| *** rpittau|afk is now known as rpittau | 07:39 | |
| noonedeadpunk | o/ | 08:02 |
|---|---|---|
| mindthecap | jrosser: I'm using Ubuntu 18.04. | 08:06 |
| mindthecap | Has anyone tried OSA with ceph with multiple clusters? I have multiple AZ-s, where i want to have separate ceph clusters on premise. I have set up single ceph cluster but now i'm stumped how to set up the inventory for the second cluster. | 08:07 |
| mindthecap | AZ=pod in my setup, if that helps (https://docs.openstack.org/openstack-ansible/queens/user/l3pods/example.html). | 08:08 |
| mindthecap | the link is from queens but hasn't changed much.. i'm on victoria. | 08:10 |
| noonedeadpunk | I think you need to create separate groups in the inventory per AZ and define group vars for that | 08:13 |
| mindthecap | yeah... i need to do something like that but don't know exactly how because ceph-ansible uses groups in configuration as well. For example mons config uses groups to find osd host ip-s. | 08:18 |
| jrosser | i think you are going to need 3 independant ceph-ansible inventories completely seperate from OSA | 08:19 |
| jrosser | sorry N not 3 | 08:19 |
| jrosser | mindthecap: generally everything about ceph gets easier as the cluster grows - beware making multiple small ones which then turn out less resilient than one larger one | 08:24 |
| mindthecap | i have multi datacenter deployment so i figured cluster per datacenter would be optimal. | 08:25 |
| jrosser | oh well thats a bit different | 08:26 |
| mindthecap | OSA supports multi site / multi pod deployments and i would really like to have ceph backend to support it aswell... | 08:26 |
| noonedeadpunk | and you have networks shared between AZs or they would be different? | 08:27 |
| mindthecap | networks are per AZ/pod. | 08:29 |
| jrosser | does glance even have an AZ concept? | 08:30 |
| jrosser | thats going to be tough with ceph if it doesnt | 08:30 |
| mindthecap | i can point glance to the ceph cluster in specific AZ, can't i? | 08:31 |
| jrosser | if you go with ceph as your storage then usually glance uses ceph as its storage | 08:32 |
| jrosser | which is a huge advantage as creating a volume to boot from a glance image is a no-op snapshot with no actual data copied | 08:32 |
| mindthecap | yeah... that's best case. Glance supports multiple stores aswell.. | 08:35 |
| noonedeadpunk | afaik glance don't have AZ support, but it has multistorage, which probably can be leveraged, dunno.. | 09:56 |
| noonedeadpunk | but eventually, if you have AZs that has nothing in common - why even make them AZs? | 09:56 |
| noonedeadpunk | I mean you can split services in regions then, with shared keystone | 09:57 |
| noonedeadpunk | and deploy them independently. You still can have single deploy host if you want (just name folder /etc/openstack_deploy_<region> and set OSA_CONFIG_DIR before running openstack-ansible) | 09:58 |
| noonedeadpunk | yeah, that would require extra hardware for controllers, but you want depend on connection between datacenters for operating a cloud | 09:59 |
| noonedeadpunk | since you will have local mysql, rabbit, etc | 09:59 |
| mindthecap | yeah... maybe i need to look at regions.. | 10:41 |
| mindthecap | do you have any reference? | 10:41 |
| noonedeadpunk | I think we don't :( | 10:58 |
| noonedeadpunk | but eventually we have openstack_region variable, that can be set for group of hosts for example | 10:58 |
| mindthecap | yeah but every deployments sets up keystone by default | 10:59 |
| jrosser | i kept it simple and did shared nothing between deployments | 11:02 |
| jrosser | and integrated them all with an external identity provider | 11:02 |
| mindthecap | does that mean that you have horizon integration aswell? | 11:06 |
| noonedeadpunk | yeah, for horizon you can jsut configure regions and relevant keystone endpoints for these regions | 11:07 |
| noonedeadpunk | it's set as variable | 11:07 |
| noonedeadpunk | basically horizon is super independent and can be deployed as a standalone thing at all | 11:08 |
| noonedeadpunk | but yeah, I also prefer shared nothing tbh | 11:08 |
| jrosser | i have N horizons in mine | 11:09 |
| jrosser | lots of choices | 11:10 |
| noonedeadpunk | What I did, I had also N horizons, but each was configured for all regions, so you had a dropdown (with default to current region) but user could pick up any from the list if needed | 11:10 |
| noonedeadpunk | so yeah, lots of choices :) | 11:10 |
| jrosser | did that need the mgmt network spanned across them all or did the horizons use the external endpoints, kind of "from outside" | 11:11 |
| noonedeadpunk | they used public endpoints | 11:11 |
| noonedeadpunk | as for region configuration you set endpoint for keystone anyway | 11:11 |
| jrosser | i already have to patch horizon for wrong endpoints with websso | 11:12 |
| noonedeadpunk | well, we don't use horizon in prod nowadays :( | 11:15 |
| opendevreview | Namrata proposed openstack/openstack-ansible stable/wallaby: Bump role SHA for octavia in stable/wallaby https://review.opendev.org/c/openstack/openstack-ansible/+/800624 | 12:01 |
| MrClayPole | noonedeadpunk: Thanks for your help yesterday. Looks like we may have gnocchi dependencies causing the build to fail. We are just wondering what the best way to override the pip versions. Should we use "repo_build_upper_constraints_overrides" from https://opendev.org/openstack/openstack-ansible-repo_build/src/branch/stable/rocky/templates/requirements_constraints.txt.j2#L15 or "gnocchi_pip_packages" from https:// | 12:55 |
