| noonedeadpunk | NeilHanlon: are you aware of any progress with libmlx5.so.1(MLX5_1.24)(64bit) for Rocky? | 09:19 |
|---|---|---|
| noonedeadpunk | I think now another thing got broken for ovs2.17 https://zuul.opendev.org/t/openstack/build/0296175692cb41c68f4f9401ac043efd | 09:22 |
| noonedeadpunk | hm, but it's fine on master... | 09:23 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible-repo_server master: Add TLS support to repo_server backends https://review.opendev.org/c/openstack/openstack-ansible-repo_server/+/876429 | 09:27 |
| opendevreview | Merged openstack/openstack-ansible-os_horizon master: Fix horizon_enable_ssl logic https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/879514 | 11:48 |
| NeilHanlon | noonedeadpunk: I met with the NFV Sig and offered to rebuild on top of Rocky, so that's something on my plate... I will try and see about making some progress on it in the next couple of weeks | 12:48 |
| noonedeadpunk | aha, ok, thanks! jsut decided to remind about this thingy so that it didn't slip :) | 12:49 |
| NeilHanlon | heh, thank ya | 12:49 |
| noonedeadpunk | as it would be great to get ovs3.1 working on Rocky eventually | 12:51 |
| NeilHanlon | agreed! | 12:54 |
| NeilHanlon | https://pagure.io/centos-infra/issue/1121 this is the proposal btw | 12:55 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move repo_packages to group_vars https://review.opendev.org/c/openstack/openstack-ansible/+/881824 | 13:01 |
| noonedeadpunk | Looks like it's merged for couple of weeks now? | 13:02 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move repo_packages to group_vars https://review.opendev.org/c/openstack/openstack-ansible/+/881824 | 13:04 |
| NeilHanlon | noonedeadpunk: yeah i just need to do some work now to actually build the software in those tags | 13:24 |
| *** dviroel_ is now known as dviroel | 14:30 | |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Move repo_packages to group_vars https://review.opendev.org/c/openstack/openstack-ansible/+/881824 | 14:58 |
| spatel | is .img format is raw here? - https://cloud-images.ubuntu.com/jammy/current/ | 15:37 |
| mgariepy | i do not think so. | 15:37 |
| mgariepy | if you download it you can check it out | 15:38 |
| spatel | I also think its not raw. I believe need to convert using qemu-img | 15:38 |
| mgariepy | i usualy download the qcows image and convert it to RAW before uploading it to glance when i need it. | 15:38 |
| spatel | hmm | 15:39 |
| mgariepy | you can also use cinder stuff to auto-convert images once. | 15:39 |
| spatel | But on that link there is no qcow2 format, i think only img available | 15:39 |
| mgariepy | https://docs.openstack.org/cinder/latest/admin/image-volume-cache.html | 15:40 |
| spatel | How auto convert? | 15:40 |
| mgariepy | depends on your usecase. | 15:40 |
| spatel | I have ceph storage and planning to use copy-in-write using raw image | 15:41 |
| mgariepy | the first time the image is used cinder will convert it put it in cache and then next time it's used it will use the cache. | 15:41 |
| mgariepy | it also cleanup alone all the stuff i've been using this for years without any issue. | 15:41 |
| spatel | .img is QCOW2 - jammy-server-cloudimg-amd64.img: QEMU QCOW2 Image (v2), 2361393152 bytes | 15:48 |
| spatel | damn why they use so many name for same format | 15:49 |
| mgariepy | meh. | 15:50 |
| mgariepy | with the size it was kinda obvious that it was not raw. | 15:50 |
| NeilHanlon | `qemu-img info` _usually_ gives some helpful info | 16:01 |
| NeilHanlon | otoh, qcow is just raw without the zero padding | 16:01 |
| NeilHanlon | (in fact you can convert an artifact from qcow to raw back to qcow and receive the exact original artifact) | 16:02 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible master: Revert "Add magnum tempest URL" https://review.opendev.org/c/openstack/openstack-ansible/+/881665 | 16:20 |
