if shit hits the fan w/ zuul v3 I'll be around . | 16:06 |
| *** Dinesh_Bhor has quit IRC | 16:07 | |
| prometheanfire | haven't used reno in a while, is it just plain text or can '`' be used as code blocks? | 16:07 |
| cloudnull | yes. | 16:08 |
| prometheanfire | sgtm | 16:08 |
| cloudnull | :D | 16:08 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Run gate playbooks in parallel https://review.openstack.org/497742 | 16:10 |
| *** pcaruana has quit IRC | 16:15 | |
| openstackgerrit | Matthew Thode proposed openstack/openstack-ansible-os_keystone master: Add security headers to web accessable services. https://review.openstack.org/507189 | 16:17 |
| *** dachary has joined #openstack-ansible | 16:17 | |
| dachary | I'd like to use http://docs.ansible.com/ansible/latest/intro_dynamic_inventory.html#example-openstack-external-inventory-script but I don't know what is the expected content of /etc/ansible/openstack.yaml ? | 16:17 |
| * dachary digging https://docs.openstack.org/os-client-config/latest/ | 16:18 | |
| evrardjp | dachary: yes, just using the standard one provided by ansible, and then using your clouds.yaml should be good enough | 16:18 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Ensure container facts are set on container create https://review.openstack.org/507191 | 16:18 |
| dachary | I don't have any clouds.yaml ? The only thing I have is an openrc.sh with environment variables | 16:19 |
| dachary | https://docs.openstack.org/os-client-config/latest/reference/index.html mentions config_files but does not document it | 16:21 |
| dachary | the plot thickens :-) | 16:21 |
| *** gkadam has joined #openstack-ansible | 16:24 | |
| dachary | I have an example here which I could blindly mimic https://github.com/openstack-infra/shade/blob/master/shade/tests/unit/fixtures/clouds/clouds.yaml | 16:24 |
| dachary | ah, https://docs.openstack.org/python-openstackclient/latest/configuration/index.html#configuration-files this is better ! | 16:25 |
| *** chhavi has joined #openstack-ansible | 16:29 | |
| logan- | yup dachary that's the doc you need :) | 16:33 |
| dachary | a few hops but not too difficult either ;-) | 16:33 |
| dachary | logan-: evrardjp thanks | 16:33 |
| *** chhavi has quit IRC | 16:42 | |
| *** gkadam has quit IRC | 16:42 | |
| *** chhavi has joined #openstack-ansible | 16:43 | |
| *** chhavi has quit IRC | 16:43 | |
| *** vishwanathj has joined #openstack-ansible | 16:45 | |
| *** armaan has quit IRC | 16:48 | |
| *** armaan has joined #openstack-ansible | 16:49 | |
| *** drifterza has quit IRC | 16:51 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_sahara stable/pike: Add support for the openSUSE Leap distributions https://review.openstack.org/505826 | 16:51 |
| *** armaan has quit IRC | 16:53 | |
| *** electrofelix has quit IRC | 16:56 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_sahara stable/pike: Fix ssl https://review.openstack.org/505280 | 17:00 |
| *** drifterza has joined #openstack-ansible | 17:07 | |
| *** jvidal has quit IRC | 17:09 | |
| *** masber has quit IRC | 17:10 | |
| openstackgerrit | Merged openstack/openstack-ansible-os_heat stable/pike: Fix ssl https://review.openstack.org/505775 | 17:13 |
| *** drifterza has quit IRC | 17:32 | |
| *** drifterza has joined #openstack-ansible | 17:34 | |
| *** drifterza has quit IRC | 17:35 | |
| mgariepy | anyone around having a newton/xenial install ? and could reboot a ctrl for a test ? | 17:35 |
| mgariepy | 14.2.8, on xenial, rebooting a controller containers comes back without network. | 17:36 |
| admin0 | mgariepy: seen it before .. forgot how i solve it :D | 17:36 |
| mgariepy | ifdown/up lxcbr0 + restart of container fix it. | 17:36 |
| admin0 | manually rebooting the containers worked as well | 17:37 |
