*** cshen has joined #openstack-ansible | 01:07 | |
*** cshen has quit IRC | 01:12 | |
*** prometheanfire has joined #openstack-ansible | 01:45 | |
*** macz_ has joined #openstack-ansible | 02:42 | |
*** macz_ has quit IRC | 02:46 | |
*** dmsimard has quit IRC | 02:49 | |
*** mmethot_ has joined #openstack-ansible | 02:50 | |
*** dmsimard has joined #openstack-ansible | 02:50 | |
*** mmethot has quit IRC | 02:53 | |
*** cshen has joined #openstack-ansible | 03:07 | |
*** cshen has quit IRC | 03:12 | |
*** cshen has joined #openstack-ansible | 05:08 | |
*** cshen has quit IRC | 05:12 | |
*** evrardjp has quit IRC | 05:33 | |
*** evrardjp has joined #openstack-ansible | 05:33 | |
*** cp- has quit IRC | 05:34 | |
*** yasemind has joined #openstack-ansible | 05:34 | |
*** cp- has joined #openstack-ansible | 05:40 | |
*** macz_ has joined #openstack-ansible | 05:46 | |
*** macz_ has quit IRC | 05:50 | |
*** cp- has quit IRC | 05:51 | |
*** cp- has joined #openstack-ansible | 05:56 | |
*** ianychoi_ has joined #openstack-ansible | 05:57 | |
*** ianychoi has quit IRC | 06:00 | |
*** pto_ has joined #openstack-ansible | 06:20 | |
*** cshen has joined #openstack-ansible | 06:22 | |
*** pto has quit IRC | 06:22 | |
*** cshen has quit IRC | 06:26 | |
*** lemko has joined #openstack-ansible | 06:31 | |
*** lemko2 has quit IRC | 06:32 | |
*** miloa has joined #openstack-ansible | 06:33 | |
*** SiavashSardari has joined #openstack-ansible | 07:06 | |
*** jbadiapa has joined #openstack-ansible | 07:25 | |
*** rpittau|afk is now known as rpittau | 07:26 | |
*** lemko0 has joined #openstack-ansible | 07:29 | |
*** lemko has quit IRC | 07:30 | |
*** lemko0 is now known as lemko | 07:30 | |
*** macz_ has joined #openstack-ansible | 07:34 | |
*** macz_ has quit IRC | 07:39 | |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_cinder stable/ussuri: Set correct permissions for rootwrap.d https://review.opendev.org/762789 | 07:44 |
---|---|---|
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_cinder stable/train: Set correct permissions for rootwrap.d https://review.opendev.org/762790 | 07:44 |
*** sshnaidm|off is now known as sshnaidm | 07:46 | |
*** pcaruana has joined #openstack-ansible | 07:49 | |
*** rpittau is now known as rpittau|bbl | 07:52 | |
SiavashSardari | noonedeadpunk could you please help me with the openrc role issue? | 07:53 |
noonedeadpunk | SiavashSardari: do you mean that when you remove service, it will drop endpoint as well? | 07:53 |
SiavashSardari | yes exactly. I already tested that in my setup | 07:54 |
noonedeadpunk | hm.... | 07:54 |
*** cshen has joined #openstack-ansible | 07:56 | |
noonedeadpunk | yeah, you're absolutely right | 07:57 |
noonedeadpunk | however, _service_catalog is independent from _service_endpoints? | 07:59 |
noonedeadpunk | I mean in general, it doesn't mean it makes sense in specificly cinder role | 07:59 |
noonedeadpunk | I will try to make proper fix to the collection and then we can use new version I guess | 08:00 |
noonedeadpunk | but yeah, agree that your solution makes sense as well, but I'd rather do it on side of the collection | 08:02 |
SiavashSardari | yeah you're right I didn't know about openstack-ansible-test we should take a look at that and find a solution that works for all projects | 08:03 |
noonedeadpunk | I mean I guess module makes wrong assuption about what do they recieve here as endpoint. So they should just return changed: false and that;s it | 08:04 |
noonedeadpunk | https://opendev.org/openstack/ansible-collections-openstack/src/tag/1.0.0/plugins/modules/endpoint.py#L198-L199 | 08:04 |
SiavashSardari | yeah that's a better approach, and I think there is no validity check in the module as well, I mean what if the endpoint provided here is not a valid endpoint at all? | 08:09 |