| opendevreview | Merged openstack/openstack-ansible-os_horizon master: Rename horizon_enable_ssl to horizon_backend_ssl https://review.opendev.org/c/openstack/openstack-ansible-os_horizon/+/879516 | 16:21 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible master: Revert "Add magnum tempest URL" https://review.opendev.org/c/openstack/openstack-ansible/+/881665 | 16:23 |
| opendevreview | Damian Dąbrowski proposed openstack/openstack-ansible master: [DNM] Do not define cluster template labels https://review.opendev.org/c/openstack/openstack-ansible/+/881566 | 16:24 |
| opendevreview | Stuart Grace proposed openstack/ansible-role-python_venv_build master: Fix venv_build_targets when used with multi-arch https://review.opendev.org/c/openstack/ansible-role-python_venv_build/+/881848 | 16:50 |
| spatel | What is wrong here with ceph glance permission? - https://paste.opendev.org/show/bW82D6jeTB9aZVKBMCDF/ | 17:21 |
| MrR | allow rx pool=volumes should be images | 17:27 |
| noonedeadpunk | spatel: I usually use profile rbd to be frank | 17:31 |
| noonedeadpunk | that should work though | 17:32 |
| spatel | MrR ? I do have both image/volume - allow rwx pool=images, allow rx pool=volumes" | 17:32 |
| spatel | noonedeadpunk what do you mean profile rbd? | 17:33 |
| mgariepy | yeah noonedeadpunk is right you need profile rbd. | 17:33 |
| spatel | sorry I am new in ceph and still learning so point me to profile and how does it work? | 17:34 |
| noonedeadpunk | it's documented here https://docs.ceph.com/en/latest/rados/operations/user-management/#authorization-capabilities | 17:34 |
| noonedeadpunk | and here's example on usage https://docs.ceph.com/en/mimic/rbd/rados-rbd-cmds/#create-a-block-device-user | 17:35 |
| mgariepy | https://github.com/ceph/ceph-ansible/blob/main/group_vars/all.yml.sample#L644 | 17:35 |
| spatel | Thanks! reading.. but in my example even it has all permission but still getting error of permission | 17:38 |
| mgariepy | you don't have the profile `rbd` for mon and osd.. | 17:39 |
| mgariepy | Gives a user permissions to manipulate RBD images. When used as a Monitor cap, it provides the minimal privileges required by an RBD client application. When used as an OSD cap, it provides read-write access to an RBD client application. | 17:41 |
| MrR | spatel glance doesn't need access to volumes, both should be images, at least, thats how it is in my case and works | 17:45 |
| spatel | MrR do you have cinder? | 17:48 |
| MrR | yes | 17:48 |
| noonedeadpunk | yeah, glance doesn't need access to volumes, that's true. It satisfies it's needs with `class-read object_prefix rbd_children` | 17:49 |
| spatel | mgariepy I can upload image on glance but can't delete so why do you think because of profile its causing issue. Let me dig into it.. cinder/nova everything works but glance not | 17:50 |
| mgariepy | not sur the mon 'r' is equivalent to profile rbd. | 17:53 |
| jrosser | don’t we have an example set of these in our vars for the ceph users? | 17:54 |
| jrosser | there must at least be a sufficiently good default set for the ceph AIO | 17:55 |
| mgariepy | spatel, waht are your caps for cinder and nova ? | 17:56 |
| mgariepy | for glance i have mon: profile rbd. and osd: profile rbd for each pool | 17:57 |
| MrR | On a seperate note, what i came for is i'm currently struggling with networking, neutron/ovn are running and i can create networks etc.. but i can't get the "public" network setup working properly. My internet facing network is currently one flat network (i have this setup as br-outrside/bond1 on each node right now if that matters), i know it's a combination of -network in openstack_user_config.yml and | 18:00 |
| MrR | neutron_provider_networks in user_variables.yml but i just can't get it right. With how i have it now, i can boot instances on the public network and see its ip pop up on my physical hardware, but when i try and add a floating ip to a neutron created network it claims it gets an ip but doesnt pop up on my physical hardware and i can't actually access that ip either so i know something somewhere is wrong. What am i | 18:00 |
| MrR | missing? | 18:00 |
| MrR | bah i keep forgetting paste works weird | 18:00 |
