| jrosser | morning | 08:35 |
|---|---|---|
| *** priteau_ is now known as priteau | 08:58 | |
| noonedeadpunk | mornings | 09:00 |
| noonedeadpunk | ah. all rechecks are issued - thanks :) | 09:28 |
| noonedeadpunk | Hm, I wonder if that is correct for zfs scenario: https://zuul.opendev.org/t/openstack/build/030e6482a2094616bf2a44866495ca75/log/logs/host/syslog.txt#3092 | 09:33 |
| jrosser | noonedeadpunk: on my nodes /var/lib/machines is ext4, and i think that the base image is then copied to pool/var/lib/lxc when it is imported into lxc | 09:56 |
| noonedeadpunk | yeah, likely you're right.... | 09:57 |
| jrosser | a tar file is created then imported into lxc here https://github.com/openstack/openstack-ansible-lxc_hosts/blob/master/tasks/lxc_cache_create.yml#L71 | 09:58 |
| jrosser | `pool/var/lib/lxc/ubuntu-focal-amd64 369M 836G 369M /var/lib/lxc/ubuntu-focal-amd64/rootfs` | 09:58 |
| noonedeadpunk | that is exactly the task that fails fwiw | 09:59 |
| jrosser | yes i was just seeing that here https://zuul.opendev.org/t/openstack/build/9933567162054355a53c5e137127e38e/log/job-output.txt#7928 | 10:00 |
| jrosser | `--quiet` may not be helping there to see what is wrong | 10:01 |
| jrosser | maybe we need a hold on that job | 10:03 |
| jrosser | at least the command that it tried is printed, so can be done manually to see what is going on | 10:03 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Add variable to control no_log in db_setup role https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869546 | 10:10 |
| noonedeadpunk | Well, I can't reproduce the issue in aio - I tried 5 times for sure | 10:14 |
| noonedeadpunk | But yeah, removing --quiet might be helpful | 10:16 |
| admin1 | radosgw-admin bucket list will return something like: 711138fc95764303b83002c567ce0972/demo ( where the uuid is with the bucket name) .. I have an openstack where ceph is done separately using ceph-ansible and even using the exact same configs in ceph.conf, bucket list returns just the bucket name without the UUID .. what this is doing is | 10:17 |
| admin1 | when I try to make the bucket public, it says NoSuckBucket ... checking if anyone has seen this before or knows a fix | 10:17 |
| admin1 | the first one (working) is from an AIO . | 10:18 |
| opendevreview | Merged openstack/openstack-ansible-os_octavia stable/xena: Use focal amphora test image by default https://review.opendev.org/c/openstack/openstack-ansible-os_octavia/+/869037 | 10:21 |
| noonedeadpunk | Are you sure that uuid is bucket name and not project uuid? | 10:32 |
| noonedeadpunk | jrosser: nasty typo in 869546 | 10:34 |
| jrosser | doh | 10:34 |
| noonedeadpunk | btw I don't really see how to drop quiet from lxc command :( | 10:34 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Add variable to control no_log in db_setup role https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869546 | 10:34 |
| noonedeadpunk | it's hardcoded, doh https://github.com/ansible-collections/community.general/blob/main/plugins/modules/lxc_container.py#L826 | 10:35 |
| noonedeadpunk | the hold it is | 10:36 |
| admin1 | noonedeadpunk, it is project uuid in aio .. and i am trying to replicate that somehow | 10:51 |
| noonedeadpunk | Are you sure you have rgw_swift_account_in_url = true in your conf? | 10:52 |
| noonedeadpunk | Also keystone endpoint format matters there | 10:52 |
| admin1 | endpoint is https://domain.com:8080/swift/v1/AUTH_%(tenant_id)s | 10:53 |
| admin1 | and config is exactly the same as AIO -> https://lists.ceph.io/hyperkitty/list/ceph-users@ceph.io/thread/EY47EPAAUL5W6U2MQNWEO5F3ITFHUZB3/ | 10:53 |
| admin1 | using openstackcli, and also horizon, i can create buckets, upload objects, download objects etc and it works .it breaks when the bucket is set to mode public | 10:55 |
| noonedeadpunk | Hm, we have `rgw_enable_apis = swift` not `rgw_enable_apis = swift,s3` in AIO? | 10:55 |