| admin0 | individually | 17:37 |
| mgariepy | stop/start of container didn't work in my case, dnsmasq wasn't working. | 17:37 |
| *** chyka has quit IRC | 17:39 | |
| *** chyka has joined #openstack-ansible | 17:39 | |
| admin0 | all my migration fail with: SSH permission denied (public key) .. can someone help ? i tried to regenerate also .. but it does not seem to work at all | 17:46 |
| logan- | how are you executing ssh | 17:47 |
| admin0 | i am not .. i am just pressing the migrate button | 17:47 |
| admin0 | logs give something like: ssh -o BatchMode=yes 10.250.2.38 mkdir -p /var/lib/nova/instances/ec4c81cf-db84-4a83-b5e2-9bfae9bf3bfe | 17:47 |
| admin0 | i login to every compute node . do a su - nova .. and then try to ssh <another compute node> . they should work, but in this case, all give permission denied | 17:47 |
| admin0 | in my other cluster, i can ssh as nova from every compute node to every other compute node .. | 17:48 |
| admin0 | in this one, struggling .. | 17:48 |
| logan- | how do the contents of /var/lib/nova/.ssh look | 17:48 |
| logan- | key present? authorized_keys present and populated? | 17:49 |
| admin0 | logan-: error here: https://gist.github.com/a1git/7caba4c90b5fcc3a9f92a32ca13242c1 | 17:51 |
| admin0 | pasting others | 17:51 |
| admin0 | i have bl-11 ( blade-11) and bl-9 ( blade-9) .. 9 can ssh 11, 11 gets permisison deined on 9 | 17:53 |
| admin0 | trying to migrate instances from 11 to 9 | 17:53 |
| admin0 | logan-: https://gist.github.com/a1git/d41726e0cbcaabf049760a32d0738e80 .. all keys on all hosts are populated | 17:54 |
| admin0 | hmm.. bl-9 does not have any other hosts on the authorized_keys entry | 17:55 |
| admin0 | it just has itself only | 17:55 |
| admin0 | few hosts have all entries on authorized_keys file | 17:56 |
| admin0 | most hosts only have themselves | 17:56 |
| admin0 | what does this do? openstack-ansible setup-openstack.yml --tags nova-key --limit compute_hosts — fix the keys ? | 17:57 |
| admin0 | would manually rm-rf the nova/.ssh folder force/trigger something ? | 17:59 |
| admin0 | ansible compute_hosts -m shell -a "wc -l /var/lib/nova/.ssh/authorized_keys" — this returns an uneven number count .. 1 6 3 2 4 7 8 | 18:00 |
| *** poopcat has joined #openstack-ansible | 18:01 | |
| logan- | which tag is this on | 18:01 |
| logan- | i will have to look at the task to see what it is doing | 18:01 |
| admin0 | stable/ocata | 18:01 |
| admin0 | i will step out for a bit | 18:01 |
| admin0 | wil be back in a few hours | 18:01 |
| xdfil | Hmmm | 18:01 |
| xdfil | migration issues on Ocata/ | 18:02 |
| admin0 | but yes, i have been struggling with the keys for a few days now | 18:02 |
| xdfil | oh its not the same as my issue | 18:02 |
| admin0 | i had one more feature request , to bring libvirt.conf under ansible-managed so that i can do maybe live migration | 18:02 |
| xdfil | My issues is the CPU features | 18:03 |
| xdfil | For some reason CMT is being a problem | 18:04 |
| xdfil | all my previous Flavors will not migrate | 18:04 |
| xdfil | But new Flavors will migrate | 18:04 |
| xdfil | I need to manually adjust instance metadata in SQL but I can't figure out the query | 18:05 |
| *** albertcard1 has joined #openstack-ansible | 18:05 | |
| *** poopcat has quit IRC | 18:06 | |
| odyssey4me | xdfil in order for an instance to live migrate, both compute nodes must have the same CPU capabilities, or you need to set nova to some sort of compatibility mode | 18:06 |
| xdfil | I'm blaming libvirt from UCA | 18:07 |
| xdfil | both nodes are identical | 18:07 |
| xdfil | but I notice on the new favors created post-Ocata the CPU mode=none and pre-Ocata CPU mode=host-model | 18:08 |
| *** mrch has quit IRC | 18:08 | |