| mgariepy | MrR, how did you configure your nodes ? did you create the br-outsite manually or you added it to your config ? | 18:04 |
| mgariepy | can you paste your config ? | 18:05 |
| MrR | https://paste.opendev.org/show/b2VR20soyaIWRbrE1Gg2/ | 18:05 |
| MrR | was just doing that | 18:05 |
| MrR | i created the br-outside manually | 18:05 |
| mgariepy | openstack tells you that the port is active in ovn ? | 18:06 |
| MrR | not added it to any openstack/ansible config, it was just direct access point on bond1 | 18:06 |
| MrR | yes | 18:06 |
| mgariepy | ovn-nbctl show and ovn-sbctl show | 18:06 |
| mgariepy | also when pasting yaml please try to keep the indentation :D | 18:07 |
| mgariepy | you might want to hide some ips in the output | 18:07 |
| MrR | will do, pulled it from a txt document as i meant to address this weeks ago haha | 18:08 |
| MrR | neither command works assuming i'm running this on the host, ovn-sbctl: unix:/var/run/ovn/ovnsb_db.sock: database connection failed (No such file or directory) is the error | 18:10 |
| MrR | yet network agent list shows me ovn/metadata as up | 18:10 |
| MrR | this is on debian | 18:11 |
| mgariepy | need to run from the controllers. | 18:11 |
| mgariepy | where `network-northd_hosts` is. | 18:12 |
| mgariepy | probably on the master one also. | 18:12 |
| mgariepy | are you deployed with lxc containers? | 18:14 |
| mgariepy | or on metal ? | 18:14 |
| MrR | lxc, running from northd container gets me a result | 18:15 |
| MrR | https://paste.opendev.org/show/bDvpo0sn3p6XM4g2ZJrw/ | 18:23 |
| MrR | be assured that xx.xx and xx.xx.x match haha | 18:23 |
| MrR | where they should anyway | 18:23 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-config_template master: Add support for start/end_string arguments https://review.opendev.org/c/openstack/ansible-config_template/+/881879 | 18:26 |
| MrR | i'm on 26.0.1 if that also matters as i noticed 26.1 is being worked on but i assume not yet ready as a proper release | 18:27 |
| mgariepy | can you paste the config of the compute 01? | 18:27 |
| MrR | same commands? | 18:28 |
| mgariepy | in theory the external port should be on this one. | 18:28 |
| mgariepy | ovs-vsctl show | 18:30 |
| mgariepy | for a start. | 18:30 |
| MrR | ovn- is only working on the node i pulled it from in the paste, both other nodes give me ovn-sbctl: unix:/var/run/ovn/ovnsb_db.sock: database connection failed () | 18:34 |
| MrR | ovs-vsctl show on the host right? | 18:34 |
| MrR | thats the only place it works anyway | 18:34 |
| mgariepy | yep | 18:35 |
| mgariepy | ovS-vsctl ;p | 18:35 |
| MrR | https://paste.opendev.org/show/bqrl5di2WwVgTTlVIeF3/ | 18:36 |
| MrR | as they are down i'm guessing i've missed a step | 18:38 |
| MrR | although i can access both those ips | 18:38 |
| jamesdenton | to use ovn-nbctl or ovn-sbctl on the non-master nodes you'll need to use: ovn-sbctl --db=tcp:<master ip>:6642 show, for example | 18:47 |
| spatel | jamesdenton there is a better way ( ovn-sbctl --no-leader-only show ) | 19:23 |
| spatel | I found my glance issue. There was a VM running using that image. as soon as i delete vm then i was allow to delete image. is that normal? | 19:25 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-config_template master: Add support for start/end_string arguments https://review.opendev.org/c/openstack/ansible-config_template/+/881879 | 19:25 |
| spatel | jamesdenton we should document ovn tips-n-tricks somewhere in OVN section | 19:29 |
| jamesdenton | yes, please do! that's a neat one | 19:30 |
| spatel | +1 | 19:30 |
| MrR | Thanks for that, as the first command didnt work, spatel's does | 19:30 |
| MrR | already pasted my results above tho | 19:31 |
| mgariepy | i got sidetracked with other stuff but i wonder how the playbook did configure ovn/ovs for your compute. usually with ovn you probably don't need to have the bridge pre-created. | 19:32 |