| noonedeadpunk | yeah, only swift, you're right | 10:56 |
| admin1 | what i have is one aio up and running, and i have replicated all configs from there | 10:57 |
| noonedeadpunk | yeah. I see | 10:57 |
| noonedeadpunk | maybe smth in openrc.... | 10:58 |
| noonedeadpunk | but horizon is valid argument | 10:58 |
| noonedeadpunk | Eventually, I faced issues with trying to make bucket public, but it was quite a while ago and iirc related to bug in rgw that was fixed | 10:59 |
| hamburgler | hey all! Was wondering with the openstack-ansible zed release as of December if the OVN deployment is considered production ready now? TIA :) | 11:48 |
| opendevreview | Merged openstack/openstack-ansible-os_octavia stable/wallaby: Use focal amphora test image by default https://review.opendev.org/c/openstack/openstack-ansible-os_octavia/+/869038 | 12:06 |
| admin1 | noonedeadpunk, does a service name make a difference ? service name swift vs radosgw ? | 12:09 |
| moha7 | Is it necessary to `have nova_nova_conf_overrides:\n DEFAULT:\n force_config_drive: true` in the user_variables file? I don't have it currently and metadata (as in hostname) is not assigned to the instances! the log an instance (Cirros): checking http://169.254.169.254/2009-04-04/instance-id, failed 1/20: up 5.69. request failed | 12:09 |
| admin1 | i think that is the diff i have | 12:10 |
| jrosser | radosgw is providing the swift service | 12:10 |
| jrosser | moha7: that should not be necessary | 12:11 |
| noonedeadpunk | moha7: it depends on your networking actually. In case metadata is accessed through net - there should be either l3 router or dhcp server exist for the network | 12:11 |
| jrosser | ^ neturon l3 router / dhcp agent | 12:11 |
| admin1 | i mean in endpoint list, i added object-storage with service name: switft . in AIO i see it added as radosgw | 12:12 |
| admin1 | so maybe that is what is causing mine to not work .. ( not creating buckets with the project uuid ) | 12:12 |
| admin1 | that is, if service-name makes a difference | 12:13 |
| jrosser | i would not expect that to make a difference | 12:14 |
| admin1 | changed to radosgw .. did not made any diff | 12:20 |
| noonedeadpunk | and I assume ceph versions output is the same? | 12:50 |
| noonedeadpunk | pleasent thing here is that aio works :D | 12:50 |
| moha7 | which one is correct for: `neutron_plugin_base` --> `neutron.services.ovn_l3.plugin.OVNL3RouterPlugin` or `networking_ovn.l3.l3_ovn.OVNL3RouterPlugin`? | 12:52 |
| opendevreview | Merged openstack/openstack-ansible stable/zed: Add gate_log_requirements function https://review.opendev.org/c/openstack/openstack-ansible/+/869025 | 12:53 |
| moha7 | jrosser: then what's the reason of `checking http://169.254.169.254/2009-04-04/instance-id, failed`, any previous experience? (ref: https://serverfault.com/a/1114711 ) | 12:56 |
| jrosser | if for some reason 169.254.169.254 is not accessible from your instance then cloud-init will fail to get the metadata | 12:57 |
| jrosser | you can choose to use either http or config drive | 12:57 |
| jrosser | the default in openstack-ansible is http | 12:58 |
| moha7 | noonedeadpunk: Regarding accessing metadata through net or locally, what's the default behavior? I did not set any special config in user_variables.yml | 12:58 |
| jrosser | moha7: the default is to use http | 12:59 |
| jrosser | there would not be anything in user variables as cloud-init is baked into the images you use | 12:59 |
| jrosser | it will try the http method | 12:59 |
| jrosser | moha7: do you have a neutron router in your network? | 13:00 |
| moha7 | no, I don't created a router yet. | 13:02 |
| moha7 | didn't have* | 13:02 |
| moha7 | Is it related to the issue? | 13:03 |
| jrosser | yes | 13:06 |
| noonedeadpunk | damn, I'm again struggling without host_containers /o\ | 13:06 |