*** macz_ has joined #openstack-ansible | 08:12 | |
*** macz_ has quit IRC | 08:17 | |
SiavashSardari | noonedeadpunk I think the root cause is here https://opendev.org/openstack/ansible-collections-openstack/src/tag/1.0.0/plugins/modules/endpoint.py#L161-L162 | 08:17 |
noonedeadpunk | yeah, you;re right..... | 08:18 |
noonedeadpunk | already come up to this as well.... | 08:18 |
noonedeadpunk | let me check with folks how do they see it | 08:19 |
SiavashSardari | ok then. if you have the time I would appreciate your input on the openrc role as well. | 08:20 |
noonedeadpunk | yeah, will try to take a look now | 08:21 |
openstackgerrit | PerToft proposed openstack/openstack-ansible master: Added Netplan examples for a Openstack Infrastructure node. Updated with reviews. https://review.opendev.org/760090 | 08:40 |
noonedeadpunk | SiavashSardari: I think issue here is in double not https://review.opendev.org/#/c/762090/2/templates/openrc.j2 | 08:47 |
noonedeadpunk | So by default in openrc you will have `export OS_SYSTEM_SCOPE=all` | 08:47 |
noonedeadpunk | oh wait | 08:49 |
*** andrewbonney has joined #openstack-ansible | 08:49 | |
*** tosky has joined #openstack-ansible | 08:49 | |
noonedeadpunk | in default value actuall, as you will have `not true` | 08:49 |
noonedeadpunk | sodefault should be false here | 08:49 |
noonedeadpunk | but that is barely readable tbh.... | 08:49 |
noonedeadpunk | maybe it's better just to make 2 templates? | 08:51 |
SiavashSardari | I know I didn't like that there is too variables, that is not very readable but I couldn't think of anything else | 08:51 |
SiavashSardari | I wanted to create another template but they will be actually the same | 08:52 |
noonedeadpunk | let's probably create another tempalte and do iterate on `Create openrc file` | 08:52 |
SiavashSardari | i mean kinda the same | 08:52 |
noonedeadpunk | yeah, they will... | 08:52 |
noonedeadpunk | I'd say we can try using overrides here, but we can't drop keys from config with it | 08:53 |
noonedeadpunk | so having this complexity vs having 2 simmilar templates - I'd vote for templates | 08:54 |
SiavashSardari | I used vars but it will not get passed to the actual template. maybe I did it wrong | 08:54 |
noonedeadpunk | well it does at the moment and that's why tests are failing | 08:54 |
noonedeadpunk | that's what I saw http://paste.openstack.org/show/800041/ | 08:55 |
noonedeadpunk | so solution here would be `{% if not _not_openrc_system_scope | default(false) | bool %}` | 08:55 |
SiavashSardari | interesting... | 08:55 |
SiavashSardari | I tested that in my setup and it worked, why it is not working here?? | 08:56 |
noonedeadpunk | but imo, we both spent some time to understand it while knowing why we do this, but with some time this might raise great confusion while debugging smth | 08:56 |
noonedeadpunk | um, I just applied your patch in my sandbox | 08:57 |
SiavashSardari | anyway if you think adding a new template is a better approach I will post a patch with new teplate | 08:57 |
noonedeadpunk | well, we can ask jrosser as well :) | 08:57 |
noonedeadpunk | I think it's just more simple and more readable approach | 08:58 |
SiavashSardari | yep I agree, let's wait for his comment on this issue | 08:59 |
SiavashSardari | I'm ok with a new template as well, it was the first solution that came to my mind too. I just didn't wand to add duplicate (ish) templates | 09:00 |
noonedeadpunk | Yeah, don't like this as well, but have no other ideas tbh as well | 09:01 |
SiavashSardari | on another matter, we don't have openstack-ansible-ops in opendev? just github? | 09:03 |
noonedeadpunk | we have | 09:04 |