| odyssey4me | the instance metadata regarding the libvirt cpu model to use is something that's taken from nova.conf as far as I know, so I'm not sure that modifying the DB will help | 18:08 |
| odyssey4me | yes, that's nothing to do with UCA libvirt -that was acode change in nova I think, or in our code | 18:08 |
| xdfil | ah | 18:09 |
| *** mrtenio has joined #openstack-ansible | 18:09 | |
| odyssey4me | I think we shifted to letting nova figure out the right cpu mode instead of forcing our own | 18:09 |
| fellypefca | Hey everyone :) | 18:09 |
| xdfil | well I am in a bit of a bind | 18:10 |
| odyssey4me | xdfil https://github.com/openstack/openstack-ansible-os_nova/commit/27872a8cad4e28788ab2879e31613fe9745f87f8 | 18:10 |
| odyssey4me | xdfil if you want to change it to use host-model again, you can just use the config override | 18:10 |
| fellypefca | I'm trying to run a OSA master but its broken three times in TASK [haproxy_server : Create haproxy service config files]. Is there some major issue or probably is my scenario? | 18:10 |
| xdfil | so this is different than using the nova.conf libvirt overrides | 18:11 |
| xdfil | ? | 18:11 |
| odyssey4me | xdfil hmm, I see that's in pike - not ocata | 18:11 |
| odyssey4me | xdfil yes, this is a nova.conf override | 18:11 |
| xdfil | that what I've been toying with until now, stuff in the [libvirt] section of nova.conf | 18:11 |
| xdfil | I will try this thanks | 18:12 |
| odyssey4me | xdfil what's the cpu_mode in nova.conf on that compute node? | 18:12 |
| xdfil | the only place I see cpu_mode is under [libvirt] | 18:13 |
| xdfil | its currently set to host=passthrough | 18:13 |
| xdfil | its currently set to host-passthrough | 18:14 |
| xdfil | I've tried setting every possible option | 18:14 |
| odyssey4me | just setting the setting may do nothing except for new instances | 18:15 |
| jwitko | you guys should take a break and come play a game of ##uno | 18:15 |
| odyssey4me | lol, I'm going to take a break and go home | 18:16 |
| odyssey4me | cheerio, night all | 18:16 |
| cloudnull | night odyssey4me | 18:16 |
| cloudnull | fellypefca: do you have an example of the error ? | 18:18 |
| xdfil | odyssey4me: thanks. I will continue trying to figure out how to change it in the SQL per instance | 18:18 |
| xdfil | these are production servers, I dont want to ask for a maintenance window because openstack reasons | 18:19 |
| xdfil | I assume if a resize the instances, that will fix them also | 18:19 |
| xdfil | but resize=reboot no? | 18:19 |
| cloudnull | yes | 18:20 |
| cloudnull | and they'll migrate to a new host. | 18:20 |
| xdfil | thanks cloundnull | 18:20 |
| xdfil | cloudnull* lol | 18:20 |
| cloudnull | can you do an offline migration ? | 18:20 |
| xdfil | Maybe, but I dont want to take them down | 18:20 |
| cloudnull | assuming that all you're looking to do is move to a new node? | 18:20 |
| cloudnull | oh. | 18:20 |
| xdfil | I ust want the peice of mind that I can move them if I need to | 18:21 |
| cloudnull | ok. | 18:21 |
| xdfil | cloudnull: missed you at PTG btw | 18:22 |
| cloudnull | yea. would have been great to be there. | 18:22 |
| xdfil | heard you were on a better vacation though | 18:22 |
| cloudnull | i was :D | 18:22 |
| xdfil | what can be better than being in a small room with a bunch of developers | 18:23 |
| xdfil | you totaly missed out on the train | 18:23 |
| xdfil | the train was amazing | 18:23 |
| cloudnull | I heard that was amazing! | 18:23 |
| xdfil | lol | 18:23 |
| xdfil | toot toot!!! | 18:24 |
| cloudnull | hotel next to active rail road tracks | 18:24 |
| cloudnull | sounds like a match made in heaven | 18:24 |
| cloudnull | so reading back. | 18:24 |
| cloudnull | new instances on new flavors migrate fine | 18:25 |
