*** yuanying has quit IRC | 00:00 | |
*** yuanying has joined #openstack-kolla | 00:01 | |
*** Pavo has quit IRC | 00:01 | |
*** jamesPR has quit IRC | 00:02 | |
*** manheim has joined #openstack-kolla | 00:03 | |
*** manheim has quit IRC | 00:07 | |
*** jtriley has quit IRC | 00:08 | |
*** bjolo has quit IRC | 00:11 | |
*** Pavo has joined #openstack-kolla | 00:14 | |
*** salv-orlando has joined #openstack-kolla | 00:15 | |
*** caoyuan_ has quit IRC | 00:18 | |
*** salv-orlando has quit IRC | 00:19 | |
*** Pavo has quit IRC | 00:26 | |
*** hongbin has quit IRC | 00:27 | |
openstackgerrit | Steven Dake proposed openstack/kolla master: Consume centos-pike-testing mariadb version https://review.openstack.org/468632 | 00:32 |
---|---|---|
openstackgerrit | Steven Dake proposed openstack/kolla master: Consume centos-pike-testing mariadb version https://review.openstack.org/468632 | 00:32 |
sdake | kfox1111 inc0 sbezverk can you fellas ack https://review.openstack.org/#/c/464031/ and its child (the anisble version pin removal) thanks | 00:35 |
openstackgerrit | Steven Dake proposed openstack/kolla master: Consume centos-pike-testing mariadb version https://review.openstack.org/468632 | 00:41 |
*** zhubingbing has joined #openstack-kolla | 00:46 | |
*** cuongnv has joined #openstack-kolla | 00:48 | |
*** kbaegis has joined #openstack-kolla | 00:50 | |
*** jtriley has joined #openstack-kolla | 00:58 | |
*** chas has joined #openstack-kolla | 01:02 | |
*** masber has quit IRC | 01:06 | |
*** chas has quit IRC | 01:06 | |
*** tovin07_ has joined #openstack-kolla | 01:06 | |
*** masber has joined #openstack-kolla | 01:06 | |
*** kbaegis has quit IRC | 01:11 | |
*** hrw has quit IRC | 01:32 | |
Reepicheep | bmaltais: that is actually how it works. It took me way to long to figure that out | 01:34 |
*** hrw has joined #openstack-kolla | 01:34 | |
*** Pavo has joined #openstack-kolla | 01:35 | |
Reepicheep | Basically the globals overrides everything to simplify the kolla-ansible setup. But unfortunately that is not always what you expect | 01:36 |
Reepicheep | Inc0 had recommended to me to just handle all interfaces in the inventory file. | 01:37 |
Reepicheep | After doing that it works much better | 01:37 |
*** Pavo has quit IRC | 01:38 | |
Reepicheep | You can do cute stuff with host groups if you have a lot of duplicate parameters to set | 01:39 |
bmaltais | Reepicheep: Look like it is actually kolla-ansible thing: https://bugs.launchpad.net/kolla-ansible/+bug/1693598 | 01:39 |
openstack | Launchpad bug 1693598 in kolla-ansible "api_interface_address in group_vars always uses api_interface from globals.yml and ignores inventory setting" [High,Triaged] | 01:39 |
Reepicheep | bmaltais: yep | 01:39 |
bmaltais | Reepicheep: Workaround is to comment out those variables from globals.yml and define them in the inventory file/ | 01:40 |
bmaltais | Reepicheep: Glad I bumped in that bug report ;-) | 01:40 |
bmaltais | Reepicheep: The documentation could be clearer about that. | 01:40 |
Reepicheep | Well you got it faster than I did.. I was told the reason and still didn't understand it correctly for another day :/ | 01:41 |
bmaltais | Reepicheep: Darn. Talk about learning the hard way ;-) | 01:42 |
*** ddyer has joined #openstack-kolla | 01:45 | |
Reepicheep | bmaltais: The thing was I was getting the error with bootstrap-servers. I happen to get it working more by accident then understanding. | 01:46 |
Reepicheep | And thought I had it fixed | 01:46 |
bmaltais | Reepicheep: Alway happen. What I would like to clarify is do you need to specify those variables in every sections or just the 1st one where you define the node | 01:47 |
Reepicheep | But it was only for bootstrap-servers because it mainly cared about api_interface | 01:47 |
Reepicheep | And I happen to have that one commented out | 01:47 |
bmaltais | Reepicheep: Like chasing a wild goose. You think you get it but you don't | 01:48 |
Reepicheep | You should be able to do it with the host groups for ansible inventory | 01:48 |
bmaltais | Reepicheep: In all groups where the node is listed or just [compute] and [controller] | 01:49 |
bmaltais | Reepicheep: I guess I could figure it out by trial and error | 01:49 |
bmaltais | Reepicheep: SO far I only add it in the compute and controller groups | 01:50 |
Reepicheep | You can set group:vars for a group of hosts and apply those groups as children to other groups | 01:50 |
Reepicheep | On a technical stand point i don't know the answer to the repeat host in multiple groups question | 01:51 |
Reepicheep | Someone else may know | 01:52 |
bmaltais | Reepicheep: THis is what the top of my inventory file look like: http://paste.openstack.org/show/610916/ | 01:53 |
bmaltais | Reepicheep: How would you add the group:var in there to simplify it? | 01:53 |
Reepicheep | There is some good documentation for it, sorry I don't have the link handy. (I'm currently on my phone) | 01:53 |
bmaltais | OK | 01:54 |
*** Pavo has joined #openstack-kolla | 01:54 | |
Reepicheep | Yeah. I see what you are asking. Let me see if I can get the link | 01:55 |
Reepicheep | bmaltais: http://docs.ansible.com/ansible/intro_inventory.html#groups-of-groups-and-group-variables | 01:57 |
Reepicheep | Isn't technology crazy ;) | 01:57 |
bmaltais | Reepicheep: Cool, now it make sense! | 01:58 |
bmaltais | Reepicheep: So much easier when you can see a good example about the syntax | 01:58 |
*** rwsu has quit IRC | 02:01 | |
*** ipsecguy_ has joined #openstack-kolla | 02:04 | |
*** ipsecguy has quit IRC | 02:06 | |
*** Pavo has quit IRC | 02:08 | |
duonghq | zhubingbing, ping | 02:10 |
duonghq | can you review this: https://review.openstack.org/#/c/456513/ | 02:11 |
*** salv-orlando has joined #openstack-kolla | 02:16 | |
zhubingbing | sure | 02:18 |
*** salv-orlando has quit IRC | 02:22 | |
*** rwsu has joined #openstack-kolla | 02:51 | |
*** chas has joined #openstack-kolla | 03:03 | |
*** chas has quit IRC | 03:08 | |
duonghq | zhubingbing, thanks | 03:17 |
masber | good morning, I am planning to deploy designate to my kolla environment. My question is can I use same interface for dns_interface and neutron_external_interface? | 03:30 |
Reepicheep | masber: I do not believe you can do that | 03:32 |
*** janki has joined #openstack-kolla | 03:38 | |
Reepicheep | masber: afaik it will use the ip of the node that is already configured on the dns_interface | 03:41 |
Reepicheep | But I have yet to try it... it's on my list | 03:41 |
masber | yep, I assigned it same as the api_interface one | 03:42 |
masber | I am testing it now | 03:42 |
*** mandre has quit IRC | 03:55 | |
*** mandre has joined #openstack-kolla | 03:56 | |
*** jamesbenson has joined #openstack-kolla | 03:57 | |
*** jamesbenson has quit IRC | 03:57 | |
*** jamesbenson has joined #openstack-kolla | 03:58 | |
*** jamesbenson has quit IRC | 04:06 | |
*** jamesbenson has joined #openstack-kolla | 04:07 | |
*** salv-orlando has joined #openstack-kolla | 04:19 | |
*** jamesben_ has joined #openstack-kolla | 04:21 | |
*** bmaltais has quit IRC | 04:23 | |
*** salv-orlando has quit IRC | 04:23 | |
*** jamesben_ has quit IRC | 04:24 | |
*** jamesben_ has joined #openstack-kolla | 04:25 | |
*** jamesbenson has quit IRC | 04:25 | |
masber | Reepicheep, do you know by chance where should I put the designate-sink custom configuration folder? is it in the kolla or network nodes? | 04:27 |
*** jamesben_ has quit IRC | 04:29 | |
*** salv-orlando has joined #openstack-kolla | 04:39 | |
*** skramaja has joined #openstack-kolla | 04:39 | |
*** chas has joined #openstack-kolla | 05:04 | |
Reepicheep | masber: this may be helpful if you have not seen it: http://egonzalez.org/deploy-openstack-designate-with-kolla-ansible/ egonzalez has a good article on it | 05:07 |
*** chas has quit IRC | 05:09 | |
Reepicheep | egonzalez also mentioned the other day that he was wanting to work on some improvements regarding designate | 05:10 |
*** korzen has joined #openstack-kolla | 05:40 | |
*** unicell has quit IRC | 05:43 | |
*** coolsvap has joined #openstack-kolla | 05:52 | |
*** jaosorior_away is now known as jaosorior | 06:19 | |
*** salv-orlando has quit IRC | 06:19 | |
chandankumar | hello | 06:25 |
chandankumar | i am trying to setup kolla allinone using vagrant on fedora host machine | 06:25 |
chandankumar | i am getting error while mounting nfs https://paste.fedoraproject.org/paste/MC-PypsEk2jLEGLl9VJGOV5M1UNdIGYhyRLivL9gydE= | 06:27 |
chandankumar | Any help to troubleshoot it? | 06:27 |
masber | Reepicheep, thank you that web page really worked! I have designate working now | 06:34 |
*** jamesbenson has joined #openstack-kolla | 06:34 | |
*** jamesbenson has quit IRC | 06:38 | |
*** daidv has joined #openstack-kolla | 06:44 | |
*** matrohon has joined #openstack-kolla | 06:50 | |
*** shardy has joined #openstack-kolla | 07:05 | |
*** salv-orlando has joined #openstack-kolla | 07:11 | |
*** jamesbenson has joined #openstack-kolla | 07:16 | |
*** jamesbenson has quit IRC | 07:20 | |
*** salv-orl_ has joined #openstack-kolla | 07:27 | |
*** salv-orlando has quit IRC | 07:30 | |
*** pcaruana has joined #openstack-kolla | 07:33 | |
*** manheim has joined #openstack-kolla | 07:37 | |
*** pcaruana has quit IRC | 07:39 | |
*** pcaruana has joined #openstack-kolla | 07:39 | |
*** rmart04 has joined #openstack-kolla | 07:46 | |
*** egonzalez has joined #openstack-kolla | 07:47 | |
*** chas has joined #openstack-kolla | 07:47 | |
manheim | hello | 07:48 |
*** chas has quit IRC | 07:52 | |
*** chas has joined #openstack-kolla | 07:53 | |
hrw | morning | 07:53 |
egonzalez | sup | 07:54 |
openstackgerrit | Duong Ha-Quang proposed openstack/kolla stable/newton: Fix Creating Ceilometer MongoDB database failed https://review.openstack.org/456513 | 07:57 |
*** jamesbenson has joined #openstack-kolla | 07:57 | |
*** matrohon has quit IRC | 07:58 | |
*** tvignaud has quit IRC | 07:58 | |
hrw | duonghq: that's for case when mongodb complains that 'mongodb' user already exists? | 08:00 |
*** masber has quit IRC | 08:01 | |
*** jamesbenson has quit IRC | 08:02 | |
*** matrohon has joined #openstack-kolla | 08:03 | |
duonghq | hrw, no | 08:07 |
manheim | https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/neutron/templates/ml2_conf.ini.j2#L47 this should have an endif, no? | 08:14 |
*** tvignaud has joined #openstack-kolla | 08:16 | |
sdake | sup peeps | 08:16 |
sdake | egonzalez i foudn part of the problem - the extend-start script was bustola | 08:16 |
hrw | sdake: yo | 08:18 |
sdake | sup hrw | 08:18 |
*** openstackgerrit has quit IRC | 08:18 | |
sdake | manheim the endif is : https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/neutron/templates/ml2_conf.ini.j2#L70 | 08:23 |
manheim | oh, i see | 08:23 |
*** skramaja is now known as skramaja_afk | 08:25 | |
manheim | how can I negate a bool in j2? like enable_hyperv | bool? | 08:26 |
*** masber has joined #openstack-kolla | 08:26 | |
manheim | enable_hyperv !| bool? | 08:26 |