| jrosser | moha7: the neutron router is the thing that understands where 169.254.169.254 is | 13:06 |
| jrosser | and this relies on the gateway of your network having a route to that IP | 13:07 |
| noonedeadpunk | I need suuuuper simple thing..... | 13:09 |
| mgariepy | hey i'm back ! | 13:09 |
| mgariepy | happy new year everyone. | 13:10 |
| jrosser | opendev.org cloning is sooooo slow for met today | 13:10 |
| noonedeadpunk | And I made it working for Xena ;( | 13:12 |
| moha7 | I would appreciate it if it's possible to see sample config files (`user_variables.yml`, `openstack_user_config.yml`) of your lab environments (Zed with the OVN networking stack.). | 13:12 |
| jrosser | moha7: did you make an all-in-one? this is the reference | 13:13 |
| moha7 | jrosser: I created an internal network (for example, named netint) that is necessary to create an instance. In this case, it can not inject hostname and key pairs (metadata) into the instance, but the instance selects an IP from the `netint` subnet range. | 13:17 |
| moha7 | Then, I created a router, named r1, that one of its hand is in the *netin* network, what about the other hand of router? | 13:17 |
| jrosser | that does not really matter for metadata | 13:18 |
| jrosser | did you get an IP from dhcp? | 13:18 |
| moha7 | jrosser: I tried several times, but I have more problems with AIO and finally it was not deployed! At least now I have brought up the Muktinode environment without errors, but there is no reference for its network settings | 13:19 |
| moha7 | jrosser: Yes, it gets IP from DHCP of netint | 13:20 |
| jrosser | the thing is we run the AIO multiple times every day as our CI job, it is exactly the same | 13:20 |
| jrosser | if you are having trouble with it then either 1) we have a bug with the AIO 2) the documentation is wrong 3) something in your environment | 13:20 |
| jrosser | anywan | 13:20 |
| jrosser | anyway.... | 13:20 |
| moha7 | Previously I had same issue with OVS, and solved with this solution: https://serverfault.com/a/1114711 | 13:21 |
| jrosser | if you are getting an IP with DHCP then the neutron DHCP agent will give the instance some routes | 13:21 |
| jrosser | but you don't have a router? | 13:21 |
| jrosser | sorry but i keep saying that without a neutron router then i don't think you are going to have success with the metadata over http | 13:22 |
| moha7 | 2) The documentation does not explain the issue well. I went exactly from the document and every time I did not succeed even though I posted the problems here. | 13:23 |
| noonedeadpunk | jrosser: but isn't without router metadata spawned in dhcp namespace? | 13:23 |
| jrosser | perhaps i don't understand properly | 13:24 |
| jrosser | but i have `169.254.169.254 via 132.185.118.101 dev eno1np0 proto dhcp src 132.185.118.225 metric 101` | 13:24 |
| jrosser | this route sends the metadata request to the router IP | 13:24 |
| jrosser | oh well... let me check that actually | 13:25 |
| noonedeadpunk | It depends | 13:25 |
| jrosser | if there is no router then is that the IP of the dhcp agent? | 13:25 |
| noonedeadpunk | You can force metadata from dhcp. But default is server from l3 if l3 exists and from dhcp if it doesn't | 13:25 |
| jrosser | there must be a route though, else how can you curl that IP? | 13:26 |
| moha7 | jrosser> to check if I have arouter there, on which LXC container I should run `ip netns`? I tried on utility and neutron_server and didn't returned anything! | 13:26 |
| noonedeadpunk | but route can be provided by dhcp? | 13:27 |
| jrosser | oh yes sure | 13:27 |
| noonedeadpunk | or well, it's provided by dhcp anyway | 13:27 |
| jrosser | moha7: the neutron agents are not in containers, just on the hosts | 13:28 |
| noonedeadpunk | https://docs.openstack.org/neutron/latest/configuration/dhcp-agent.html#DEFAULT.force_metadata | 13:28 |
| jrosser | moha7: if you can give specific issues with the AIO then we can try to replicate them and help | 13:28 |