| cloudnull | but old ones not so much ? | 18:25 |
| xdfil | cloudnull: correct | 18:36 |
| *** weezS has joined #openstack-ansible | 18:40 | |
| cloudnull | if you recreate the old flavors does it fix it ? | 18:42 |
| cloudnull | if you create the flavors via the cli you can recreate the flavors using the old IDs | 18:43 |
| xdfil | oh there's an idea... Right now I've commented out the setting in nova.conf.j2 and i am re-running the os-nova playbook | 18:45 |
| xdfil | I'll try your suggestion next | 18:45 |
| xdfil | I've always been uncertain what would be the effect of deleting a flavor that is in use by running instances | 18:46 |
| xdfil | if their flavor doesn't exist, all migrations would fail right? | 18:46 |
| *** fellypefca has left #openstack-ansible | 18:49 | |
| xdfil | ive forgotten: which nova hosts need to be run together? or another way of saying it would be " is it ok to run os-nova with --limit=compute_hosts" | 18:52 |
| cloudnull | if you delete the flavor and have existing instances it will show a mismatch when you do an instance show or when looking at things through horizon | 18:53 |
| cloudnull | however if you delete it and then recreate it with the existing ID it totally fine | 18:53 |
| cloudnull | IDK if that fixes the migration issue | 18:53 |
| cloudnull | but I know it does not break anything | 18:54 |
| xdfil | thanks that's good to know | 18:54 |
| cloudnull | we used to do this a lot i in the osic when we changed resource allocations which was helpful so folks working in projects didn't have to change their automation . | 18:54 |
| cloudnull | yes. your limit should be fine | 18:55 |
| cloudnull | and if you just want to update config you should be able to use the tag "nova-config" | 18:55 |
| *** dave-mccowan has quit IRC | 18:56 | |
| xdfil | oh | 18:56 |
| cloudnull | `openstack-ansible os-nova-install.yml --limit nova_compute --tags nova-config | 18:56 |
| xdfil | so I notice by default the virt type is set to qemu | 18:56 |
| cloudnull | that should make it go | 18:56 |
| cloudnull | ah,. | 18:56 |
| cloudnull | what version are you running? | 18:57 |
| xdfil | I always assumed that libvirt was automatically going to leverage kvm when it is possible | 18:57 |
| xdfil | but the other night I was reading th nova config refrence docs and they do say to set the type to kvm for best performance | 18:57 |
| xdfil | I'm still unclear on that qemu vs kvm | 18:58 |
| cloudnull | software vs hardware | 18:58 |
| xdfil | So qemu doesn't automattically use kvm when possible? | 18:58 |
| xdfil | eww I need to fix that then | 18:58 |
| xdfil | I'm using newer CPU | 18:59 |
| cloudnull | https://review.openstack.org/#/c/496254/ | 18:59 |
| cloudnull | if you're on a version of ocata before that went in, there was a bug | 18:59 |
| xdfil | 15.1.8 | 19:00 |
| cloudnull | which always forced nova_virt_type: "qemu" | 19:00 |
| cloudnull | when trying to detected the virt type | 19:00 |
| xdfil | its the most recent tag | 19:00 |
| xdfil | I'll see what happens after this play runs | 19:00 |
| cloudnull | ok | 19:00 |
| xdfil | I removed all my libvirt config overrides | 19:00 |
| xdfil | this time | 19:00 |
| xdfil | I had stuff in there for migration bandwidth and such | 19:01 |
| xdfil | none of that stuff I had in there probably makes a difference | 19:01 |
| *** hybridpollo has joined #openstack-ansible | 19:01 | |
| cloudnull | xdfil: https://github.com/openstack/openstack-ansible-os_nova/blob/stable/ocata/tasks/nova_virt_detect.yml | 19:03 |
| cloudnull | vs - https://github.com/openstack/openstack-ansible-os_nova/blob/991d3295fc2f8ecb3cfd46f680ceb1816c763291/tasks/nova_virt_detect.yml | 19:03 |
| cloudnull | so 15.1.8 has the bug | 19:03 |