noonedeadpunk | github is a mirror of the opendev | 09:04 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-openstack_openrc master: Adding support of system scoped openrc and clouds.yaml https://review.opendev.org/762090 | 09:04 |
SiavashSardari | i'm getting 404 | 09:04 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-openstack_openrc master: Adding support of system scoped openrc and clouds.yaml https://review.opendev.org/762090 | 09:05 |
SiavashSardari | this is the link in docs https://opendev.org/openstack/openstack-ansible-os_ops | 09:05 |
noonedeadpunk | https://opendev.org/openstack/openstack-ansible-ops ? | 09:05 |
noonedeadpunk | where this link is? | 09:05 |
SiavashSardari | https://docs.openstack.org/openstack-ansible-ops/latest/ | 09:06 |
SiavashSardari | well thanks for the link | 09:06 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-ops master: Fix repo link in docs https://review.opendev.org/762801 | 09:08 |
*** fanfi has joined #openstack-ansible | 09:10 | |
SiavashSardari | there are usefull playbooks in https://github.com/openstack/openstack-ansible-ops/tree/master/multi-node-aio/playbooks for service setups such as creating images and flavors and stuff. is there another role or playbook in ops repo that just setup services after OSA setup? | 09:10 |
noonedeadpunk | tbh not sure. and these playbooks are not good maintained or tested, so it's more like general repo for potentially useful stuff for ops | 09:12 |
jrosser | morning | 09:13 |
noonedeadpunk | o/ | 09:18 |
noonedeadpunk | your input on https://review.opendev.org/#/c/762090/4 would be great in terms of having 2 openrc templates or this complexity... | 09:19 |
jrosser | not not :( | 09:23 |
SiavashSardari | jrosser so no on both approachs? | 09:45 |
jrosser | well the not not is not very clear to read | 09:47 |
jrosser | i was just wondering if the contents of the 'if' were reversed and the _not_ could be removed off the _not_openrc_system_scope | 09:48 |
SiavashSardari | well the _not_ is a workaround for two problems: 1. creating two very similar templates. and 2. openrc files use d differently in OSA roles, and despite of system scoped or not, the openrc file should be project scoped (due to project_id in cinder and manila endpoints) | 09:54 |
SiavashSardari | last week noonedeadpunk suggested that it's better if we add a system scoped openrc file too | 09:56 |
SiavashSardari | so we got in this dilemma of how to tackle the two openrc file problem | 09:57 |
jrosser | i put some comments on the patch | 09:58 |
jrosser | isnt not _not_openrc_system_scope == openrc_system_scope anyway? | 09:59 |
noonedeadpunk | it is | 09:59 |
noonedeadpunk | the thing here is that we define _not_openrc_system_scope after deployment of openrc | 09:59 |
noonedeadpunk | but we could name it _openrc_system_scope surely | 09:59 |
jrosser | Create project scoped openrc file -> dest: "{{ openrc_file_dest }}.system_scope" | 10:03 |
SiavashSardari | I first named it _openrc_system_scope, I thought it will bring more confusion and I added _not '=D | 10:03 |
jrosser | aaaarrrrhhhh! | 10:03 |
noonedeadpunk | S yeah, I think maybe having 2 tempaltes is better?:) | 10:03 |
jrosser | i have no idea whats going on here! :) | 10:04 |
jrosser | well yes i see what we want to acheive | 10:04 |
jrosser | lets just not do it like this :) | 10:04 |
noonedeadpunk | I think agreed then:) | 10:04 |
SiavashSardari | oops guys, my bad. I pushed a wrong version. | 10:05 |
SiavashSardari | let me post a patch | 10:05 |
*** pcaruana has quit IRC | 10:09 | |
openstackgerrit | Siavash Sardari proposed openstack/openstack-ansible-openstack_openrc master: Adding support of system scoped openrc and clouds.yaml https://review.opendev.org/762090 | 10:13 |