manheim | if that is not true | 08:26 |
egonzalez | manheim, not enable_hyperv | bool | 08:27 |
manheim | thanks | 08:27 |
*** reidrac has joined #openstack-kolla | 08:27 | |
*** athomas has joined #openstack-kolla | 08:27 | |
*** reidrac has quit IRC | 08:29 | |
egonzalez | sdake, added a couple of comments. I think the loop was never working as expected :/ | 08:32 |
*** n0isyn0i1e is now known as n0isyn0ise | 08:33 | |
*** pbourke has quit IRC | 08:35 | |
*** pbourke has joined #openstack-kolla | 08:35 | |
chandankumar | sdake: hello | 08:38 |
chandankumar | sdake: i am trying to setup vagrant kolla dev env on fedora 25 | 08:38 |
chandankumar | sdake: https://docs.openstack.org/developer/kolla-ansible/vagrant-dev-env.html | 08:38 |
chandankumar | sdake: i am hitting this error: https://paste.fedoraproject.org/paste/l2A9BchoERgHPesDnEnPqV5M1UNdIGYhyRLivL9gydE= | 08:38 |
chandankumar | sdake: and there is no mountd.service | 08:39 |
sdake | egonzalez agreed | 08:39 |
*** jamesbenson has joined #openstack-kolla | 08:39 | |
chandankumar | sdake: Any pointer to fix it? | 08:39 |
sdake | chandankumar i dont use vagrant sorry | 08:39 |
*** openstackgerrit has joined #openstack-kolla | 08:39 | |
openstackgerrit | Merged openstack/kolla master: Add ec2-api image https://review.openstack.org/468518 | 08:39 |
chandankumar | sdake: whomt to catch? | 08:39 |
chandankumar | *whom | 08:39 |
duonghq | chandankumar, not many people use vagrant in Kolla afaik | 08:40 |
sdake | chandankumar mandre and pbourke are the two major consuemrs of that code afaik | 08:41 |
chandankumar | sdake: thanks will check with them | 08:41 |
*** reidrac has joined #openstack-kolla | 08:43 | |
*** jamesbenson has quit IRC | 08:43 | |
*** mgoddard has joined #openstack-kolla | 08:44 | |
mandre | chandankumar: it's been a long time I've used the vagrant environment, not since kolla-ansible was split out of the repo | 08:45 |
*** skramaja_afk is now known as skramaja | 08:46 | |
mandre | things have probably changed but I can try to help | 08:46 |
chandankumar | mandre: destroyed the env. | 08:46 |
chandankumar | mandre: setting up again | 08:46 |
reidrac | mandre, chandankumar seems broken now, I oppened an issue in LP as I'd like to help fixing it but I want to know if my suggestion is a good idea before starting | 08:47 |
mandre | chandankumar: haha, drastic | 08:47 |
reidrac | this one: https://bugs.launchpad.net/kolla/+bug/1693847 | 08:47 |
openstack | Launchpad bug 1693847 in kolla "Vagrant development env broken after repo split" [Undecided,New] | 08:47 |
reidrac | basically the bootstrap script expects kolla-ansible content to be in kolla repo; it should be easy to fix as I explain in the issue | 08:48 |
reidrac | (that's for master, Pike) | 08:48 |
chandankumar | mandre: reidrac couple of more fixes: dnf install @vagrant will install the vagrant , setting up nfs part https://developer.fedoraproject.org/tools/vagrant/vagrant-nfs.html | 08:49 |
chandankumar | mandre: reidrac we donot need vagrant-vbguest for libvirt setup | 08:49 |
reidrac | chandankumar: right, the docs need some work but that's not fundamentally broken I got it working withot the right instructions :) | 08:50 |
mandre | chandankumar: are you following docs at https://docs.openstack.org/developer/kolla-ansible/vagrant-dev-env.html? | 08:50 |
chandankumar | https://docs.openstack.org/developer/kolla-ansible/vagrant-dev-env.html | 08:51 |
chandankumar | YES | 08:51 |
chandankumar | sorry | 08:51 |
chandankumar | yes | 08:51 |
reidrac | I would appreciate comments on lp:1693847 | 08:52 |
mandre | reidrac: will look at it in a minute | 08:52 |
reidrac | mandre: thanks! | 08:53 |
chandankumar | reidrac: bug updated | 08:56 |
reidrac | chandankumar: thanks! | 08:57 |
mandre | chandankumar: looks like the vagrant rpm stopped pulling nfs-utils | 09:00 |
*** powerd has joined #openstack-kolla | 09:03 | |
manheim | egonzalez about this https://review.openstack.org/#/c/455684/15/ansible/roles/nova-hyperv/templates/wsgate.ini.j2 | 09:06 |
mandre | reidrac: right, so if we consider kolla to be the container image building project, it doesn't make much sense to ship a vagrant environment, it should go in kolla-ansible IMO | 09:10 |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla master: Implement blazar Docker images https://review.openstack.org/454745 | 09:11 |
reidrac | mandre: I don't have a strong opinion as long as is not duplicated; my point to suggest it shuld go in kolla is that you can still use Vagrant to develop/build the image build part alone, but without that you can't run the deploy with ansible | 09:12 |
reidrac | oh, I guess that's actually a strong opinion then :) | 09:12 |
reidrac | basically: if you don't have kolla repo you can't really do anything with kolla-vagrant, that's why I suggested it should go in kolla repo | 09:13 |
reidrac | does it make sense? | 09:13 |
mandre | reidrac: agreed it's quite useful to be able to test the images with a deployment tool but probably we should do it the other way around, i.e have vagrant in kolla-ansible and make it depend on kolla | 09:14 |
*** zhubingbing has quit IRC | 09:15 | |
reidrac | mandre: OK, fair enough; can you add a comment on the ticket? I think I'll start to work on that now | 09:16 |
mandre | reidrac: see the documentation for the vagrant env was moved to kolla-ansible, I think the code counterpart was simply forgotten | 09:18 |
mandre | will add comment to the launchpad | 09:18 |
reidrac | mandre: thanks! | 09:19 |
*** jamesbenson has joined #openstack-kolla | 09:20 | |
manheim | how can I get the IP of the host ansible is running on? | 09:20 |
mandre | reidrac: hey, I've just noticed https://review.openstack.org/#/q/topic:fix-vagrant, probably this will be abandoned with your changes | 09:21 |
manheim | like this TASK...skipping: [192.168.10.14] | 09:21 |
manheim | and not from facts | 09:21 |
reidrac | mandre: good one, thanks | 09:22 |
manheim | i want to set the IP, the the IP will not be on an interface I can control | 09:23 |
manheim | it can be on eth0 or eth1 | 09:23 |
manheim | and I need to get that IP | 09:23 |
manheim | on windows there is just a list of interfaces in the gather phase | 09:23 |
*** jamesbenson has quit IRC | 09:24 | |
*** SamYaple has quit IRC | 09:27 | |
*** eglute has quit IRC | 09:27 | |
*** kwazar has quit IRC | 09:27 | |
*** afranc has quit IRC | 09:27 | |
*** sdake has quit IRC | 09:27 | |
*** egonzalez has quit IRC | 09:27 | |
*** ddyer has quit IRC | 09:27 | |
*** sbezverk has quit IRC | 09:27 | |
*** duonghq has quit IRC | 09:27 | |
*** esmiurium has quit IRC | 09:27 | |
*** jrist has quit IRC | 09:27 | |
*** timss has quit IRC | 09:27 | |
*** cmceniry has quit IRC | 09:27 | |
*** rbergeron has quit IRC | 09:27 | |
*** markmcclain has quit IRC | 09:29 | |
*** markmcclain has joined #openstack-kolla | 09:32 | |
openstackgerrit | Martin Mágr proposed openstack/kolla master: Sensu images https://review.openstack.org/463593 | 09:35 |
*** jamesbenson has joined #openstack-kolla | 09:41 | |
*** masber has quit IRC | 09:43 | |
*** sambetts_ is now known as sambetts | 09:45 | |
*** jamesbenson has quit IRC | 09:46 | |
*** eglute has joined #openstack-kolla | 09:49 | |
*** sdake has joined #openstack-kolla | 09:49 | |
*** kwazar has joined #openstack-kolla | 09:49 | |
*** SamYaple has joined #openstack-kolla | 09:49 | |
*** afranc has joined #openstack-kolla | 09:49 | |
*** sdake is now known as Guest12055 | 09:49 | |
*** common has quit IRC | 09:51 | |
openstackgerrit | Van Hung Pham proposed openstack/kolla master: Trivial fix typos in documents https://review.openstack.org/469031 | 09:52 |
*** common has joined #openstack-kolla | 09:53 | |
*** Serlex has joined #openstack-kolla | 09:54 | |
manheim | can someone take a look at this | 09:55 |
manheim | https://paste.ubuntu.com/24713405/ | 09:55 |
manheim | and i'll explain | 09:55 |
*** tovin07_ has quit IRC | 09:57 | |
*** lpetrut has joined #openstack-kolla | 10:06 | |
chandankumar | manheim: reidrac i have created a new step, just before proceeding with kolla | 10:06 |
chandankumar | manheim: reidrac will i use vagrantfile of kolla or kolla-ansible ? | 10:07 |
manheim | i've managed to do it | 10:08 |
chandankumar | manheim: with kolla or kolla-anisble? | 10:09 |
manheim | kolla-ansible | 10:09 |
manheim | let me explain | 10:09 |
chandankumar | ack | 10:10 |
*** mickt has joined #openstack-kolla | 10:10 | |
manheim | i am want to deploy nova on windows, and i need the IP of the windows node on the interface ansible uses | 10:10 |
manheim | but I do not know if it is the first or last interface | 10:10 |
manheim | so I try ot match the IP from inventory file with an IP from ansible interfaces from windows host | 10:11 |
*** egonzalez has joined #openstack-kolla | 10:13 | |
*** ddyer has joined #openstack-kolla | 10:13 | |
*** sbezverk has joined #openstack-kolla | 10:13 | |
*** duonghq has joined #openstack-kolla | 10:13 | |
*** esmiurium has joined #openstack-kolla | 10:13 | |
*** jrist has joined #openstack-kolla | 10:13 | |
*** timss has joined #openstack-kolla | 10:13 | |
*** cmceniry has joined #openstack-kolla | 10:13 | |
*** rbergeron has joined #openstack-kolla | 10:13 | |
*** daidv has quit IRC | 10:14 | |
*** mickt has quit IRC | 10:14 | |
*** mickt has joined #openstack-kolla | 10:14 | |
*** dims has quit IRC | 10:15 | |
*** kbyrne has quit IRC | 10:17 | |
*** dims has joined #openstack-kolla | 10:18 | |
*** kbyrne has joined #openstack-kolla | 10:18 | |
reidrac | chandankumar: I'm making the change in kolla-ansible, as per mandre recommendation | 10:19 |
chandankumar | manheim: reidrac got the env up with the same error https://paste.fedoraproject.org/paste/Sbl5~vQCeXr6BP3yOyqj815M1UNdIGYhyRLivL9gydE= | 10:19 |
reidrac | I've got a patch already, I'm testing and the docs should go after that | 10:19 |
chandankumar | reidrac: the same error is comign with kolla-ansible vagrant fole | 10:19 |
chandankumar | *file | 10:20 |
reidrac | when the PR is realy you may want to play with it ;) | 10:20 |
chandankumar | reidrac: sure | 10:20 |
reidrac | *ready | 10:20 |
*** cuongnv has quit IRC | 10:23 | |
*** ArchiFleKs has quit IRC | 10:37 | |
*** ArchiFleKs has joined #openstack-kolla | 10:41 | |
reidrac | chandankumar: all test pass, built some images and deployed some containers; I'll be back online in a bit and when the docs are updated I'll submit the PR | 10:41 |
*** dasTor has quit IRC | 10:42 | |
chandankumar | reidrac: sure, Thanks :-) | 10:43 |
pomac | Jeffrey4l: your storage driver did say "btrfs" right? Just want to make sure that me and bjolo are looking at the right things? | 10:44 |
*** duritong has quit IRC | 10:50 | |
*** duritong has joined #openstack-kolla | 10:51 | |
openstackgerrit | Sam Betts proposed openstack/kolla-kubernetes master: Remove inspector files and configs from ironic-conductor https://review.openstack.org/468403 | 10:56 |