| MrClayPole | opendev.org/openstack/openstack-ansible-os_gnocchi/src/branch/stable/rocky/defaults/main.yml#L169-L176 or is there a 3rd way we've not considered? | 12:55 |
| noonedeadpunk | there's global-requirement-pins.txt in openstack-ansible repo, that is supposed to be respected and have highest prescedence | 12:56 |
| MrClayPole | Thanks, I did look at that but was worried it might affect other OSA components if I set it there. I'll make my changes there. | 13:00 |
| noonedeadpunk | oh, well, it would indeed | 13:33 |
| admin1 | hi guys .. i have never used flat .. and first time using it .. i see this when i do brctl show "br-vlan 8000.3cecef1a946c no eno1" and also this "brq09f673d0-e0 8000.fe163e26667d no tapd8edbdbd-d3" .. which is for a test instance | 14:28 |
| admin1 | how does the brq connect with the eno1 | 14:28 |
| admin1 | oh wait | 14:29 |
| admin1 | i don't have br-flat or flat in my config :D | 14:29 |
| admin1 | i do | 14:29 |
| admin1 | type: flat, net_name: flat , host_bind_override: eno1 | 14:29 |
| admin1 | container_bridge: br-vlan | 14:29 |
| noonedeadpunk | #startmeeting openstack_ansible_meeting | 15:00 |
| opendevmeet | Meeting started Tue Jul 13 15:00:45 2021 UTC and is due to finish in 60 minutes. The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot. | 15:00 |
| opendevmeet | Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. | 15:00 |
| opendevmeet | The meeting name has been set to 'openstack_ansible_meeting' | 15:00 |
| noonedeadpunk | #topic rollcall | 15:00 |
| noonedeadpunk | o/ partically here | 15:01 |
| admin1 | o/ for moral support | 15:03 |
| mgariepy | o/ somewhat busy focussing on other things | 15:04 |
| noonedeadpunk | #topic office hours | 15:05 |
| noonedeadpunk | so, I wasn't really doing much work for the project for previous week, so have close to nothing to share | 15:05 |
| noonedeadpunk | I saw awesome keystone doc that gshippey has pushed regarding federation and saml | 15:06 |
| noonedeadpunk | patches I pushed for vault repo creation and panko deprecation haven't been merged either | 15:07 |
| noonedeadpunk | one of our departments have tested V->W upgrade and it went pretty fine except octavia | 15:07 |
| noonedeadpunk | other than that - dunno, don't have much to discuss from my side. | 15:08 |
| johnsom | noonedeadpunk Something the Octavia team did or an Ansible speed bump? (asking to see if we have something to improve on the Octavia side) | 15:09 |
| noonedeadpunk | no-no, we did spelling check in variable name | 15:09 |
| noonedeadpunk | *spelling mistake | 15:09 |
| noonedeadpunk | octavia did great so far! | 15:10 |
| johnsom | Ah, ok. | 15:10 |
| johnsom | Cool, just making sure. Thanks | 15:10 |
| noonedeadpunk | with amphora on ubuntu 20.04 (and haproxy+keepalived 2+) it's really very stable | 15:10 |
| johnsom | Yeah, faster too | 15:10 |
| noonedeadpunk | Yeah. Before that it was kind of lottery - if amphora get updated or will stuck in pending_update | 15:11 |
| noonedeadpunk | so thanks for your work! | 15:11 |
| johnsom | lol, you are welcome (on behalf of the whole team). | 15:12 |
| jrosser | o/ hello - kind of not here :) travelling home from day at the datacenter | 15:27 |
| noonedeadpunk | I bet it was fun day haha | 15:27 |
| noonedeadpunk | I think I will wrap up the meeting then | 15:27 |
| *** mgoddard- is now known as mgoddard | 15:27 | |
| noonedeadpunk | if nobody wants to add smth for dicussion | 15:27 |
| jrosser | we’ve not had much chance to test W upgrade | 15:28 |
| jrosser | lab in bits to rebuild | 15:28 |
| noonedeadpunk | In our case it was to update OpenStack courses which are based on OSA | 15:29 |
| noonedeadpunk | So these were labs as well | 15:29 |
| noonedeadpunk | But at least smth other then just our CI aio | 15:29 |
| noonedeadpunk | #endmeeting | 15:44 |
| opendevmeet | Meeting ended Tue Jul 13 15:44:42 2021 UTC. Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4) | 15:44 |
| opendevmeet | Minutes: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2021/openstack_ansible_meeting.2021-07-13-15.00.html | 15:44 |
| opendevmeet | Minutes (text): https://meetings.opendev.org/meetings/openstack_ansible_meeting/2021/openstack_ansible_meeting.2021-07-13-15.00.txt | 15:44 |
| opendevmeet | Log: https://meetings.opendev.org/meetings/openstack_ansible_meeting/2021/openstack_ansible_meeting.2021-07-13-15.00.log.html | 15:44 |
| spotz | noonedeadpunk: Hey soorry been in meetings all moorning. FYI on PTO starting tomorrow. And don't forrget to send email for PTG scheduling:) | 16:01 |
| *** rpittau is now known as rpittau|afk | 16:23 | |
| -opendevstatus- NOTICE: Depends-On using https://review.opendev.org URLs are currently not working. This was due to a config change in Zuul that we are reverting and will be restarting Zuul to pick up. | 17:40 | |
| opendevreview | Georgina Shippey proposed openstack/openstack-ansible-os_keystone master: Updates to federation documentation https://review.opendev.org/c/openstack/openstack-ansible-os_keystone/+/800504 | 17:43 |
| *** zbr is now known as Guest681 | 17:50 | |
| spatel | noonedeadpunk i am start doing prepwork for my V -> W upgrade. hope we do not have anything pending | 19:19 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!