SiavashSardari | should I use recheck for zuul gate failures too? | 10:16 |
*** gshippey has joined #openstack-ansible | 10:17 | |
noonedeadpunk | yeah you can do this as well | 10:20 |
*** yann-kaelig has joined #openstack-ansible | 10:26 | |
SiavashSardari | noonedeadpunk about with_items, I like your idea but I don't know how to handle when condition and dest | 10:28 |
noonedeadpunk | https://opendev.org/openstack/openstack-ansible-os_ceilometer/src/branch/master/tasks/ceilometer_post_install.yml#L82 | 10:29 |
noonedeadpunk | just as a sample | 10:29 |
noonedeadpunk | and you add key `condition: "{{ openrc_system_scope | bool }}" to the item | 10:30 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Bump SHAs for master https://review.opendev.org/762762 | 10:57 |
*** jamesgibo has joined #openstack-ansible | 11:07 | |
openstackgerrit | PerToft proposed openstack/openstack-ansible master: Added Netplan examples for a Openstack Infrastructure node. Updated with reviews. https://review.opendev.org/760090 | 11:13 |
*** rpittau|bbl is now known as rpittau | 11:30 | |
openstackgerrit | Siavash Sardari proposed openstack/openstack-ansible-openstack_openrc master: Adding support of system scoped openrc and clouds.yaml https://review.opendev.org/762090 | 11:56 |
openstackgerrit | Merged openstack/openstack-ansible master: Add neutron missing drivers bump https://review.opendev.org/762148 | 12:10 |
*** yann-kaelig has quit IRC | 12:16 | |
admin0 | what does openstack_config: true/false do in case of ceph ? | 12:19 |
*** rfolco has joined #openstack-ansible | 12:19 | |
admin0 | if i want to use an external ceph ( have it not create mons ) , it should be true or false ? | 12:19 |
ebbex | admin0: it creates the ceph-pools | 12:22 |
admin0 | what user-variable to set to have it not create mons ? | 12:22 |
admin0 | but use existing ones | 12:22 |
admin0 | please ignore my question | 12:23 |
admin0 | was my mistake ;) | 12:23 |
admin0 | ok .. one more question .. i am on tag 21.1.0 .. new setup .. want to test magnum(kubernetes) .. does anyone have the exact image/command line to deploy a test setup ? I used the default documents .. and I always end up with timed out on kube_master ( it gets spawned up ) .. but kube minions always fail | 12:27 |
admin0 | i have not been able to get kubernetes working .. i think it could be the image i am work on ( as it requires specific version) | 12:28 |
admin0 | so checking here if anyone is using k8 on osa and can share some notes | 12:28 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Unfreeze roles https://review.opendev.org/762185 | 12:31 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Unfreeze roles https://review.opendev.org/762185 | 12:32 |
ebbex | sorry, played around with magnum and fedora-atomic, but never got it to work. from 21.1.0 you should be using fedora-coreos, remember to set 'os_distro: fedora-coreos' as a property on the image. | 12:32 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Unfreeze roles https://review.opendev.org/762185 | 12:32 |
ebbex | Now i just deploy okd4 (openshift) on it. Has a decent installer, creates bootstrap/masters/workers with terraform. | 12:34 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Unfreeze roles https://review.opendev.org/762185 | 12:34 |
noonedeadpunk | well, we got it working, but I think you might want to bump magnum a bit ahead of the osa, because it might have bugs that have been already fixed in later releases | 12:36 |
noonedeadpunk | as eventually we were always setting specific magnum version | 12:36 |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible master: Unfreeze roles https://review.opendev.org/762185 | 12:38 |
admin0 | ebbex, were you able to get it to work with fedora-coreos ? | 12:41 |