*** reidrac has quit IRC | 11:03 | |
*** dims has quit IRC | 11:05 | |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla-ansible master: Add a Kolla 'devstack' mode https://review.openstack.org/454690 | 11:05 |
*** janki has quit IRC | 11:06 | |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla-ansible master: Add a Kolla 'devstack' mode https://review.openstack.org/454690 | 11:07 |
*** janki has joined #openstack-kolla | 11:07 | |
*** kbaegis has joined #openstack-kolla | 11:07 | |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla-ansible master: Add dev mode for Murano https://review.openstack.org/467608 | 11:10 |
*** powerd has quit IRC | 11:10 | |
*** powerd has joined #openstack-kolla | 11:10 | |
*** dims has joined #openstack-kolla | 11:11 | |
*** bmaltais has joined #openstack-kolla | 11:11 | |
*** rwellum has joined #openstack-kolla | 11:13 | |
*** ashishb has joined #openstack-kolla | 11:13 | |
ashishb | Hello Everyone! | 11:13 |
ashishb | I am trying Kolla and along side magnum and Kuryr | 11:14 |
ashishb | wondering if there are Kolla docker images for Kuryr? | 11:14 |
mickt | Hi, did anyone try upgrading nova yet? | 11:14 |
chandankumar | ashishb: https://github.com/openstack/kolla/tree/master/docker/kuryr | 11:15 |
openstackgerrit | Dan Ardelean proposed openstack/kolla-ansible master: Add Hyper-V role https://review.openstack.org/455684 | 11:17 |
ashishb | chandankumar: awesome! thanks, I'll surely try it out.. just checking if it creates both Kuryr controller and agent ..? | 11:17 |
chandankumar | ashishb: sorry never tried that | 11:18 |
ashishb | chandankumar: to be more specific, I am trying to get and deploy dockerize kuryr-controller and kuryr-cni driver | 11:19 |
ashishb | ok, in anycase, I'll give this a try thanks chandankumar for pointers | 11:20 |
kbaegis | Morning all | 11:25 |
rwellum | Morning kbaegis | 11:26 |
kbaegis | ashishb: I roll my own. The build scripts certainly allow for kuryr | 11:26 |
kbaegis | ashishb: They should be pullable | 11:26 |
ashishb | kbaegis: Thanks for the info, I see some like this on docker hub: https://hub.docker.com/r/kolla/ubuntu-source-kuryr-libnetwork/ | 11:29 |
ashishb | and https://hub.docker.com/r/kolla/ubuntu-source-kuryr-base/ | 11:29 |
ashishb | hope I am looking at right place to pull those | 11:29 |
openstackgerrit | Martin Mágr proposed openstack/kolla master: Sensu images https://review.openstack.org/463593 | 11:31 |
*** Serlex has quit IRC | 11:37 | |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla-ansible master: Add a new 'outward' rabbitmq instance https://review.openstack.org/453738 | 11:44 |
*** jamesbenson has joined #openstack-kolla | 11:45 | |
*** bjolo has joined #openstack-kolla | 11:46 | |
bjolo | howdy | 11:47 |
bjolo | bmaltais, i have also tried that i dont get it to work | 11:49 |
*** reidrac has joined #openstack-kolla | 11:49 | |
bjolo | supposedly you need to override it for each machine in the inventory file, but... does not seem to work for me | 11:49 |
*** jamesbenson has quit IRC | 11:49 | |
*** jtriley has quit IRC | 11:50 | |
bjolo | what we do is to create bond interfaces even for machines that only have 1 nic | 11:50 |
egonzalez | bjolo, need to override but cannot be any reference to the var in globals.yml, globals override everything | 11:50 |
bjolo | so i have to remove it from globals totaly? | 11:51 |
bjolo | and only have it in inventory? | 11:51 |
egonzalez | bjolo, yes, is a limitation | 11:52 |
bjolo | ok gotcha | 11:52 |
bjolo | tnx for the info | 11:52 |
bjolo | egonzalez, on another note. | 11:53 |
egonzalez | can group variables at inventory, or let the most common used at group_vars/all | 11:53 |
egonzalez | instead of globals | 11:53 |
bjolo | think it was yesterday you said something about creating subinterfaces when only having one nic. | 11:53 |
bjolo | it was in br-ext context | 11:53 |
bjolo | since br-ext needs to "own" the interface right | 11:54 |
egonzalez | yep | 11:54 |
bjolo | what type of subinterfaces do you use? | 11:54 |
bjolo | to me, subinterface means eth0:0, eth0:1, etc | 11:56 |
egonzalez | never done that before, I use VMs with as much virtual interfaces as needed, but saw some blogs about that | 11:56 |
bjolo | aha ok | 11:57 |
bjolo | cause i got a litte thrown off by that | 11:57 |
bjolo | in our setup we usually have 2 nics bonded to bond0 | 11:57 |
bjolo | then we create vlan interfaces on the bonds | 11:57 |
egonzalez | http://mattinaction.blogspot.co.uk/2014/04/a-working-openstack-configuration-for.html example | 11:57 |
bjolo | i.e. bond0.101, bond0.102, etc | 11:57 |
openstackgerrit | Juan J. Martínez proposed openstack/kolla-ansible master: Fix vagrant development environment https://review.openstack.org/469068 | 11:57 |
hrw | https://review.openstack.org/#/c/468348/ - can someone +w it? | 11:58 |
bjolo | mmm need to read thru that | 11:58 |
bmaltais | bjolo: Yes, comment all reference to network interface out and define them in the inventory file on the 1st appearance of the node name | 11:58 |
hrw | so gnocchi will build for centos/source | 11:59 |
reidrac | chandankumar: you have it; see PR 469068 | 11:59 |
bjolo | the problem we ran into when using vlan interfaces is that they all get the same MAC address by default | 11:59 |
bjolo | that does not work with neutron openvswitch since they use MAC addresses in the openflow rules | 11:59 |
bmaltais | bjolo: Like in: http://paste.openstack.org/show/610955/ | 12:00 |
egonzalez | hrm, have you tried changing the mac? | 12:00 |
bjolo | so we create unique MACs for each vlan interface | 12:00 |
chandankumar | reidrac: thanks, let me try that | 12:01 |
bmaltais | bjolo: Never tried eth0:0 interface syntax in inventory file. I am not sure how it would work... | 12:01 |
bjolo | so egonzalez when you said subinterface like eth0:0 it would not work in my mind, since subinterfaces share MAC | 12:02 |
bjolo | dont think you can change that afaik | 12:02 |
bmaltais | bjolo: In centos7 I saw that you can specify a MAC for a VLAN interface using nmtui | 12:02 |
bjolo | just wanted to check if we had missunderstood something, but it seems we all agree on this :) | 12:02 |
bmaltais | bjolo: Did you try that? | 12:03 |
bjolo | yes for vlan interface no problem | 12:03 |
bjolo | but for subinterface... no | 12:03 |
bmaltais | bjolo: I see. | 12:03 |
bmaltais | Got to go. Be back | 12:03 |
*** skramaja is now known as skramaja|afk | 12:06 | |
*** skramaja|afk is now known as skramaja_afk | 12:06 | |
*** janki has quit IRC | 12:06 | |
*** korzen has quit IRC | 12:23 | |
*** Serlex has joined #openstack-kolla | 12:24 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible master: DNM: test master branch https://review.openstack.org/456140 | 12:24 |
*** rmart04 has quit IRC | 12:29 | |
*** fguillot has joined #openstack-kolla | 12:30 | |
*** dave-mcc_ has joined #openstack-kolla | 12:36 | |
*** fguillot has quit IRC | 12:36 | |
*** chas has quit IRC | 12:44 | |
*** Pavo has joined #openstack-kolla | 12:45 | |
*** chas has joined #openstack-kolla | 12:45 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla master: Test fix rally build https://review.openstack.org/469077 | 12:46 |
*** chas has quit IRC | 12:49 | |
openstackgerrit | Paul Bourke (pbourke) proposed openstack/kolla-ansible master: Add a new 'outward' rabbitmq instance https://review.openstack.org/453738 | 12:52 |
*** goldyfruit has joined #openstack-kolla | 12:59 | |
*** fguillot has joined #openstack-kolla | 13:00 | |
openstackgerrit | Mathieu Rohon proposed openstack/kolla master: Add skydive service https://review.openstack.org/459995 | 13:05 |
*** Seyong has quit IRC | 13:05 | |
openstackgerrit | Mathieu Rohon proposed openstack/kolla-ansible master: Add skydive service https://review.openstack.org/460922 | 13:06 |
*** eanylin has quit IRC | 13:14 | |
*** lucasxu has joined #openstack-kolla | 13:14 | |
*** skramaja_afk is now known as skramaja | 13:20 | |
*** skramaja is now known as skramaja_ | 13:22 | |
*** salv-orlando has joined #openstack-kolla | 13:27 | |
*** rmart04 has joined #openstack-kolla | 13:27 | |
*** skramaja_ is now known as skramaja_afk | 13:28 | |
chandankumar | reidrac: manheim still facing the same error after applying your patch https://paste.fedoraproject.org/paste/JaRDhYoNQzqoc7m~3I-qGF5M1UNdIGYhyRLivL9gydE= | 13:28 |
*** jtriley has joined #openstack-kolla | 13:28 | |
chandankumar | tested on a clean env as well as previous env | 13:28 |
chandankumar | it is coming with vagrant libvirt centos setup | 13:29 |
chandankumar | s/centos/fedora | 13:29 |
*** salv-orl_ has quit IRC | 13:30 | |
reidrac | chandankumar: that's a differnt issue, probably related to NFS not properly installed | 13:31 |
*** rmart04 has quit IRC | 13:31 | |
reidrac | chandankumar: looks more related to your comment on the ticket; IMHO that should be dealt with in a different bug report | 13:31 |
chandankumar | reidrac: any help to debug that issue? | 13:32 |
reidrac | chandankumar: looks like the mount line explicitly uses nfs version 3 and is not supported by the NFS service | 13:33 |
openstackgerrit | Merged openstack/kolla master: Sensu images https://review.openstack.org/463593 | 13:33 |
reidrac | chandankumar: that mount line is generated by valgrant-libvirt plugin; which version is running? I'm on the latest one | 13:34 |
reidrac | chandankumar: docs say you need 0.0.31 or later (that's pretty much the latest one) | 13:35 |
chandankumar | reidrac: Installed the plugin 'vagrant-libvirt (0.0.40)'! | 13:36 |
reidrac | chandankumar: which host OS was that? may be there's a package you must install for NFS v3 support | 13:37 |
*** rmart04 has joined #openstack-kolla | 13:37 | |
chandankumar | reidrac: i am on fedora 25 | 13:37 |
*** chas has joined #openstack-kolla | 13:38 | |
reidrac | chandankumar: yum search nfs and search for version 3 support; my nfs package includes both nfs 3 and 4 | 13:38 |
*** chas_ has joined #openstack-kolla | 13:39 | |
chandankumar | reidrac: yup trying out | 13:41 |
reidrac | chandankumar: dnf rather yum :D | 13:41 |
*** Pavo has quit IRC | 13:41 | |
Reepicheep | bjolo: is there a problem with a subinterface having the same mac address? They are all the same interface so frames should end up forwarded to the correct port | 13:42 |
*** rmart04 has quit IRC | 13:42 | |
*** chas has quit IRC | 13:43 | |
reidrac | chandankumar: you can also check the docs, Fedora 25 is very current, I wouldn't be surprised if they introduced a change on that | 13:43 |
openstackgerrit | Merged openstack/kolla master: gnocchi: take care of centos/source on non-x86 https://review.openstack.org/468348 | 13:46 |
*** coolsvap has quit IRC | 13:47 | |
*** jrist has quit IRC | 13:48 | |
*** chas_ has quit IRC | 13:50 | |
*** chas has joined #openstack-kolla | 13:51 | |
*** jrist has joined #openstack-kolla | 13:51 | |
*** chas_ has joined #openstack-kolla | 13:52 | |
*** chas has quit IRC | 13:55 | |
*** GonZo2000 has joined #openstack-kolla | 14:00 | |
*** GonZo2000 has quit IRC | 14:00 | |
*** GonZo2000 has joined #openstack-kolla | 14:00 | |
kbaegis | inc0: What was the string to set the ansible version on the target? | 14:01 |