| mgariepy | o wonder if this one did create the bridge in ovs and ignore the one or map it differently : https://github.com/openstack/openstack-ansible-os_neutron/blob/master/tasks/providers/setup_ovs_ovn.yml#L55-L65 | 19:35 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/ansible-config_template master: DNM https://review.opendev.org/c/openstack/ansible-config_template/+/881887 | 19:35 |
| MrR | no worries, if we are talking my br-outside bridge i created that before i had grasped networking for ansible just as a point of access on that network | 19:36 |
| spatel | make sure bond1 is not part of other bridges. I had that issue and found ovs doesn't like that. | 19:37 |
| spatel | it won't complain but you can't ping vm. | 19:37 |
| mgariepy | anyhow if you find the issue please update us on it. i'm off for the weekend now :D | 19:38 |
| spatel | mgariepy have a good weekend!! have some cold beers :) | 19:38 |
| mgariepy | hehe yeah don't worry my beers are cold and waiting. | 19:38 |
| spatel | hayyyy | 19:39 |
| MrR | just checked and i have nothing else on bond1 apart from the vlans on bond1.10 etc | 19:42 |
| MrR | i'll pop back Monday for some more assistance :) | 19:43 |
| spatel | it won't work if bond1.10 has IP etc... | 19:43 |
| spatel | If its lab then i would say split interface from bond1 and assign dedicated nic to ovs | 19:44 |
| MrR | my setup is identical to this: https://docs.openstack.org/openstack-ansible/latest/user/prod/example.html | 19:45 |
| MrR | well, in the segragation of bonds etc.. ips are different | 19:46 |
| MrR | so an extra independent bond/nic is needed to get it working? i have spare ports/cards | 19:47 |
| spatel | i would say try that out on extra nic port | 19:51 |
| MrR | i'll give that a go and report back monday | 19:51 |
| spatel | I have strong feeling issue related to that because I had similar issue and found it was nic isolation related | 19:51 |
| spatel | It works with LinuxBridge but OVS doesn't like that. | 19:52 |
| spatel | I am very curious so please report here, or i will ask you :) | 19:52 |
| MrR | i definately will, i'm at a point now where i either get this working or blow it all up lol. I try and find solutions myself but openstack in general has defiantly tested my patience. I still have at least one issue after this but it could be related to this so i'll get to that another time | 19:55 |
| MrR | oh in fact its bank holiday, might be tuesday, depends on plans | 19:55 |
| spatel | +1 | 19:56 |
| MrR | have a good weekend. i'm out, bye now | 19:58 |
| spatel | you too | 20:01 |
| damiandabrowski | seems like we have some gating issue again :| this time it's related to cloudsmith repo and its gpg key | 20:06 |
| damiandabrowski | jobs fail with this error: | 20:07 |
| damiandabrowski | Failed to update apt cache: E:Failed to fetch https://dl.cloudsmith.io/public/rabbitmq/rabbitmq-erlang/deb/ubuntu/dists/jammy/InRelease 402 Payment Required [IP: 2600:9000:2011:a000:e:f4d2:20c0:93a1 443] | 20:07 |
| damiandabrowski | E:The repository 'https://dl.cloudsmith.io/public/rabbitmq/rabbitmq-erlang/deb/ubuntu jammy InRelease' is no longer signed. | 20:07 |
| spatel | anyone know how to clean up data from pool? I did rados -p images cleanup and didn't work | 20:07 |
| damiandabrowski | example: https://zuul.opendev.org/t/openstack/build/071097b872f24ca0b55feadeab759c56/console | 20:07 |
| jrosser | damiandabrowski: maybe rabbitmq have not paid their bill there? | 20:36 |
| damiandabrowski | yeah, probably | 20:37 |
| damiandabrowski | or reached the bandwidth limit | 20:37 |
| damiandabrowski | balena was struggling with the same issue and decided to leave cloudsmith: | 20:38 |
| damiandabrowski | https://github.com/balena-io/etcher/issues/4057 | 20:38 |
| damiandabrowski | but on the other hand: I completely don't understand why rabbitmq playbook passes successfully and I saw this error during nova/cinder playbook execution | 20:41 |
| damiandabrowski | ah, cloudsmith has some issues with checksums for redhat | 20:50 |
| damiandabrowski | but i guess they might have broken something else during troubleshooting | 20:50 |
| damiandabrowski | https://status.cloudsmith.io/ | 20:50 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!