admin0 | ebbex, openshift as a service to openstack ? | 12:41 |
admin0 | to -> on .. | 12:42 |
ebbex | admin0: yeah, helps if you have octavia. | 12:53 |
admin0 | ebbex, got a howto on osa+octavia ? | 12:53 |
ebbex | unfortunately not :( | 12:54 |
*** yann-kaelig has joined #openstack-ansible | 12:56 | |
*** rh-jelabarre has joined #openstack-ansible | 13:01 | |
pto_ | If i need to run a custom playbook/task in an lxc container, is there an easy way to do it with OSA? | 13:02 |
*** pto_ is now known as pto | 13:04 | |
SiavashSardari | @pto just cd to playbooks directory and use ansible command | 13:05 |
pto | I need to run it from ansible tower (awx) | 13:06 |
SiavashSardari | I don't have enough experience with awx, maybe noonedeadpunk can help you | 13:07 |
*** damiandabrowski has joined #openstack-ansible | 13:09 | |
damiandabrowski | hi everyone! | 13:10 |
noonedeadpunk | pto: well I had some experience with AWX, but never tried it with osa. Eventually awx support dynamic inventories, so you can probably pass OSA dynamic_inventory script to awx, but you need to have same deploy host. but not sure how to pass env | 13:11 |
pto | noonedeadpunk: Thanks for the advice. I think i will run the play directly from the bare metal host (infra) and interact with the container instead | 13:11 |
noonedeadpunk | eventaully it would be isnteresting to get awx working with osa | 13:16 |
noonedeadpunk | as well as with vault :) | 13:16 |
noonedeadpunk | (I think ppl already using it) | 13:16 |
noonedeadpunk | admin0: here's good blog post about osa and octavia https://satishdotpatel.github.io//openstack-ansible-octavia/ | 13:21 |
noonedeadpunk | https://developer.rackspace.com/docs/private-cloud/rpc/master/rpc-octavia-internal/octavia-install-guide/ might be also worth reading | 13:22 |
*** jamesgibo has quit IRC | 13:26 | |
*** jamesgibo has joined #openstack-ansible | 13:32 | |
*** jamesgibo has quit IRC | 13:32 | |
jrosser | admin0: getting magnum to work is a big task, i wonder if guilhermesp can share a known working config? | 13:41 |
*** d34dh0r53 has joined #openstack-ansible | 13:45 | |
*** tbarron has joined #openstack-ansible | 13:47 | |
*** SiavashSardari has quit IRC | 13:49 | |
*** jamesgibo has joined #openstack-ansible | 13:51 | |
openstackgerrit | Dmitriy Rabotyagov (noonedeadpunk) proposed openstack/openstack-ansible-os_barbican master: Add deployment of the external libraries https://review.opendev.org/762842 | 13:59 |
*** cshen has quit IRC | 14:02 | |
*** spatel has joined #openstack-ansible | 14:10 | |
admin0 | noonedeadpunk, thanks for that link .. i am going to give it a try | 14:11 |
*** nurdie has joined #openstack-ansible | 14:25 | |
spatel | jamesdenton: can you take a look - https://review.opendev.org/#/c/762729/ | 14:25 |
jamesdenton | done | 14:46 |
spatel | jamesdenton: question, related openvswitch package. Finally today victora RDO got release and it has package name rdo-openvswitch-2.13 which support ovs+dpdk | 14:50 |
spatel | do you think for centOS we should use package name "rdo-openvswitch" instead of openvswitch? | 14:50 |
jamesdenton | i don't think so? check this out: https://review.opendev.org/#/c/762661/ | 14:53 |
spatel | why do we need to add "centos-release-nfv-openvswitch" repo now? | 14:56 |
spatel | now openvswitch-2.13 is already part of default rdo-<release> repo | 14:56 |
spatel | jamesdenton: did i missing something? | 14:57 |
jamesdenton | i think my patch preceeded the release of 2.13 in the rdo repo | 14:58 |
openstackgerrit | James Denton proposed openstack/openstack-ansible-os_neutron master: Test OVS deployments on CentOS 8 https://review.opendev.org/762661 | 15:00 |
jamesdenton | let try it | 15:00 |