kbaegis | ansible_version or something? | 14:01 |
kbaegis | looks like my inventory got overwritten :/ | 14:02 |
*** salv-orlando has quit IRC | 14:02 | |
*** jamesPR has joined #openstack-kolla | 14:02 | |
jamesPR | hi all ;) | 14:02 |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla master: Implement blazar Docker images https://review.openstack.org/454745 | 14:02 |
*** zhubingbing has joined #openstack-kolla | 14:03 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla master: ODL installed for source images https://review.openstack.org/466336 | 14:04 |
kbaegis | Think I found it: ansible_python_interpreter? | 14:04 |
*** emccormick has joined #openstack-kolla | 14:05 | |
*** chas has joined #openstack-kolla | 14:05 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla master: Implement blazar Docker images https://review.openstack.org/454745 | 14:06 |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla master: Re-enable bifrost test_build https://review.openstack.org/465012 | 14:07 |
*** chas_ has quit IRC | 14:09 | |
*** eanylin has joined #openstack-kolla | 14:10 | |
*** jamesbenson has joined #openstack-kolla | 14:16 | |
*** GonZo2000 has quit IRC | 14:19 | |
*** rmart04 has joined #openstack-kolla | 14:19 | |
openstackgerrit | Sam Betts proposed openstack/kolla-kubernetes master: Remove inspector files and configs from ironic-conductor https://review.openstack.org/468403 | 14:20 |
*** eaguilar has joined #openstack-kolla | 14:21 | |
*** tovin07 has joined #openstack-kolla | 14:22 | |
*** rmart04 has quit IRC | 14:24 | |
*** rmart04 has joined #openstack-kolla | 14:24 | |
*** tovin07 has quit IRC | 14:26 | |
*** jamesbenson has left #openstack-kolla | 14:32 | |
*** jamesbenson has joined #openstack-kolla | 14:32 | |
*** skramaja_afk has quit IRC | 14:33 | |
*** tovin07 has joined #openstack-kolla | 14:37 | |
bjolo | Reepicheep, ovs openflow rules make use of interface mac addresses for source and destinations. so having same MAC for several interfaces does not work | 14:46 |
*** hongbin has joined #openstack-kolla | 14:48 | |
Reepicheep | bjolo: what is your use case of needing a subinterface that is manged by ovs? | 14:48 |
Reepicheep | I guess I'm not following | 14:48 |
bjolo | we use vlan interfaces with uniq macs on each interface | 14:48 |
bjolo | not subinterfaces | 14:48 |
Reepicheep | yeah.. a vlan interface is different then a subinterface | 14:49 |
bjolo | network_interface: bond0 | 14:49 |
bjolo | tunnel_interface bond0.101 | 14:49 |
bjolo | etc | 14:49 |
Reepicheep | yeah.. that is very similar to how I have mine setup | 14:49 |
kbaegis | Hey guys, is anyone else trying to run kolla-ansible from a venv? | 14:50 |
kbaegis | It's really the more appropriate/cleaner way to run | 14:50 |
bjolo | kbaegis, no but i have run it from a docker container | 14:50 |
kbaegis | bjolo: Do you have a dockerfile I could look at? | 14:51 |
Reepicheep | my vlan interfaces do have the same mac addresses and I'm using ovs | 14:51 |
*** magicboiz has joined #openstack-kolla | 14:52 | |
jamesbenson | Quick question, I'm doing redeploys, how do you un-initialize the fernet keys? I've run this command and no luck: docker kill keystone_fernet && docker stop keystone_fernet && docker rm keystone_fernet; rm -f /etc/kolla/keystone-fernet/* | 14:52 |
bjolo | Reepicheep, ok does networking work? i.e. boot up a few VMs, assign floating and ingress/egress trafic works? | 14:52 |
Reepicheep | yeah it's working for me | 14:53 |
bjolo | kbaegis, i did it manually just to try. just run a ubuntu/centos container and exec into it | 14:53 |
*** Pavo has joined #openstack-kolla | 14:54 | |
Reepicheep | bjolo: http://paste.openstack.org/show/610978/ that is my interfaces on a network node | 14:54 |
Reepicheep | the only difference is I'm using a team interface for internal stuff and the bond1 is my neutron_external_interface | 14:55 |
kbaegis | I'll be very happy when kolla is re-written without python2 | 14:55 |
bjolo | Reepicheep, whats your network setup? i have the follwong http://paste.openstack.org/show/610980/ | 14:56 |
Reepicheep | bjolo: in the past I have had my neuntron_external_interface set to a vlan interface "team.852" but I quit doing that because it limited me to one external (flat) network | 14:56 |
*** magicboiz has quit IRC | 14:57 | |
bjolo | Reepicheep, ok little different | 14:57 |
bjolo | i quess in our case it is network and neutron interfaces that clashes. they cant have the same macs | 14:58 |
Reepicheep | bjolo: I have never tried to have multiple neutron_external_interfaces | 14:58 |
bjolo | Reepicheep, it works :) | 14:58 |
Reepicheep | I see where ovs is getting confused for you now | 14:58 |
Reepicheep | just as you explained | 14:58 |
bjolo | to keep down on cost and cabling, i only have 2 nics at most in any machine | 14:58 |
bjolo | i mean 2x10Gb is sufficient for most needs no matter what | 14:59 |
bjolo | especially in a LACP bond | 14:59 |
bjolo | or team | 14:59 |
Reepicheep | that is very similar to what I do.. I just added an additional pair for the network nodes | 14:59 |
bjolo | but since neutron needs to own its interfaces, i have to create vlan interfaces and treat them as flat | 15:00 |
Reepicheep | now to be upfront I have just made that change and have yet to attempt a second external network | 15:00 |
Reepicheep | I was planning on setting mine to vlan type | 15:00 |
bjolo | that works too if you have the extra nics for it :) | 15:01 |
Reepicheep | but I was hoping to try it possibly today | 15:01 |
*** mickt has left #openstack-kolla | 15:01 | |
Reepicheep | I'm in the middle of moving to some new hardware and multiple external networks is one of the goals | 15:01 |
bmaltais | bjolo: Is there a conclusion regarding vlan interface and kolla-ansible? I am looking at a setup like yours... baremetal machines with 2 10GB Nic | 15:02 |
bmaltais | bjolo: If kolla-ansible can't deploy on that with vlan interface it is a bit of an issue then... | 15:03 |
Reepicheep | bjolo: it looks like as a team interface you can set MAC addresses I'm not sure on a bond | 15:03 |
bjolo | bmaltais, kolla can deploy. no problem at all. However, if you dont set uniq MACs on your interfaces when you have our type of setup you will see weird neutron behavior | 15:04 |
bjolo | trafic will go out on wrong interface etc | 15:04 |
bjolo | Reepicheep, you can set mac on vlan bond interfaces | 15:05 |
bmaltais | bjolo: OK... So perhaps bonding is the issue if it does not allow MAC specification per vlan | 15:05 |
bjolo | bmaltais, it does allow it, but it does not generate the macs by itself | 15:05 |
Reepicheep | bjolo: yeah I'm looking that up now | 15:06 |
bjolo | by default the vlan bond interface will have the same mac as the parent bond | 15:06 |
Reepicheep | how do you manage your interfaces? | 15:06 |
bjolo | with ansible of course :) | 15:06 |
Reepicheep | lol of course | 15:06 |
bjolo | haha | 15:06 |
bjolo | *should have said puppet just to stir it up :) * | 15:07 |
*** ducttape_ has joined #openstack-kolla | 15:07 | |
* Reepicheep is still no ansible expert | 15:07 | |
Pavo | have you guys thought about moving to ansible-container? | 15:07 |
Reepicheep | the funny thing is that I do use puppet to prep my machines before I even touch kolla :| | 15:08 |
bjolo | haha ok | 15:08 |
bjolo | whatever works for you and what you feel comfortable with :) | 15:08 |
Reepicheep | my thought is.. is that something you can automate by generating a bunch of macs and getting the assign correctly? | 15:09 |
*** korzen has joined #openstack-kolla | 15:09 | |
bjolo | yes | 15:09 |
Reepicheep | of course ;) | 15:10 |
*** ducttap__ has joined #openstack-kolla | 15:10 | |
bjolo | echo $* |md5sum|sed 's/^\(..\)\(..\)\(..\)\(..\)\(..\).*$/02:\1:\2:\3:\4:\5/' | 15:11 |
Reepicheep | I would guess there may not be a way around ovs's tie to using the mac address as identifing interfaces | 15:11 |
bjolo | all 02 macs are private | 15:11 |
Reepicheep | yeah.. I was thinkng generating those and droping them into some ansible paramiter somewhere and let ansible make sure each bond.xxx get's a uniq mac | 15:12 |
bjolo | http://paste.openstack.org/show/610983/ | 15:12 |
bjolo | we have a jinja template file for each vlan interface and use the ansible lookup command in the template | 15:13 |
*** ducttape_ has quit IRC | 15:13 | |
bjolo | feed the mac generator with fqdn and interface -> uniq for sure | 15:13 |
bjolo | i.e. persistent-mac.sh foo.bar.com bond0.123 | 15:14 |
bjolo | i.e. persistent-mac.sh foo.bar.com bond0.456 | 15:15 |
bjolo | i.e. persistent-mac.sh foo2.bar.com bond0.123 | 15:15 |
*** jamesben_ has joined #openstack-kolla | 15:17 | |
Reepicheep | bjolo: on a side note when you have multiple neutron_external_interfaces you just different --physical-network s when you do an openstack network create? | 15:17 |
bjolo | yes | 15:18 |
bjolo | pbourke wrote the code for it | 15:18 |
Reepicheep | that's simple enough.. I have not even thought about that | 15:18 |
*** jamesbenson has quit IRC | 15:18 | |
pbourke | most of it was vbel and jmcarthy, I just upstreamed it :) | 15:19 |
Reepicheep | I took the "I already have 4 interfaces at each node what's anouther pair of NICs for just two nodes" approach ;) | 15:19 |
bjolo | http://paste.openstack.org/show/610987/ | 15:19 |
bjolo | they then gets configured as physnet1, physnet2, etc | 15:20 |
Reepicheep | ok.. neutron_bridge_name also | 15:20 |
Reepicheep | does it also work with team interfaces or just bond interfaces? | 15:21 |
pbourke | Pavo: it was brought up before, i didnt like the idea | 15:21 |
Pavo | why so pbourke | 15:21 |
bjolo | on any compute node | 15:22 |
bjolo | http://paste.openstack.org/show/610988/ | 15:22 |
pbourke | Pavo: the Dockerfile syntax is standard, I dont see the benefit of rewriting everything just to be locked further into the ansible ecosystem | 15:22 |
bjolo | Reepicheep, i have hundreds of compute nodes. mostly HP blades. adding mezzanine cards and passthru switches would cost us a fortune | 15:23 |
Pavo | I can see that pbourke | 15:23 |
Pavo | was just wondering | 15:23 |
pbourke | Pavo: there may have been other thoughts on the blueprint also, it was a while back so I cant remember | 15:23 |
Reepicheep | bjolo: true.. I would try to do it with just two NICs in your situation | 15:24 |
*** manheim has quit IRC | 15:24 | |
Reepicheep | bjolo: that is a slick trick though .. I'm going to have to remember that | 15:24 |
bjolo | tnx :) | 15:25 |
Pavo | I currently only have 4 lines of code in my Dockerfiles that copy a ansible play book inside and then runs the roles inside during build | 15:25 |
*** tovin07 has left #openstack-kolla | 15:25 | |
inc0 | good morning | 15:30 |