| cloudnull | andymccr: evrardjp: when is the next release? | 19:04 |
| xdfil | hmm so checkout stable/Ocata? | 19:04 |
| cloudnull | xdfil: if you checkout the head of the stable branch it'll work as expected. | 19:05 |
| cloudnull | you'll need to get the roles again (run bootstrap-ansible.sh) | 19:05 |
| xdfil | right, it makes me a little nervous I've had bad luck with checking out stable before | 19:06 |
| xdfil | I may have forgotten to bootstrap though | 19:06 |
| cloudnull | you could just clone that one role into place | 19:06 |
| xdfil | it was a few months back | 19:06 |
| cloudnull | `rm -rf /etc/ansible/roles/os_nova; git clone https://git.openstack.org/openstack/openstack-ansible-os_nova /etc/ansible/roles/os_nova; pushd /etc/ansible/roles/os_nova; git checkout stable/ocata; popd` | 19:08 |
| cloudnull | or something similar | 19:08 |
| *** hybridpollo has quit IRC | 19:09 | |
| xdfil | So you think i'm being over-paranoid about checking out stable? | 19:11 |
| cloudnull | no | 19:12 |
| xdfil | ok :) | 19:12 |
| cloudnull | there be dragons. :) | 19:12 |
| cloudnull | I mean it should be good to go | 19:12 |
| cloudnull | and there should be no issues with it | 19:12 |
| cloudnull | but if this is prod you might want to stick to a single sha/tag | 19:13 |
| cloudnull | in this case, to fix the one issue, i think its probably best to simply get the one updated role you need. | 19:13 |
| cloudnull | on the next tag upgrade you go to the deviation will be fixed for you assuming you run bootstrap-ansible.sh | 19:14 |
| *** hybridpollo has joined #openstack-ansible | 19:22 | |
| *** dave-mccowan has joined #openstack-ansible | 19:24 | |
| *** dave-mccowan has quit IRC | 19:32 | |
| openstackgerrit | Matthew Thode proposed openstack/openstack-ansible-os_keystone master: Add security headers to web accessable services. https://review.openstack.org/507189 | 19:33 |
| *** dave-mccowan has joined #openstack-ansible | 19:33 | |
| ivve | any of you guys got experience with extremly slow image uploads via horizon to glance using ceph as glance backend? i have enabled legacy mode as well as images_allow_location = true and glance-api.conf got show_image_direct_url = true as well.. but uploading a 2.4G raw image takes like 1 hour | 19:44 |
| cloudnull | only w/ horizon? | 19:45 |
| cloudnull | the cli is faster? | 19:45 |
| ivve | didn't have this issue in ocata | 19:45 |
| ivve | gonna try cli and export it out from ceph | 19:45 |
| ivve | (to compare) | 19:45 |
| cloudnull | ok | 19:45 |
| cloudnull | I've not seen any major slow downs | 19:45 |
| ivve | quite surprised myself | 19:46 |
| cloudnull | but I've also not tested pike in prod quite yet | 19:46 |
| ivve | its insanely slow | 19:46 |
| cloudnull | I have a cluster upgrade I'll be doning later this week and I'll try it out. | 19:46 |
| ivve | usually 2.4g goes really fast, even over 1gbit.. if i put it via cli it should be fast with 10gbit | 19:46 |
| ivve | trying now, sec | 19:46 |
| cloudnull | both are from the same workstation ? | 19:47 |
| cloudnull | cli and horizon from the same computer ? | 19:47 |
| ivve | nvm found the issue | 19:47 |
| ivve | :P | 19:47 |
| cloudnull | what was it ? | 19:48 |
| ivve | firewall for sure | 19:48 |
| cloudnull | ah . | 19:48 |
| ivve | scp is just as slow | 19:48 |
| cloudnull | oh. damn firewalls getting in the way of all that is fun :D | 19:48 |
| ivve | haha yea | 19:49 |
| ivve | :D | 19:49 |
| ivve | wireless network is also shit | 19:51 |
| ivve | :D | 19:51 |
| ivve | damn i have stuff in my way here | 19:51 |
| ivve | [ 3] 0.0-11.4 sec 1.50 MBytes 1.11 Mbits/sec | 19:51 |
| ivve | yeaaaaaaahh | 19:51 |
| *** pbandark has quit IRC | 19:51 | |