spatel | hmm | 15:01 |
jamesdenton | spatel this is the issue I had Friday: v | 15:01 |
jamesdenton | http://paste.openstack.org/show/800059/ | 15:01 |
*** cshen has joined #openstack-ansible | 15:02 | |
spatel | In NFV repo package name is openvswitch2.13 so try "dnf install openvswitch2.13" | 15:03 |
jamesdenton | should openvswitch not be some sort of alias? | 15:03 |
jamesdenton | i'd rather not get in to version specific names if we can avoid it | 15:03 |
spatel | Today rdo announced victoria and they are saying in RDO repo package name is rdo-openvswitch-2.13 http://mirror.centos.org/centos/8/cloud/x86_64/openstack-victoria/Packages/r/ | 15:03 |
spatel | If you install rdo-repo and try dnf install rdo-openvswitch then it will up rdo-openvswitch-2.13 | 15:04 |
spatel | Big problem is after centOS 8.2 default openvswitch+dpdk support dropped out and that us why RDO came in picture | 15:05 |
openstackgerrit | Merged openstack/openstack-ansible-os_neutron master: Remove firewall_driver from securitygoup section due to duplication in agents config file https://review.opendev.org/761815 | 15:12 |
*** macz_ has joined #openstack-ansible | 15:20 | |
*** macz_ has quit IRC | 15:20 | |
*** macz_ has joined #openstack-ansible | 15:20 | |
noonedeadpunk | version specific sound like a trouble | 15:22 |
noonedeadpunk | we need to indetify latest then unless some specific version is set I guess | 15:22 |
spatel | noonedeadpunk: we don't need to stick version into file. | 15:23 |
spatel | just use "dnf install rdo-openvswitch" | 15:23 |
spatel | let me run some test and verify in my lab | 15:26 |
*** dave-mccowan has joined #openstack-ansible | 15:34 | |
*** redrobot has quit IRC | 15:35 | |
spatel | noonedeadpunk: this is what i did, Install fresh CentOS 8 machine and installed "dnf install centos-release-openstack-victoria" | 15:38 |
spatel | Then i installed "dnf install rdo-openvswitch" | 15:38 |
spatel | and voila | 15:38 |
-spatel- [root@localhost yum.repos.d]# ovs-vswitchd --version | 15:39 | |
-spatel- ovs-vswitchd (Open vSwitch) 2.13.2 | 15:39 | |
-spatel- DPDK 19.11.3 | 15:39 | |
*** NewJorg has quit IRC | 15:39 | |
*** NewJorg has joined #openstack-ansible | 15:40 | |
spatel | jamesdenton: noonedeadpunk i don't think we need to mess with any versioning, i found even i do "dnf install openvswitch" its pick up rdo-openvswitch package name so we are good | 15:44 |
*** dave-mccowan has quit IRC | 15:45 | |
spatel | we don't need to do anything there | 15:45 |
jamesdenton | that's without the victoria repo? | 15:45 |
spatel | no | 15:45 |
spatel | you have to install victoria repo. | 15:45 |
jamesdenton | ok, i think that's why i chose the centos-release-nfv-openvswitch repo instead | 15:46 |
spatel | hmm that make sense | 15:46 |
jamesdenton | http://paste.openstack.org/show/800061/ | 15:48 |
spatel | nice! | 15:49 |
jamesdenton | just trying to keep it as neutral as possible | 15:50 |
openstackgerrit | James Denton proposed openstack/openstack-ansible-os_neutron master: Test OVS deployments on CentOS 8 https://review.opendev.org/762661 | 15:51 |
*** Guest64097 has joined #openstack-ansible | 15:51 | |
jamesdenton | so that is not addressing the DPDK component directly, simply fixing openvswitch deployment and testing on centos8 | 15:51 |
spatel | yes.. we should focus on basic OVS deployment | 15:52 |
spatel | DPDK would be add-on | 15:52 |
jamesdenton | yes, i'm sure there are corrections to be made there, too, as you've pointed out | 15:53 |
*** Guest64097 is now known as redrobot | 15:53 | |
spatel | i have replied you to your comments | 15:59 |
admin0 | Could not find the requested service magnum-api: host -- does magnum need more than one declaration ? i have magnum-infra_hosts | 16:00 |