Pavo | morning inc0 | 15:31 |
jmorgan1 | good morning | 15:31 |
egonzalez | inc0, can you take a look at https://review.openstack.org/#/c/468632/ | 15:32 |
*** zhubingbing has quit IRC | 15:32 | |
egonzalez | morning btw | 15:32 |
inc0 | egonzalez: gates are pretty red for one | 15:33 |
egonzalez | there is a bug in neutron too atm :( | 15:33 |
inc0 | and something strange happened to our bootstrap servers it seems | 15:34 |
egonzalez | related neutron bug https://bugs.launchpad.net/neutron/+bug/1694420 | 15:35 |
openstack | Launchpad bug 1694420 in neutron "AttributeError: 'NoneType' object has no attribute 'port_security_enabled" [Critical,Confirmed] | 15:35 |
inc0 | ah, no, it's prechecks | 15:35 |
inc0 | our prechecks are wrong | 15:35 |
*** lucasxu has quit IRC | 15:36 | |
*** lucasxu has joined #openstack-kolla | 15:37 | |
egonzalez | inc0, No module named jinja2? | 15:37 |
inc0 | yup | 15:37 |
*** Pavo has quit IRC | 15:37 | |
inc0 | we don't need jinja2 on nodes other than deploy | 15:38 |
inc0 | and jinja2 is requirement in ansible | 15:38 |
inc0 | so yeah...no need for this precheck | 15:39 |
Guest12055 | sup peeps | 15:40 |
Guest12055 | whisky tango foxtrot | 15:40 |
*** Guest12055 is now known as sdake | 15:40 | |
*** sdake has joined #openstack-kolla | 15:40 | |
*** zhubingbing has joined #openstack-kolla | 15:40 | |
openstackgerrit | Eduardo Gonzalez proposed openstack/kolla-ansible master: Revert "Add jinja2-py version check in pre_checks" https://review.openstack.org/469167 | 15:41 |
sdake | inc0 can you ack https://review.openstack.org/#/c/468672/ | 15:41 |
inc0 | thanks egonzalez | 15:42 |
sdake | inc0 can you ack https://review.openstack.org/#/c/464031/ | 15:42 |
sdake | inc0 love you long time on those acks :) | 15:42 |
sdake | one is easy one is hard :) | 15:42 |
sdake | what is this "equalto" message i see | 15:42 |
sdake | ansible 2.3.0.0 | 15:42 |
sdake | i think its a jinja2 problem? | 15:42 |
kbaegis | Hey guys, I'm able to run kolla-ansible inside a venv, but it hangs | 15:48 |
kbaegis | https://hastebin.com/epinigehiw.md | 15:48 |
sdake | inc0 can you w+ those changes oto plz | 15:48 |
kbaegis | https://hastebin.com/dacagabeya.md | 15:49 |
kbaegis | That's more relevant information | 15:49 |
inc0 | sry sdake I'm still hungry and before coffee;) | 15:50 |
reidrac | kbaegis: https://bugs.launchpad.net/kolla/+bug/1693847 ; there's a PR linked in one comment, give it a go | 15:53 |
openstack | Launchpad bug 1693847 in kolla "Vagrant development env broken after repo split" [Medium,In progress] - Assigned to Juan J. Martínez (jjmartinez) | 15:53 |
*** lucasxu has quit IRC | 15:54 | |
reidrac | kbaegis: uh, I was too quick; it may not be related; ignore me! | 15:54 |
sdake | inc0 after you have a red bull, really need your eyeballs on this: https://review.openstack.org/#/c/468632/12 | 15:55 |
sdake | inc0 mariadb is hopelessly broken even in older versions because of extend start behavior | 15:55 |
inc0 | sdake: and I need this;) | 15:55 |
inc0 | https://review.openstack.org/#/c/469167/ | 15:55 |
kbaegis | reidrac: :) Thanks | 15:56 |
inc0 | and I need everyones eyes on this https://review.openstack.org/#/c/466007/72 :) | 15:56 |
openstackgerrit | Merged openstack/kolla-kubernetes master: Remove pin on ansible<2.1 https://review.openstack.org/468672 | 16:00 |
*** tonanhngo has joined #openstack-kolla | 16:00 | |
openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes master: Add a destroy workflow for kolla-kubernetes https://review.openstack.org/464031 | 16:01 |
*** chas has quit IRC | 16:02 | |
trozet | sean-k-mooney: hi | 16:03 |
sean-k-mooney | hi | 16:03 |
trozet | sean-k-mooney: are you cool with this? https://review.openstack.org/#/c/467712/ | 16:03 |
*** chas has joined #openstack-kolla | 16:03 | |
*** matrohon has quit IRC | 16:03 | |
sean-k-mooney | trozet: i dont really see any advatage to this | 16:04 |
trozet | sean-k-mooney: it's just to isolate extensions/plugins from all being in the same container | 16:05 |
trozet | sean-k-mooney: to avoid one sub-project breaking all of the neutron server | 16:05 |
sean-k-mooney | yes but that is the opiste of how the plugins were ment to work in kolla | 16:05 |
trozet | sean-k-mooney: how was it meant to work? include all? | 16:05 |
sean-k-mooney | trozet: i understand your main concurn was db migration | 16:05 |
sean-k-mooney | trozet: enable specic plugins at build time | 16:06 |
trozet | sean-k-mooney: ah | 16:06 |
trozet | sean-k-mooney: this is mainly for tripleO, we want to isolate all the different plugins into their own neutron server container | 16:06 |
sean-k-mooney | trozet: https://github.com/openstack/kolla/blob/master/doc/image-building.rst#plugin-functionality | 16:06 |
trozet | sean-k-mooney: so if you think it doesnt quite fit the kolla architecture, then I guess we will need to build it int eh tripleo-common project or something | 16:07 |
sean-k-mooney | trozet: what happens if i want sriov and odl in that case | 16:07 |
*** chas has quit IRC | 16:07 | |
trozet | sean-k-mooney: SRIOV is in tree with neutron so that one is OK, but yeah your concern is valid for using multiple plugins at the same time in the future | 16:07 |
sean-k-mooney | trozet: well the intent with the plugins were that you would use a template overdide and the kolla-build.conf to define what plugins were enabled | 16:08 |
*** egonzalez has quit IRC | 16:09 | |
*** kbaegis has quit IRC | 16:10 | |
sean-k-mooney | trozet: ignoreing the ml2 mech driver i think you will also have problems with the advanced servics | 16:11 |
sean-k-mooney | looking at https://review.openstack.org/#/c/467712/3/docker/neutron/neutron-server-opendaylight/Dockerfile.j2 | 16:11 |
sean-k-mooney | your binbary install wont support lbaas for example | 16:11 |
sean-k-mooney | trozet: for tripple0 do you intend to use prebuilt contianer or will trippleO build the contianers as part of the deployment? | 16:15 |
*** korzen has quit IRC | 16:16 | |
*** akorzeni__ has joined #openstack-kolla | 16:16 | |
trozet | sean-k-mooney: yeah was going to add octavia as well | 16:17 |
trozet | sean-k-mooney: well we want to use prebuilt, but if kolla doesnt want this specific container then I guess we will build this one | 16:18 |
sean-k-mooney | trozet: well i dont speak for kolla so you should get input form the core team but for me i think the plugin mechanisum and tempalte overrides should allow you to do this without modifying the in tree container | 16:19 |
sean-k-mooney | trozet: sdake inc0 Jeffrey4l pbourke ... whould be better able to say if https://review.openstack.org/#/c/467712/3 makes sense form a kolla core team point of view | 16:21 |
*** akorzeni__ has quit IRC | 16:22 | |
*** jrich523 has joined #openstack-kolla | 16:22 | |
*** korzen has joined #openstack-kolla | 16:22 | |
*** eaguilar has quit IRC | 16:24 | |
jmorgan1 | trozet: is your patch opnfv specific? as there was some discussion on creating a sub-project wihtin kolla for networking related projects (like opnfv) to put code into | 16:26 |
*** zhubingbing has quit IRC | 16:27 | |
*** zhubingbing has joined #openstack-kolla | 16:27 | |
sean-k-mooney | jmorgan1: are you refering to scarab poc? | 16:27 |
jmorgan1 | sean-k-mooney: i don't think so, there was some discussion at openstack boston to create something like openstack/kolla-networking for opnfv and other netowkring specific projects to house kolla based support | 16:28 |
sean-k-mooney | jmorgan1: i would personally be more interested in a subproject for deploying infra components in general then just resticting to networking | 16:29 |
*** kbaegis has joined #openstack-kolla | 16:29 | |
*** salv-orlando has joined #openstack-kolla | 16:29 | |
jmorgan1 | sean-k-mooney: need to check with the boss (inc0) on getting this started | 16:29 |
sean-k-mooney | jmorgan1: yes i have spoken to him on and off about this. one of the questions is is it in kolla scope or not. arguabley it could in the scope of a subproject but i have always wondered should kolla be deploying mysql or ceph for example in main kolla/kolla-ansible/kolla-k8s | 16:31 |
*** david-lyle has joined #openstack-kolla | 16:32 | |
sean-k-mooney | jmorgan1: those peices need to be deployed by something and useing kollas architecture/tooling to deploy them the same way we deploy openstack makes sense but you need specalist non openstakck knolage to deploy them correctly | 16:32 |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes master: permit rabbitmq-mgmt to select rabbitmq as endpoint https://review.openstack.org/467771 | 16:33 |
openstackgerrit | Kevin Fox proposed openstack/kolla-kubernetes master: openvswitch not required to be kernel module https://review.openstack.org/467769 | 16:33 |
*** itlinux has joined #openstack-kolla | 16:34 | |
jmorgan1 | sean-k-mooney: sorry, not sure of the details as I'm still new to kolla, will need to check with Michal on his thoughts for implmentation | 16:36 |
*** tonanhngo has quit IRC | 16:36 | |
sean-k-mooney | kfox1111: in relation to https://review.openstack.org/#/c/467769/3 out of interest does kolla-k8s allow me to override the ovs-datapath type via the ml2_conf.ini? | 16:38 |
sean-k-mooney | kfox1111: if the ovs kernel module is not loaded this needs to be set to netdev | 16:38 |
*** rwallner has joined #openstack-kolla | 16:38 | |
kfox1111 | kolla-k8s doesn't have much by way of configmap stuff yet. it just is a copy of kolla-ansible's templates. | 16:39 |
kfox1111 | so, whatever kolla-ansible can do for config gen. | 16:39 |
*** kbaegis1 has joined #openstack-kolla | 16:40 | |
sean-k-mooney | kfox1111: does it support external config dirctory like ansible? | 16:40 |
*** kbaegis has quit IRC | 16:40 | |
kfox1111 | unfortunately not. The ansible soltuion involved ansible patching up the config files at the time its injected into the remote machine. we're not using that plumbing. | 16:40 |
sean-k-mooney | kfox1111: that is how you set that currently with kolla ansible. e.g. config override of [OVS] datapath_type=netdev | 16:41 |
kfox1111 | can tweak the file in /etc/kolla/neutron before uploading as a configmap though. | 16:41 |
sean-k-mooney | ok cool | 16:41 |
*** lpetrut has quit IRC | 16:42 | |
*** lucasxu has joined #openstack-kolla | 16:43 | |
sean-k-mooney | kfox1111: one last question on https://review.openstack.org/#/c/467769/3/helm/microservice/neutron-openvswitch-agent-daemonset/templates/openvswitch-agent-daemonset.yaml | 16:43 |
*** pcaruana has quit IRC | 16:43 | |
*** chas has joined #openstack-kolla | 16:43 | |
sean-k-mooney | are you trying to ensure the ovsdb is running or ovs-vswitchd? | 16:43 |
sdake | sean-k-mooney kolla-kubernetse (for the moment) does work with external config dirs | 16:44 |
sdake | sean-k-mooney i use it daily | 16:44 |
sdake | kfox1111 the ansible code retained config merging | 16:44 |
kfox1111 | sean-k-mooney: probably should check for both really. | 16:44 |