| noonedeadpunk | Oh, sorry, https://docs.openstack.org/neutron/latest/configuration/dhcp-agent.html#DEFAULT.enable_isolated_metadata | 13:28 |
| jrosser | well anyway those both default to false | 13:29 |
| noonedeadpunk | but we have that enabled by default | 13:29 |
| jrosser | oh :) | 13:29 |
| noonedeadpunk | https://opendev.org/openstack/openstack-ansible-os_neutron/src/branch/master/templates/dhcp_agent.ini.j2#L19 | 13:29 |
| jrosser | anyway isnt this all different for OVN anyway? | 13:29 |
| noonedeadpunk | it is ) | 13:30 |
| moha7 | jrosser: Sure, I'll install AIO again and reports issues here | 13:30 |
| jrosser | moha7: you are using the Zed release? | 13:31 |
| moha7 | Yes, stable/zed | 13:31 |
| moha7 | > neutron agents are on the hosts | 13:32 |
| moha7 | So, `ip netns` on the controller hosts? | 13:32 |
| jrosser | well yes but i don't know how this works for OVN at all | 13:33 |
| noonedeadpunk | it's all in OVN somewhere somehow... | 13:35 |
| noonedeadpunk | worth summoning jamesdenton | 13:36 |
| moha7 | The author of this blog post https://satishdotpatel.github.io/openstack-ansible-multinode-ovn/ uses this option `neutron_metadata_checksum_fix: False` in its user_variables.yml. I'm going to add it there, maybe solved! | 13:36 |
| moha7 | `ip netns` has no output on all hosts. @jamesdenton | 13:37 |
| noonedeadpunk | blogpost ^ it's not fully relevant for Zed | 13:37 |
| jrosser | i'm not sure `neutron_metadata_checksum_fix` even exists? | 13:39 |
| jrosser | codesearch suggests that it does not | 13:39 |
| moha7 | )': I need a network config file as a reference! | 13:40 |
| jrosser | not for an AIO :) | 13:40 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Limit maximum number of threads for parallel git clone https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869564 | 13:48 |
| jrosser | ^ ampere fix | 13:49 |
| andrewbonney | noonedeadpunk: might there be something wrong with the git sha bump script? Just noticed the stable/zed hash for Neutron is quite a way behind, and didn't move forward in https://github.com/openstack/openstack-ansible/commit/b2b61f792eae06a3b5042065f9ef0f5439ca356e | 14:03 |
| moha7 | by adding `neutron_metadata_checksum_fix` to user_variables.yml file and running `os-neutron-install.yml`, now the metadata issue solved! | 14:03 |
| moha7 | spatel: ^ | 14:04 |
| spatel | noonedeadpunk i am having issue with zed - https://paste.opendev.org/show/bXer4dyXV5911o8aWBrU/ | 14:04 |
| noonedeadpunk | andrewbonney: we had a bug in Neutron so I manually bumped it behind | 14:04 |
| spatel | moha7 +1 | 14:04 |
| andrewbonney | Ah fair enough - I must have missed it | 14:04 |
| noonedeadpunk | specifically https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/869042 | 14:05 |
| noonedeadpunk | wtf is neutron_metadata_checksum_fix.... | 14:05 |
| jamesdenton | mornin | 14:07 |
| moha7 | spatel: Have you updated the blog post you wrote to reflect the latest changes in Zed? It does not have some features, for example: `network-northd_hosts: *controller_hosts` and `network-gateway_hosts: *compute_hosts` | 14:07 |
| spatel | moha7 i started playing with zed but stuff in git clone issue which i want to discuss with noonedeadpunk | 14:08 |
| jamesdenton | i do not think it's updated, yet. seeing issues, moha7? | 14:08 |
| spatel | My blog lab only tested on wallaby release. I want to test on zed | 14:10 |
| moha7 | I deployed a multinode env successfully. I had an issue with injecting metadata in instances that is solved by `neutron_metadata_checksum_fix: false` as is used in that blog posts. The only remaining issue is the provider network that seems there's no config set in the neutron settings! | 14:10 |
| jamesdenton | in ml2_conf.ini you don't see anything under ml2_type_vlan? | 14:12 |
| spatel | moha7 what is the deal of provider network? | 14:12 |