jamesdenton | thanks spatel. | 16:01 |
spatel | jamesdenton: one more change we need to do for dpdk which is change group = "hugetlbfs" in /etc/libvirt/qemu.conf file. i don't know how to do that with ansible. | 16:08 |
spatel | qemu group need to run as hugetlbfs to bind dpdk socket with ovs | 16:08 |
spatel | otherwise you will see permission error in logs | 16:09 |
jamesdenton | k | 16:09 |
jamesdenton | https://github.com/openstack/openstack-ansible-os_nova/blob/master/templates/qemu.conf.j2 | 16:09 |
jamesdenton | may have to span both playbooks | 16:10 |
spatel | Yes but we need to pass variable in user-variables.yml | 16:10 |
spatel | or put some condition id dpdk=enabled then go changed qemu.conf permission also | 16:11 |
spatel | s/id/if | 16:11 |
spatel | i wan't sure so i didn't submit patch, lets see what do you think about it | 16:12 |
*** jamesdenton has quit IRC | 16:13 | |
*** jamesdenton has joined #openstack-ansible | 16:13 | |
jamesdenton | spatel we already have 'ovs_dpdk_support: True/False' that can be used. But we also need to consider differences between OSes here. Not sure the same changes apply across the board | 16:15 |
spatel | That is why we have variable files for each OS platform to tune it according. | 16:16 |
spatel | we have two option changed group permission of OVS from hugetlbfs to qemu or tell other service to use hugetlbfs | 16:18 |
*** jamesgibo has quit IRC | 16:30 | |
*** rpittau is now known as rpittau|afk | 17:41 | |
fanfi | folks, what I need to set (minimum settings) for installing os-magnum service | 17:50 |
guilhermesp | hey there | 17:52 |
guilhermesp | mgnum is an adventure | 17:52 |
guilhermesp | i can share some thoughts tho | 17:52 |
guilhermesp | whats up admin0 ? | 17:52 |
fanfi | :)..thx | 17:52 |
guilhermesp | fanfi: magnum is kinda of straight forward to deploy. you just need to define the host inventories for the service and deploy it. The adventure is: using the right coreos image and set the right labels depending on what you are expecting to have on your k8s clusters | 17:57 |
guilhermesp | i can say now that for train, for example, auto healing ( using magnum auto healer and not draino ) is not working with the tags that are set by default | 17:58 |
fanfi | so it's not usesfull that you expected ? | 18:00 |
openstackgerrit | Jonathan Rosser proposed openstack/openstack-ansible master: Add default simple key to secrets https://review.opendev.org/759082 | 18:10 |
admin0 | guilhermesp, i have magnum setup ( as in the api and containers and the process itself) .. but struggling with the image and the correct syntax to bring it up | 18:26 |
admin0 | like what image to use, what networking overlay etc | 18:26 |
admin0 | issue is it times out on deployment of kube_master ( which at the end deploys) .. but int he stack logs, both kube_master and minions are marked as failed | 18:26 |
*** macz_ has quit IRC | 18:50 | |
*** macz_ has joined #openstack-ansible | 18:51 | |
*** andrewbonney has quit IRC | 19:10 | |
*** miloa has quit IRC | 19:12 | |
guilhermesp | which version are you running admin0 ? | 19:16 |
guilhermesp | and which image? ( fedora-atomic or coreos ) | 19:17 |
guilhermesp | fanfi: it is once you get the right combination of labels :) | 19:17 |
admin0 | i used fedora-atomic | 19:20 |
admin0 | but i was suggested to use core-os | 19:20 |
guilhermesp | yep, fedora-atomic is no longer supported | 19:21 |
guilhermesp | which magnum version? | 19:21 |
guilhermesp | coreos uses ignition instead of cloud init | 19:22 |
guilhermesp | i'd suggest to use magnum >=train | 19:23 |
guilhermesp | and switch to coreos | 19:23 |
admin0 | i am on tag 21.1.0 | 19:27 |
admin0 | not sure which version of magnum it is | 19:27 |