kfox1111 | sdake: interesting. maybe because its resolving to localhost? | 16:44 |
kfox1111 | it at least can be overridden that way...... hmm.... | 16:45 |
sdake | kfox1111 right - it only needs to resolve to localhost | 16:45 |
kfox1111 | cool. | 16:45 |
openstackgerrit | Christian Berendt proposed openstack/kolla master: In set_configs.py support perm parameter in permissions section https://review.openstack.org/468472 | 16:45 |
sean-k-mooney | ok so you should be able to congiure the neutron agent to use ovs in netdev mode so without the kernel | 16:45 |
sean-k-mooney | kfox1111: loaded=$(timeout 5s ovs-vsctl show >/dev/null 2>&1 && echo 1 || echo 0); is really only checking that the ovsdb is working | 16:45 |
sdake | kfox1111 or inc0 can you ack https://review.openstack.org/#/c/464273/ | 16:46 |
kfox1111 | sean-k-mooney: I didn't write that check. I wrote the one below that does an ovs-appctl version check. | 16:46 |
sdake | any mariadb nerds in the channel? :) | 16:47 |
kfox1111 | before the ovs-appctl there was a startup race. after I added it, it seemd to be fixed. has been stable ever since. | 16:47 |
kfox1111 | sdake: what do you want to know? | 16:48 |
*** chas has quit IRC | 16:48 | |
sean-k-mooney | kfox1111: hum ok appctl shoudl conneect to the vswitchd so ok then ya that patch should be fine. | 16:48 |
sdake | kfox1111 https://review.openstack.org/#/c/468632/ | 16:49 |
sdake | kfox1111 comments on the comments would help :) | 16:50 |
*** manheim has joined #openstack-kolla | 16:50 | |
openstackgerrit | Steven Dake proposed openstack/kolla master: GATECHECK of kolla repo https://review.openstack.org/469183 | 16:51 |
sdake | kfox1111 basically that bash as written doesn't work properly | 16:51 |
sdake | grep returns 1 - which causes the extend-start to exit immediately | 16:51 |
sdake | kfox1111 that patch works in my environment (with kolla-ansible and kolla-kubernetes) however it is not passing the gate | 16:51 |
sdake | not sure if the gate is cratered, as i saw a similar failure in an unrelated patch | 16:52 |
*** jemcevoy has joined #openstack-kolla | 16:54 | |
*** Mannheim has joined #openstack-kolla | 16:54 | |
*** manheim has quit IRC | 16:55 | |
*** jtriley has quit IRC | 16:57 | |
*** itlinux has quit IRC | 16:58 | |
openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes master: Remove pathfinder from call path of secret-generator https://review.openstack.org/464273 | 17:02 |
openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes master: Remove pathfinder from call path of secret-generator https://review.openstack.org/464273 | 17:03 |
sdake | kfox1111 that review https://review.openstack.org/#/c/468632/ is partly a bash question and partly a mysql question - right up your alley :) | 17:04 |
*** reidrac has quit IRC | 17:05 | |
*** mgoddard has quit IRC | 17:07 | |
*** jtriley has joined #openstack-kolla | 17:12 | |
*** harlowja has joined #openstack-kolla | 17:14 | |
sean-k-mooney | sdake: left my two cents on that review hopefully it helps but usally the exit code of grep is used to tell if the predicate matched. | 17:14 |
sdake | sean-k-mooney that works great unless set errexit is set :) | 17:15 |
sean-k-mooney | hehe ya we save the vaule of that in our devstack plugin and restore it after we finish preforming actions | 17:15 |
sean-k-mooney | there are only some part of our plugin wehre we want that to be set so we enable it just for those | 17:16 |
bmaltais | Does kolla-ansible have a dependancy on NetworkManager on centos7? | 17:16 |
sean-k-mooney | bmaltais: not specifically | 17:17 |
sean-k-mooney | bmaltais: it does require that somthing configures the ips and interfaces | 17:17 |
*** lpetrut has joined #openstack-kolla | 17:17 | |
bmaltais | OK. Was having issues with bond interface going up and down. I think it was caused by network manager | 17:18 |
sean-k-mooney | ah well kolla wont touch the host level networking in that respect excepth for adding the neutron external interface to ovs/lb | 17:18 |
*** eaguilar has joined #openstack-kolla | 17:19 | |
*** rwallner has quit IRC | 17:22 | |
Reepicheep | yeah bmaltais what ever you set as neutron_external_interface is the only one it kolla really configures | 17:23 |
Reepicheep | other then the vip when you have haproxy.. | 17:23 |
sdake | bmaltais i turn network manager off | 17:24 |
Reepicheep | if the interface is going up and down it should show up in your system logs / journeld | 17:24 |
sdake | bmaltais or use the NM_* whatever it is in the sysconfig files to disable network manager managmeent of the interfaces | 17:25 |
*** rwallner has joined #openstack-kolla | 17:25 | |
sbezverk | sdake: which bash not working you are referring to? | 17:26 |
*** rmart04 has quit IRC | 17:26 | |
openstackgerrit | Steven Dake proposed openstack/kolla-kubernetes master: Remove pathfinder from call path of secret-generator https://review.openstack.org/464273 | 17:27 |
sdake | inc0 sbezverk can you merge https://review.openstack.org/#/c/464031/ | 17:28 |
*** emccormick has quit IRC | 17:28 | |
sdake | sbezverk this patch: https://review.openstack.org/#/c/468632/ | 17:28 |
bmaltais | Tank's all | 17:28 |
bmaltais | Thank's | 17:29 |
*** Serlex has quit IRC | 17:29 | |
*** lpetrut has quit IRC | 17:29 | |
*** Pavo has joined #openstack-kolla | 17:30 | |
inc0 | sdake: https://review.openstack.org/#/c/469167/ <- neutron has a bug | 17:33 |
inc0 | thats why we can't spawn vm | 17:33 |
inc0 | can you merge it regardless? it breaks multinode gates totally | 17:33 |
*** Pavo has quit IRC | 17:33 | |
inc0 | as in, it breaks Kolla on multinode | 17:34 |
inc0 | not only in gates | 17:34 |
*** jrich523 has quit IRC | 17:34 | |
*** sambetts is now known as sambetts|afk | 17:35 | |
sdake | inc0 that may explain why my mariadb fix is blowing up in the gates | 17:36 |
sdake | is there an eta on a revert from the neutron team? | 17:36 |
kbaegis1 | inc0: Kolla-ansible doesn't like running setup in a venv. Any idea why? | 17:36 |
*** unicell has joined #openstack-kolla | 17:36 | |
openstackgerrit | Merged openstack/kolla-kubernetes master: Add a destroy workflow for kolla-kubernetes https://review.openstack.org/464031 | 17:37 |
kbaegis1 | inc0: The setup playbook fails to run and just hangs indefinitely | 17:37 |
inc0 | kbaegis1: hold on, I'll get to you in a second;) | 17:38 |
inc0 | sdake: https://bugs.launchpad.net/neutron/+bug/1694420 | 17:39 |
openstack | Launchpad bug 1694420 in neutron "AttributeError: 'NoneType' object has no attribute 'port_security_enabled" [Critical,Confirmed] | 17:39 |
inc0 | kbaegis1: logs plz | 17:39 |
openstackgerrit | Merged openstack/kolla-ansible master: Revert "Add jinja2-py version check in pre_checks" https://review.openstack.org/469167 | 17:39 |
kbaegis1 | inc0: https://hastebin.com/cuzuwocewo.md | 17:40 |
inc0 | kbaegis1: ok that's ansible issue, nothing kolla specific, but let's try to debug | 17:41 |
kbaegis1 | inc0: Might be my inventory? https://hastebin.com/uzufutoyix.bash | 17:42 |
inc0 | kbaegis1: try to run ansible -m setup -i /path/to/inv all | 17:42 |
kbaegis1 | inc0: Also hangs, returns nothing | 17:43 |
*** emccormick has joined #openstack-kolla | 17:44 | |
inc0 | ps aux on the node to see what's happening? | 17:44 |
inc0 | as I said, this goes beyond kolla, you might want to ask in #ansible | 17:44 |
*** mgoddard has joined #openstack-kolla | 17:44 | |
*** eaguilar has quit IRC | 17:46 | |
*** eaguilar has joined #openstack-kolla | 17:47 | |
*** Pavo has joined #openstack-kolla | 17:47 | |
*** korzen has quit IRC | 17:49 | |
*** egonzalez has joined #openstack-kolla | 17:50 | |
sean-k-mooney | kbaegis1: check the mtu between you and the remote node | 17:51 |
kbaegis1 | its localhost | 17:52 |
kbaegis1 | (AIO) | 17:52 |
inc0 | sean-k-mooney: so, I'm looking at neutron server for odl | 17:52 |
sean-k-mooney | oh i had the same behvaior last week when my mtu was wrong and it could not copy the output | 17:52 |
kbaegis1 | Here's where it's hanging: https://hastebin.com/akidaqoxel.vbs (lvs) | 17:52 |
inc0 | I'm on the fence, I'm not sure if delta is big enough to varrant whole new image | 17:52 |
sean-k-mooney | inc0: everything they are doing can be done with template origde and most of what its doing is not related to odl | 17:53 |
inc0 | yeah | 17:54 |
inc0 | and source build is almost identical | 17:55 |
sean-k-mooney | so what markus has already merged should support this usecase form an image point of view | 17:55 |
sean-k-mooney | the main usecase discribed in the commit refers to "extensions that may cause DB upgrade failures" | 17:56 |
inc0 | trozet sean-k-mooney mgkwill how about we discuss this on ML? | 17:56 |
inc0 | having a lot of non-DRY images will cause headache with support | 17:57 |
sean-k-mooney | inc0: yep if we merge this image we should also have another for ovn and every other contoler | 17:58 |
inc0 | yeah, I don't want that | 17:58 |
inc0 | we need to figure this problem in more..systematic model | 17:58 |
sean-k-mooney | inc0: https://github.com/openstack/kolla/blob/master/doc/image-building.rst#plugin-functionality should in theroy allready adress it | 17:59 |
inc0 | well overrides were created to solve this precise issue | 17:59 |
inc0 | I think they are enough to not create new image | 17:59 |
sean-k-mooney | inc0: my approch would be to build the same image multiple times with different tags | 18:00 |
inc0 | but we might need better infrastructure for shipping overrides | 18:00 |
*** Pavo has quit IRC | 18:00 | |
kbaegis1 | inc0: lvm crashed/was non-responsive. Gah. There went my morning ;) | 18:00 |
inc0 | or different names based of override | 18:00 |
sean-k-mooney | then you can change the tag in the global.yml to select if you want the backend you wnat | 18:00 |
sean-k-mooney | ya names works too | 18:00 |
inc0 | names will be concise with rest of kolla | 18:00 |
sean-k-mooney | i dont think you can change the name on the commandline though where as tags can be | 18:01 |
inc0 | no, we need to chagne build script | 18:01 |
*** myoung is now known as myoung|biab | 18:01 | |
sean-k-mooney | if we were to do that it would be usefully to also be able to build all versions at the same time | 18:02 |
sean-k-mooney | with overrides as they are today that will not work but it should be possible to add | 18:02 |
inc0 | yeah, I'm wondering if we could override container name in override file itself | 18:03 |
sean-k-mooney | ya thats basically what i was thinking with one override file per backend | 18:03 |
sean-k-mooney | inc0: that still leaves a problem though if you want multiple backends | 18:04 |
sean-k-mooney | or more likely multiple advanced services | 18:04 |
sean-k-mooney | there really isnt a difference between enableing the odl driver vs sfc/lbaas/*aas | 18:05 |
inc0 | sean-k-mooney: do you feel like protoryping something? | 18:11 |
inc0 | or I can do it later too | 18:11 |
inc0 | just throw few ideas how to solve it in general usecase | 18:11 |