| ivve | i should have a lady following me all day at work tomorrow yelling shame like in GoT for blaming glance/ceph | 19:52 |
| cloudnull | hahaha | 19:52 |
| ivve | SHAME! | 19:52 |
| ivve | :( | 19:52 |
| cloudnull | lol | 19:53 |
| cloudnull | it happens | 19:53 |
| ivve | :D | 19:53 |
| ivve | now graphs are larger in the firewall *happy* | 19:53 |
| cloudnull | woot! | 19:53 |
| ivve | at least 80mbit :P | 19:53 |
| ivve | and ill finally see if i can get this stuff to run proper | 19:54 |
| admin0 | back | 19:59 |
| admin0 | logan-: ping | 19:59 |
| ivve | shieeeeeeeeeet | 20:01 |
| ivve | it works :D | 20:01 |
| ivve | now the hard test.. spawn 23908295 instances at once | 20:01 |
| ivve | holy shit, shame on me for doubting this shit :D | 20:03 |
| ivve | pike and ceph lumi works pretty well.. some test remain though! | 20:03 |
| cloudnull | sweet! | 20:05 |
| ivve | this time i have mtu's set to 9000. hopefully no bandwidth issue between instances | 20:10 |
| *** esberglu has quit IRC | 20:13 | |
| cloudnull | if you have had the mtu set to <9000 you may need to set the bridge interfaces, tap devices, etc to the proper MTU. just to make sure you've not got any instances not using 9000 | 20:16 |
| *** esberglu_ has joined #openstack-ansible | 20:16 | |
| *** acormier has joined #openstack-ansible | 20:18 | |
| *** esberglu_ has quit IRC | 20:21 | |
| *** albertcard1 is now known as poopcat | 20:21 | |
| cloudnull | ivve: https://cloudnull.io/2016/08/converting-veth-interfaces-to-jumbo-frames/ | 20:22 |
| cloudnull | ^ shameless plug | 20:22 |
| ivve | cloudnull: saved! thanks | 20:24 |
| *** vakuznet has quit IRC | 20:24 | |
| ivve | however i install with overridden mtu values at 9000 (this time) and it looks good on the hosts | 20:25 |
| *** mbuil has quit IRC | 20:26 | |
| cloudnull | good | 20:27 |
| *** smatzek has quit IRC | 20:29 | |
| *** esberglu has joined #openstack-ansible | 20:29 | |
| *** armaan has joined #openstack-ansible | 20:42 | |
| *** tasker has joined #openstack-ansible | 20:43 | |
| tasker | I'm going over the "newton/upgrade-guide/manual-upgrade" guide and wondered if the order of the upgrade scripts matters. must I do them in the order written in that document or can I alter the steps some? | 20:44 |
| mhayden | logan-: did you want to change your vote on https://review.openstack.org/490222 ? | 20:45 |
| mhayden | there's a reno proposed now | 20:45 |
| tasker | for example, can I run "hostanme-compatibility" and "lxc-container-tz-sync" prior to configuring the galera containers? | 20:45 |
| mhayden | logan-: same thing on https://review.openstack.org/490223 | 20:45 |
| -openstackstatus- NOTICE: The infra team is bringing Zuul v3 online; expect service disruptions and please see https://docs.openstack.org/infra/manual/zuulv3.html for more information. | 20:49 | |
| * mhayden wheees | 20:50 | |
| prometheanfire | ithappening | 20:51 |
| *** woodard has quit IRC | 20:52 | |
| mhayden | https://media.giphy.com/media/5mBE2MiMVFITS/giphy-downsized.gif | 20:55 |
| * mhayden thinks he did that right | 20:55 | |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-pip_install stable/pike: Optimize pip_install for CentOS https://review.openstack.org/507271 | 21:02 |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-openstack_hosts stable/pike: Optimize openstack_hosts for CentOS https://review.openstack.org/507273 | 21:04 |
| *** gouthamr has quit IRC | 21:04 | |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-tests stable/pike: [TEST] Test RPM repo improvements https://review.openstack.org/507274 | 21:07 |
| *** armaan has quit IRC | 21:08 | |