| moha7 | Everything works other than the external network. Not possible to create external network saying: ` Error: Failed to create network netext. Details: Invalid input for operation: physical_network 'provider' unknown for VLAN provider network. Neutron server returns request_ids: ['req-88e75dd5-f176-4de2-8582-4b160b9761f2'] ` | 14:12 |
| moha7 | Should I have create any bridge manually? | 14:13 |
| jamesdenton | in openstack_user_config.yml did you setup provider_networks? | 14:13 |
| jamesdenton | or did you do that in user_variables.yml? | 14:13 |
| opendevreview | Merged openstack/openstack-ansible stable/yoga: Disable sahara tempest tests https://review.opendev.org/c/openstack/openstack-ansible/+/867152 | 14:14 |
| moha7 | jamesdenton: http://ix.io/4kJ5 <-- openstack_user_config.yml | 14:15 |
| jamesdenton | perfect. see the one whose container_bridge is br-provider? | 14:15 |
| jamesdenton | and see how the net name is 'vlan'? | 14:15 |
| jamesdenton | net_name is the label | 14:15 |
| moha7 | user_variables.yml: http://ix.io/4kJ6 | 14:15 |
| jamesdenton | the provider label. so, create your provider network with --provider-physical-network vlan instead of --provider-physical-network provider | 14:16 |
| moha7 | I didn't get this : "see the one whose container_bridge is br-provider", May you explain it more? | 14:16 |
| jamesdenton | container_bridge: "br-provider" | 14:16 |
| jamesdenton | line 38 | 14:17 |
| jamesdenton | https://paste.opendev.org/show/bdxw1T43FYhc1wON4mpf/ | 14:17 |
| jamesdenton | line 44 is 'net_name'. | 14:17 |
| jrosser | spatel: i am just doing stable/zed AIO and i have no problem with git clone at all | 14:18 |
| spatel | I tried on 3 machine and encounter same error | 14:19 |
| spatel | I am using ubuntu 20.04.5 release | 14:19 |
| moha7 | Yesssssssss! the name was 'vlan'. Created. | 14:20 |
| spatel | if i switch to 25.x.x tag then no error but switch to 26.0.0 encounter git error | 14:20 |
| spatel | I am clueless :( | 14:20 |
| jamesdenton | moha7 nice! | 14:20 |
| moha7 | Thanks | 14:20 |
| jrosser | spatel: what is `neutron_metadata_checksum_fix` | 14:20 |
| jamesdenton | sure. If you run 'ovs-vsctl list open_vswitch', and look at external_ids, you'll see the label:interface mapping in ovn-bridge-mappings. ie. ovn-bridge-mappings="vlan:br-provider" | 14:21 |
| spatel | I am trying to recall.. what the heck is checksum_fix | 14:22 |
| jrosser | spatel: well this is the thing - i can't find it | 14:22 |
| spatel | moha7 can you remove neutron_metadata_checksum_fix and try.. because you said you encounter error | 14:23 |
| spatel | Its been long time now do i have no recollection of that flag. | 14:23 |
| moha7 | Ok, I remove `neutron_metadata_checksum_fix` again and see what happens for the metadata | 14:24 |
| jamesdenton | FYI: with OVN you should see some 'ovnmeta' namespaces, likely on the computes. The traditional DHCP and Metadata agents should not be running, though someone mentioned they might be installed and i need to look into that | 14:25 |
| jrosser | jamesdenton: i'm just running a stable/zed AIO so can check some of this | 14:26 |
| jamesdenton | kk | 14:26 |
| spatel | jrosser let me give you full output of bootstrap-ansible | 14:28 |
| spatel | jrosser holy crap! it works now.. without error | 14:30 |
| spatel | that is freaking odd... | 14:30 |
| spatel | should i blame internet for it? | 14:30 |
| spatel | or git? | 14:31 |
| jamesdenton | the internet is a series of tubes. and sometimes those tubes have problems. | 14:35 |
| spatel | Very odd that yesterday 3 time i got same error on 3 different machine.. but this morning every just works. scary | 14:36 |
| spatel | jamesdenton we have no way to migrate from lxb to ovn correct? (even with small downtime) | 14:37 |
| jamesdenton | https://www.jimmdenton.com/migrating-lxb-to-ovn/ | 14:37 |
| jamesdenton | no automated way | 14:37 |