*** david-lyle has quit IRC | 18:13 | |
*** jtriley has quit IRC | 18:13 | |
sean-k-mooney | inc0: am im not sure how much time ill be able to spend on it but ill take a look when if i get a chance. if you have something ill be happy to revew also. kolla build system is resonably flexable so i hope it would not be that much effort to extend to cover this | 18:15 |
inc0 | sean-k-mooney: kk I'll work on it then | 18:15 |
inc0 | trozet sorry for -2:( I'll let you know when I get prototype up | 18:16 |
trozet | inc0: sorry I missed all the discussion | 18:17 |
sean-k-mooney | inc0: trozet from a vendor point of view it makes senese to have a way to make contianer with know validated sets of plugins so if we can adress this generically im sure it will be useful for anyone downstream of kolla | 18:17 |
inc0 | trozet: in short we were discussing neutron-server-odl change | 18:17 |
trozet | sean-k-mooney, inc0: right so we needa nother one for ovn, etc | 18:18 |
inc0 | I think we need better way to ship[ all these variations without adding container | 18:18 |
inc0 | thats exactly my point | 18:18 |
trozet | sean-k-mooney, inc0: what is your solution? | 18:18 |
inc0 | trozet: my personal favorite right now is to modify override mechanism in a way that we'll be able to create new container from override | 18:18 |
inc0 | then you for example create override with just changes to neutron-server and it will land as a separate container | 18:19 |
sean-k-mooney | trozet: effectly use template overrides for each variation and extend the buildsystem to build multiple copyies of the container for each varation and allow the override to spcify the containers name | 18:19 |
trozet | inc0: is the advantage that this is more simple than duplicating the template for each new container? | 18:19 |
inc0 | trozet: advantage is that if for example we'd need to fix a bug in neutron container we won't forget about 4 different containers | 18:20 |
inc0 | that share same code | 18:20 |
trozet | inc0: but i already use the neutron server as my base container, so wouldnt i pick up any change you make to the that? | 18:21 |
inc0 | well there is a lot of overlap | 18:21 |
inc0 | like start.sh | 18:21 |
trozet | inc0: so the advantage is we have a smaller template file and get rid of the start | 18:22 |
inc0 | maybe it's matter of doing sub-container differently, thing is, we need a solution soon | 18:22 |
openstackgerrit | Merged openstack/kolla-kubernetes master: permit rabbitmq-mgmt to select rabbitmq as endpoint https://review.openstack.org/467771 | 18:22 |
inc0 | well, not as much "smaller" as DRY | 18:22 |
*** iniazi_ has joined #openstack-kolla | 18:22 | |
inc0 | issue I'm afrtaid of is that we'll have ton of small variations of a container | 18:23 |
inc0 | with neutron, cinder | 18:23 |
inc0 | all the heavy plugin based services | 18:23 |
inc0 | that can easily get out of hand to handle | 18:24 |
trozet | inc0: but wont you have that even with an override? | 18:24 |
trozet | inc0: still producing many containers right | 18:24 |
inc0 | we can either use FROM heavier or ship templates | 18:24 |
inc0 | trozet: many containers is fine, lots of overlapping code is not | 18:24 |
sean-k-mooney | inc0: well shiping template in contrib would be ease to manage | 18:24 |
*** jamesPR has quit IRC | 18:25 | |
trozet | inc0: ok so you will work on a better mechanism, will you let me know once that is done please? | 18:25 |
inc0 | we also need to handle different "quality" of containers - as in, if it breaks because odl plugin gets broken upstream, we can't block building containers in kolla | 18:25 |
*** iniazi has quit IRC | 18:25 | |
inc0 | trozet: here's the deal, I'll remove -2 and let's make a race;) I'll explore avenue of overrides, could you try to remove code duplication? | 18:26 |
trozet | inc0: i thought it already builds the ones it can build now, at least thats what I saw in my tests | 18:26 |
*** david-lyle has joined #openstack-kolla | 18:26 | |
trozet | inc0: btw neutron server wouldnt build for ubuntu | 18:26 |
inc0 | yeah that's what I'm saying, we already have issues | 18:26 |
trozet | inc0: LOL i'll lose the race | 18:26 |
trozet | inc0: im too slow :) | 18:27 |
sean-k-mooney | trozet: ya the binary build on ubunutu was broken a few days ago not sure if its fix | 18:27 |
sean-k-mooney | trozet: ubuntu source should work though | 18:27 |
inc0 | this is not about speed;) what I want is few ideas on table to solve this issue | 18:27 |
trozet | inc0: is it not possible for me to point to the neutron-server start.sh in its directory? | 18:27 |
sean-k-mooney | trozet: can i ask why are you adding sfc/bgpvpn/l2 gw to the odl container specifcally | 18:28 |
trozet | sean-k-mooney: those are features we are focusing on in ODL right now | 18:28 |
sean-k-mooney | trozet: sure but if they made sense to include for odl why would you not include them in the base neutron server contianer | 18:29 |
*** tonanhngo has joined #openstack-kolla | 18:29 | |
inc0 | trozet: docker doesn't work like that, but since start.sh is already copied in container, you don't need to re-copy it | 18:29 |
sean-k-mooney | trozet: effectly it feels like you are trying to do too things at once, enable odl and enable other service you would like to use with odl | 18:29 |
trozet | sean-k-mooney: the consensus was in OOO that users dont want to use those with ovs agent | 18:29 |
inc0 | just remove whole thing from your container | 18:29 |
trozet | inc0: oh yeah good point | 18:30 |
kbaegis1 | https://hastebin.com/ejuziqejos.vbs <— Weird failures from inside the venv. "can't parse" | 18:30 |
sean-k-mooney | well networking sfc makes sense without odl but this remindes me the networking-sfc intergration without odl si currently broken in kolla | 18:30 |
inc0 | trozet: I switched -2 to -1 to not block your work, since I voiced my concerns;) | 18:31 |
inc0 | this is hard problem to solve and might require refactor of build script | 18:31 |
inc0 | which I'd be ok with as long as it makes sense | 18:31 |
inc0 | funny enough, if we would allow adding containers from overrides themselves | 18:32 |
inc0 | we could theoretically make most of containers just overrides:)_ | 18:32 |
jemcevoy | inc0: I am having an issue with the nova.conf file... I have two generations of nVidia Grid Cards the K2 and the M60 which requires two pci_alias entries in the nova.conf but only the second entry make it to the controller nodes. | 18:33 |
jemcevoy | http://paste.openstack.org/show/611006/ has the entries I have on the ansible node and the entry that gets to the runtime on the controller | 18:34 |
inc0 | hmm...ini supports multiple keys like that? | 18:34 |
trozet | inc0: right I was going to say isnt neutron server an override of base neutron | 18:34 |
jemcevoy | Yes I have done it with a K1 and a K2 GPUs | 18:35 |
trozet | inc0: so it's really however you want to do it... I personally think the current structure is fine | 18:36 |
inc0 | jemcevoy: it might be bug in either our merge_config or oslo.config | 18:36 |
kbaegis1 | jemcevoy: Well it's certainly a violation of .ini | 18:36 |
inc0 | trozet: let's try by removing overlapping code from regular neutron-server ok? | 18:36 |
trozet | inc0: sure I will do that | 18:37 |
sean-k-mooney | inc0: novas conf are not ini files | 18:38 |
sean-k-mooney | ini files dont support repeated fields but .conf files do | 18:38 |
inc0 | yeah I guess that's us merging conf as ini | 18:40 |
inc0 | that will make this discussion about confs in central database much more interesting too | 18:40 |
sean-k-mooney | inc0: well we use config parser but yes i think i know where the bug will be | 18:40 |
sean-k-mooney | we convert the config into a dictionary and thend merge the dictionaries | 18:41 |
sean-k-mooney | inc0: https://github.com/openstack/kolla-ansible/blob/master/ansible/action_plugins/merge_configs.py#L53-L54 | 18:42 |
sean-k-mooney | inc0: we just need to see if the key already exites and if it does we need to convert it to a list of values | 18:43 |
*** mgoddard has quit IRC | 18:43 | |
*** numans has quit IRC | 18:43 | |
inc0 | sean-k-mooney: that won't do, we need to be able to both add keys and override existing | 18:44 |
*** jtriley has joined #openstack-kolla | 18:44 | |
sean-k-mooney | inc0: hum good point ... | 18:44 |
kbaegis1 | So I'm having problems running reconfigure or deploy from inside a py2.7 venv: https://hastebin.com/ejuziqejos.vbs "Can not parse the inner module output". Syslog https://hastebin.com/anonapofel.php | 18:45 |
jemcevoy | inc0: I believe multiples are acceptable when marked (Multi-valued) https://docs.openstack.org/newton/config-reference/compute/config-options.html#id29 | 18:46 |
*** numans has joined #openstack-kolla | 18:46 | |
kbaegis1 | Anyone know the syntax to run a command through the toolbox? | 18:46 |
kbaegis1 | I can access mysql locally from the container | 18:46 |
sean-k-mooney | jemcevoy: correct and the pci whiltleis is allowed to be repeted | 18:46 |
kbaegis1 | login works w/ password | 18:46 |
inc0 | jemcevoy: you might have to edit kolla code unfortunately | 18:48 |
inc0 | this is hard problem and we need to fix it | 18:48 |
inc0 | kbaegis1: docker exec -it kolla_toolbox bash | 18:48 |
inc0 | and you're inside container | 18:48 |
jemcevoy | I got around the whitelist by having seperate configs for the compute hosts with K2 and M60's but the scheduler looka like I will need to hack the configs on the three controller noedes | 18:48 |
inc0 | I wonder how that works with nova --config-file conf-first-pci.conf --config-file conf-second-pci.conf | 18:49 |
inc0 | will it add? or override? | 18:49 |
sean-k-mooney | inc0: i belive override | 18:50 |
sean-k-mooney | inc0: that said i have not tried it | 18:50 |
sean-k-mooney | inc0: i know repeating works in a single file | 18:50 |
kbaegis1 | inc0: I can do that; can't the toolbox container run commands for other containers though? | 18:52 |
sean-k-mooney | inc0 that said becasue the key is declared as a multistingopt i may handel it differenly https://github.com/openstack/nova/blob/master/nova/conf/pci.py#L54-L119 | 18:52 |
*** numans has quit IRC | 18:53 | |
inc0 | kbaegis1: not this way, it uses API as any other good service | 18:53 |
sean-k-mooney | jemcevoy: can you try passthrough_whitelist = [{"product_id":"0001", "vendor_id":"8086"}, | 18:53 |
sean-k-mooney | {"product_id":"0002", "vendor_id":"8086"}] | 18:53 |
kbaegis1 | inc0: Gotcha. Any idea from my logs why the database create fails? (It's already created, but outside the venv) | 18:54 |
sean-k-mooney | jemcevoy: so instead of multiple seperate options declare an array of them | 18:54 |
kbaegis1 | inc0: Do I have to do a destroy;deploy? | 18:54 |
inc0 | kbaegis1: show me exact logs please | 18:54 |
kbaegis1 | inc0: https://hastebin.com/anonapofel.php includes syslog, deploy and reconfigure | 18:55 |
*** matrohon has joined #openstack-kolla | 18:55 | |
jemcevoy | sean-k-mooney: I will try defining a list as you say.... Will let you know after the reconfigure finishes | 18:56 |