| *** thorst_ has quit IRC | 21:19 | |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-openstack_hosts stable/pike: Optimize openstack_hosts for CentOS https://review.openstack.org/507273 | 21:22 |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-pip_install stable/pike: Optimize pip_install for CentOS https://review.openstack.org/507271 | 21:23 |
| logan- | done mhayden thanks | 21:24 |
| openstackgerrit | Major Hayden proposed openstack/openstack-ansible-pip_install master: Fix missing space in RPM command https://review.openstack.org/507279 | 21:25 |
| mhayden | logan-: woot | 21:25 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Ensure container facts are set on container create https://review.openstack.org/507191 | 21:26 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts stable/newton: Set kernel params to ensure iptables are never left open https://review.openstack.org/490223 | 21:32 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts stable/newton: Add reno for bridge-nf-call change https://review.openstack.org/506366 | 21:32 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts stable/ocata: Set kernel params to ensure iptables are never left open https://review.openstack.org/490222 | 21:32 |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts stable/ocata: Add reno for bridge-nf-call change https://review.openstack.org/506365 | 21:32 |
| cloudnull | anyone around that might want to give this a review https://review.openstack.org/#/c/499886/ | 21:32 |
| cloudnull | I'd love to get moving on these https://review.openstack.org/#/q/topic:hyperconverged+(status:open+OR+status:merged) if at all possible | 21:33 |
| *** jheroux has quit IRC | 21:36 | |
| *** thorst has joined #openstack-ansible | 21:39 | |
| *** jwitko has quit IRC | 21:41 | |
| *** dave-mccowan has quit IRC | 21:43 | |
| *** thorst has quit IRC | 21:45 | |
| *** shardy has quit IRC | 21:47 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Converge neutron agents onto Baremetal https://review.openstack.org/454450 | 21:55 |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: [WIP] Further hyper-converge our container usage https://review.openstack.org/499396 | 21:58 |
| *** john51 has joined #openstack-ansible | 21:59 | |
| *** woodard has joined #openstack-ansible | 22:00 | |
| *** john51_ has quit IRC | 22:01 | |
| *** thorst has joined #openstack-ansible | 22:06 | |
| *** gouthamr has joined #openstack-ansible | 22:11 | |
| *** rromans has quit IRC | 22:14 | |
| *** galstrom is now known as galstrom_zzz | 22:27 | |
| *** weezS has quit IRC | 22:28 | |
| *** acormier has quit IRC | 22:35 | |
| *** acormier has joined #openstack-ansible | 22:35 | |
| *** acormier has quit IRC | 22:40 | |
| *** dave-mccowan has joined #openstack-ansible | 22:44 | |
| *** rromans has joined #openstack-ansible | 22:44 | |
| *** chyka has quit IRC | 22:56 | |
| openstackgerrit | Kevin Carter (cloudnull) proposed openstack/openstack-ansible master: Run gate playbooks in parallel https://review.openstack.org/497742 | 23:00 |
| *** dave-mcc_ has joined #openstack-ansible | 23:05 | |
| openstackgerrit | Merged openstack/openstack-ansible-openstack_hosts stable/pike: Optimize openstack_hosts for CentOS https://review.openstack.org/507273 | 23:05 |
| *** mcarden has joined #openstack-ansible | 23:05 | |
| openstackgerrit | Merged openstack/openstack-ansible-pip_install master: Fix missing space in RPM command https://review.openstack.org/507279 | 23:05 |
| *** dave-mccowan has quit IRC | 23:07 | |
| *** acormier has joined #openstack-ansible | 23:14 | |
| *** acormier has quit IRC | 23:18 | |
| *** chyka has joined #openstack-ansible | 23:39 | |
| *** hybridpollo has quit IRC | 23:46 | |
| *** masber has joined #openstack-ansible | 23:48 | |
| *** weezS has joined #openstack-ansible | 23:49 | |
| *** weezS has quit IRC | 23:57 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!