admin0 | and in a new one, tag 21.2.0 | 19:27 |
guilhermesp | so you should basically be able to run clusters with coreos | 19:32 |
admin0 | do you have the cli commands on exactly how to create it ? | 19:34 |
djhankb | I've been trying to get magnum working myself, however I'm using Train 20.0.1 | 19:34 |
djhankb | In my scenario, I am running into an issue getting the resources created in Heat, specifically the VM instances, I believe they are bombing out with some sort of authentication issue. | 19:35 |
djhankb | Sounds similar to what admin0 is describing | 19:38 |
guilhermesp | https://www.irccloud.com/pastebin/Azc7WEJd/ | 19:40 |
guilhermesp | admin0: tried to keep as simple as i could :) | 19:40 |
openstackgerrit | Merged openstack/openstack-ansible-os_barbican master: Reduce number of processes on small systems https://review.opendev.org/759895 | 19:42 |
*** klamath_atx has joined #openstack-ansible | 19:43 | |
guilhermesp | I'd suggest also in case the master is stuck somehow, log into it and check the /var/log/heat-config/* logs and see what failed | 19:45 |
*** gshippey has quit IRC | 19:50 | |
fanfi | guilhermesp oki ...i will try to set it up :) ...thanks for any addition samples commands | 20:00 |
admin0 | guilhermesp, it will take me till today to decode that :) | 20:07 |
admin0 | will try definately | 20:07 |
admin0 | and paste the success/failure tomorrow | 20:07 |
guilhermesp | nice nice | 20:11 |
*** macz_ has quit IRC | 20:18 | |
*** macz_ has joined #openstack-ansible | 20:19 | |
spatel | noonedeadpunk: look like we got enough votes here - https://review.opendev.org/#/c/762185/ | 20:44 |
djhankb | guilhermesp: Thanks for posting that - I just was testing it out and I am still running into an error: (Using the UI) Resource CREATE failed: AuthorizationFailure: resources.kube_masters.resources[0].resources.kube-master-bfv: Authorization failed. | 20:50 |
djhankb | I am not sure if this is an issue with the image, or something else. I've verified that my heat stack works by using the "hello world" HOT file to build a machine | 20:51 |
djhankb | I feel like I have gotten further than I had before using those Labels that you posted, but still failing on the creation of the kube_master | 20:52 |
admin0 | create in progress going on for 28 minutes .. is that normal | 21:01 |
*** MickyMan77 has joined #openstack-ansible | 21:02 | |
djhankb | admin0: if you go to heat stacks, you can click on the stack to view progress to see where it is at in the process | 21:06 |
guilhermesp | not normal admin0 it should be max 10 min. Are you able to login to master and check heat-config logs? | 21:07 |
guilhermesp | maybe you're hitting docker pull limit rate? | 21:07 |
*** hindret has quit IRC | 21:13 | |
*** hindret has joined #openstack-ansible | 21:13 | |
MickyMan77 | Why does openstack add .novalocal to the fqdn hostname and add search domain "openstacklocal novalocal" in the resolve.conf on all my instances ? | 21:17 |
MickyMan77 | Can I change that with some settings like, neutron_dns_domain: "testdomain.local" ? | 21:18 |
djhankb | MickyMan77: I *think* that's set in user_variables.yml - dhcp_domain: foo.bar.com | 21:30 |
MickyMan77 | thx, i will check that.. | 21:42 |
*** jbadiapa has quit IRC | 22:38 | |
*** tosky_ has joined #openstack-ansible | 22:52 | |
*** tosky is now known as Guest41419 | 22:55 | |
*** tosky_ is now known as tosky | 22:55 | |
*** Guest41419 has quit IRC | 22:55 | |
*** cshen has quit IRC | 22:56 | |
spatel | what is -undefined- in inventory-manage.py -l ? | 22:59 |
*** spatel has quit IRC | 23:04 | |
*** cshen has joined #openstack-ansible | 23:52 | |
*** cshen has quit IRC | 23:57 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!