inc0 | kbaegis1: hmm its strange | 18:57 |
kbaegis1 | inc0: I'm good at strange | 18:57 |
inc0 | try specyfyuing ansible_python_interpreter | 18:58 |
inc0 | ansible playbook -e ansible_python_interpreter=/path/to/venv/interpreter | 18:59 |
inc0 | you can specify that in inventory too | 18:59 |
*** numans has joined #openstack-kolla | 18:59 | |
kbaegis1 | inc0: Already did: ansible_python_interpreter='/source/kolla/ansible-git/py2-venv/bin/python' | 18:59 |
inc0 | :S | 19:00 |
inc0 | I'm out of ideas... | 19:00 |
openstackgerrit | OpenStack Proposal Bot proposed openstack/kolla-ansible stable/ocata: Updated from global requirements https://review.openstack.org/452870 | 19:00 |
kbaegis1 | inc0: *shrug* Thanks, I'll try some other vodoo out :) | 19:01 |
kbaegis1 | oh | 19:01 |
*** myoung|biab is now known as myoung | 19:05 | |
*** manheim has joined #openstack-kolla | 19:05 | |
*** itlinux has joined #openstack-kolla | 19:17 | |
kfox1111 | inc0: sbezverk: sdake: https://github.com/kubernetes/community/pull/659 | 19:17 |
kfox1111 | we need to see if we can get some prioritized focus on that. | 19:17 |
inc0 | kfox1111: this is about network namespaces sharing? | 19:17 |
kfox1111 | yeah. | 19:18 |
inc0 | we need to add sig-openstack to this | 19:18 |
kfox1111 | how does that get done? | 19:19 |
inc0 | dunno, I just commented it | 19:19 |
kfox1111 | k | 19:19 |
inc0 | either they'll add it or they'll say I'm noob at opensource and rtfm and point me to docs which will say how to do it | 19:20 |
kfox1111 | hehe. | 19:21 |
*** Pavo has joined #openstack-kolla | 19:21 | |
*** sayantan_ has joined #openstack-kolla | 19:24 | |
sean-k-mooney | inc0: normally you just click a little gear icon beside lable and add it but i think they have resticted it so you need to have the member or contributor tag | 19:24 |
*** salv-orl_ has joined #openstack-kolla | 19:27 | |
*** salv-orlando has quit IRC | 19:30 | |
*** sayantan_ has quit IRC | 19:30 | |
*** chas has joined #openstack-kolla | 19:34 | |
*** itlinux has quit IRC | 19:38 | |
sbezverk | kfox1111: in that pull you mentioned I do not see any code changes, am I missing something? | 19:52 |
kbaegis1 | inc0: So here's an actual incompatability. Not sure how I fixed this last time: https://hastebin.com/asihayaxoy.rb | 19:52 |
kbaegis1 | sysctl package will not install for several linux systems | 19:55 |
*** shardy has quit IRC | 19:59 | |
*** egonzalez has quit IRC | 20:00 | |
*** dave-mcc_ has quit IRC | 20:04 | |
*** manheim has quit IRC | 20:04 | |
*** klindgren has joined #openstack-kolla | 20:06 | |
*** unicell has quit IRC | 20:08 | |
*** unicell has joined #openstack-kolla | 20:08 | |
inc0 | kbaegis1: if you'd be able to make a change that would make it more flexible | 20:08 |
inc0 | I'd gladly approve it;)_ | 20:08 |
*** fguillot has quit IRC | 20:11 | |
*** matrohon has quit IRC | 20:13 | |
*** sgrasley1 has left #openstack-kolla | 20:19 | |
jemcevoy | inc0 sean-k-mooney : Defining as an array ie [{dict} | 20:22 |
jemcevoy | inc0 sean-k-mooney : Defining as an array ie [{dict},{dict}] did not work.... Infact the dates on the config files did not change when I ran reconfigure | 20:24 |
inc0 | that's...strange | 20:24 |
jemcevoy | I'll try again changing the order of the array to see if that has an effect | 20:25 |
openstackgerrit | Andy Smith proposed openstack/kolla-ansible master: Add support for hybrid messaging backends https://review.openstack.org/468966 | 20:29 |
*** Guest63352 has joined #openstack-kolla | 20:30 | |
*** eaguilar has quit IRC | 20:32 | |
*** Guest63352 is now known as ryanender | 20:33 | |
*** lucasxu has quit IRC | 20:36 | |
*** lucasxu has joined #openstack-kolla | 20:36 | |
bjolo | a little outside kolla... is it possible to explicitly deny a project access to a network? neutron rbac docs only talks about giving acces for specific projects, and use * for all projects. | 20:39 |
bjolo | i have hundreds of projects and all have access to the default external network. now i have one project that should only have access to another external network but no access to the default external | 20:40 |
sbezverk | bjolo: depending on your network gear and ip allocation it could be done with access list.. but again I can only confirm for cisco ;) | 20:42 |
bjolo | i dont want to have to give every project explicit rights to the default external network. (today it has wildcard i.e. all projects have access) | 20:42 |
bjolo | sbezverk, im talking neutron rbac :) | 20:42 |
bjolo | https://docs.openstack.org/ocata/networking-guide/config-rbac.html | 20:42 |
sbezverk | bjolo: ok then ignore my comment ;) | 20:43 |
bjolo | tnx anyway :) | 20:43 |
*** fooliouno has joined #openstack-kolla | 20:43 | |
bjolo | sean-k-mooney your a neutron dev right? ^^ | 20:43 |
Reepicheep | hmm speaking of neutron.. I have a bad habbit of typing it "nuetron" | 20:44 |
bjolo | aka new tron :) | 20:44 |
Reepicheep | usually not a big deal unless it's when I'm trying to create /etc/kolla/config/neutron | 20:44 |
Reepicheep | then I spend way to much time troubleshooting why my ml2_conf.ini file is not updating :| | 20:45 |
bjolo | haha typos are a bitch man | 20:46 |
*** jtriley has quit IRC | 20:46 | |
*** rwallner has quit IRC | 20:48 | |
*** sgrasley has joined #openstack-kolla | 20:48 | |
*** rwallner has joined #openstack-kolla | 20:48 | |
*** Mannheim has quit IRC | 20:49 | |
*** powerd has quit IRC | 20:49 | |
*** powerd has joined #openstack-kolla | 20:50 | |
*** manheim has joined #openstack-kolla | 20:51 | |
*** rwallner has quit IRC | 20:52 | |
*** david-lyle has quit IRC | 20:53 | |
*** david-lyle has joined #openstack-kolla | 20:54 | |
*** emccormick has quit IRC | 20:55 | |
*** ryanender has quit IRC | 20:59 | |
*** matrohon has joined #openstack-kolla | 21:00 | |
*** lucasxu has quit IRC | 21:03 | |
inc0 | kfox1111 sdake https://review.openstack.org/#/c/469248/ | 21:04 |
openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-ansible master: Enable multi-scenerio gates https://review.openstack.org/466815 | 21:07 |
openstackgerrit | Michal Jastrzebski (inc0) proposed openstack/kolla-ansible master: Enable multinode gate https://review.openstack.org/466007 | 21:07 |
inc0 | trozet: so I'm thinking if you could remove all the non-neutron-odl specific stuff from Dockerfile that should just work | 21:09 |
inc0 | and be failry straightforward and elegant | 21:09 |
kbaegis1 | Still the same thing :/ https://hastebin.com/izakeqacaw.md I can't seem to get it deployed from a venv | 21:10 |
inc0 | kbaegis1: python2.7 everywhere? | 21:11 |
kbaegis1 | inc0: I've tried every conceivable combination: 3.5->2.7 target, 3.5->3.5 target, 3.4->3.4 target, etc | 21:11 |
kbaegis1 | Nothing friggin works | 21:11 |
*** david-lyle has quit IRC | 21:12 | |
inc0 | I'm deeply saddened to read that | 21:12 |
kbaegis1 | And all toolbox returns is the braindead error: Can not parse the inner module output: localhost | SUCCESS => {\n \"changed\": false, \n \"user\": \"haproxy\"\n | 21:12 |
kbaegis1 | I had it working running pip as root, but that breaks gentoo system utitlies | 21:12 |
inc0 | so let's try that: | 21:12 |
kbaegis1 | So if I'm documenting it, I have to do it in a venv | 21:12 |
kbaegis1 | well, with 3.4 I keep getting 'NameError: name 'basestring' is not defined' | 21:16 |
*** goldyfruit has quit IRC | 21:17 | |
kbaegis1 | on all 3 versions of python I have | 21:17 |
inc0 | it seems that you have some strange regex module in python oO | 21:18 |
kbaegis1 | Ahh, right | 21:19 |
inc0 | https://github.com/openstack/kolla-ansible/blob/afdd11b9a22ecca70962a4637d89ad50b7ded2e5/ansible/library/kolla_toolbox.py#L86-L90 | 21:19 |
inc0 | https://hastebin.com/izakeqacaw.md line 29 | 21:20 |
kbaegis1 | Yeah, that was what I had to find a workaround for | 21:21 |
kbaegis1 | Which got overwritten :/ | 21:21 |
*** lucasxu has joined #openstack-kolla | 21:26 | |
*** lucasxu has quit IRC | 21:31 | |
kbaegis1 | Well, wayyyyy too many version dependencies | 21:31 |
kbaegis1 | ansible, pip, docker, kolla- not to mention python | 21:32 |
kbaegis1 | Should've just gone with an iterator to try all available options ;) | 21:32 |
kbaegis1 | Unfortunately I know one actually works, as I've proven it and it's driving me mad | 21:32 |
*** pradk has joined #openstack-kolla | 21:42 | |
*** salv-orl_ has quit IRC | 21:50 | |
*** salv-orlando has joined #openstack-kolla | 21:50 | |
*** salv-orlando has quit IRC | 21:54 | |
*** jamesben_ has quit IRC | 22:04 | |
*** jamesben_ has joined #openstack-kolla | 22:07 | |
*** chas has quit IRC | 22:08 | |
*** jamesben_ has quit IRC | 22:12 | |
*** pradk has quit IRC | 22:17 | |
*** powerd has quit IRC | 22:20 | |
*** emccormick has joined #openstack-kolla | 22:21 | |
inc0 | kbaegis1: or you know...use package manager;) | 22:25 |
inc0 | ahh I forgot, gentoo | 22:25 |
*** fooliouno has quit IRC | 22:29 | |
*** powerd has joined #openstack-kolla | 22:29 | |
*** inc0 has quit IRC | 22:32 | |
*** inc0 has joined #openstack-kolla | 22:33 | |
*** ducttape_ has joined #openstack-kolla | 22:47 | |
*** ducttap__ has quit IRC | 22:48 | |
*** jamesbenson has joined #openstack-kolla | 22:51 | |
*** tonanhngo has quit IRC | 22:55 | |
*** jamesbenson has quit IRC | 22:55 | |
*** GonZo2000 has joined #openstack-kolla | 22:57 | |
openstackgerrit | Merged openstack/kolla-ansible stable/ocata: Updated from global requirements https://review.openstack.org/452870 | 22:58 |
*** iniazi has joined #openstack-kolla | 23:01 | |
*** iniazi_ has quit IRC | 23:02 | |
*** tonanhngo has joined #openstack-kolla | 23:04 | |
*** tonanhngo has quit IRC | 23:08 | |
*** kbaegis1 has quit IRC | 23:10 | |
*** itlinux has joined #openstack-kolla | 23:11 | |
*** tonanhngo has joined #openstack-kolla | 23:16 | |
*** kbaegis has joined #openstack-kolla | 23:19 | |
*** kbaegis has quit IRC | 23:19 | |
*** kbaegis has joined #openstack-kolla | 23:19 | |
*** kbaegis has quit IRC | 23:19 | |
*** kbaegis has joined #openstack-kolla | 23:19 | |
*** kbaegis has quit IRC | 23:19 | |
*** kbaegis has joined #openstack-kolla | 23:20 | |
*** kbaegis has quit IRC | 23:20 | |
*** kbaegis has joined #openstack-kolla | 23:20 | |
*** kbaegis has quit IRC | 23:20 | |
kfox1111 | review just landed. https://review.openstack.org/#/c/469265/ | 23:20 |
kfox1111 | interested parties should weigh in. | 23:21 |
*** tonanhngo has quit IRC | 23:21 | |
*** Pavo has quit IRC | 23:26 | |
sdake | inc0 popcorn! | 23:34 |
jemcevoy | inc0 sean-k-mooney: I did mange to make kolla-ansible change the nova.conf files with the pci_alias as an array or list and that causes nova to throw an exception.PciInvalidAlias http://paste.openstack.org/show/611033/ | 23:35 |
*** GonZo2000 has quit IRC | 23:37 | |
*** manheim has quit IRC | 23:40 | |
*** fguillot has joined #openstack-kolla | 23:56 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!