| spatel | so there are no way we can put your blog steps in playbook :) | 14:41 |
| jamesdenton | Well, someone prob could. There's nothing too crazy from what i recall. I would not want OSA responsible for that | 14:42 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-os_ironic master: Update IPA image for the Zed release https://review.opendev.org/c/openstack/openstack-ansible-os_ironic/+/869570 | 14:49 |
| jrosser | andrewbonney: ^ | 14:49 |
| noonedeadpunk | jrosser: oh... zfs thingy is interesting.... | 14:53 |
| jrosser | hmm | 14:53 |
| noonedeadpunk | Seems like pool name is trimmed somehow https://paste.openstack.org/show/bH7ZMrIXYbkj4IoZCGNm/ | 14:54 |
| noonedeadpunk | But yes, I think it depends on volume or not https://opendev.org/openstack/openstack-ansible/src/branch/master/tests/roles/bootstrap-host/tasks/prepare_loopback_zfs.yml#L29-L37 | 14:55 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Add reminder to contributor docs to update amphora/IPA images https://review.opendev.org/c/openstack/openstack-ansible/+/869574 | 14:55 |
| jrosser | we gave a custom pool name i think? | 14:55 |
| noonedeadpunk | Yup | 14:56 |
| noonedeadpunk | Will push patch now | 14:57 |
| jrosser | well here for the loopback case https://github.com/openstack/openstack-ansible/blob/e697bed2cea5ea5e49ae7f03c10650b8aec77bc8/tests/roles/bootstrap-host/tasks/prepare_loopback_zfs.yml#L30 | 14:57 |
| jrosser | argh https://github.com/openstack/openstack-ansible/blob/e697bed2cea5ea5e49ae7f03c10650b8aec77bc8/tests/roles/bootstrap-host/tasks/prepare_data_disk.yml#L117 | 14:57 |
| jrosser | i see | 14:57 |
| jrosser | why did this not break a load before | 14:58 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Sync ZFS pool names https://review.opendev.org/c/openstack/openstack-ansible/+/869575 | 15:07 |
| noonedeadpunk | I think we've "fixed" another path not that long ago | 15:08 |
| noonedeadpunk | jrosser: for https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869546 - I guess we want also same patches for service_setup and mq_setup? | 15:15 |
| noonedeadpunk | or maybe worth using same var for all these cases? | 15:15 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/zed: Prevent bootstrap failure when all roles/collections are overriden https://review.opendev.org/c/openstack/openstack-ansible/+/869458 | 15:17 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/yoga: Prevent bootstrap failure when all roles/collections are overriden https://review.opendev.org/c/openstack/openstack-ansible/+/869459 | 15:17 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/zed: Define name for all collections in a-r-r https://review.opendev.org/c/openstack/openstack-ansible/+/869460 | 15:18 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-plugins master: Unify vars for glusterfs RHEL variants and remove rocky-8 workaround. https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/866116 | 15:26 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-plugins stable/zed: Update TOX_CONSTRAINTS_FILE for stable/zed https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/867498 | 15:26 |
| jrosser | darman: did you find why your deployment was slow in the end? | 16:03 |
| spatel | noonedeadpunk If i deploy ldap backed keystone on multi-region cloud just for identity. in that case how does horizon provide multi-region support? | 16:18 |
| spatel | In that case can i have drop-down menu in horizon to select region and my token will get pass automatically to other region? | 16:18 |
| noonedeadpunk | spatel: you will still need to configure horizon from what I can recall and tell it which regions should it supprot | 16:21 |
| noonedeadpunk | I'm not sure if there is autodetection of regions - maybe there is but I just never used that | 16:22 |
| spatel | Ok, i will put those region entry in horizon but my question is how does they pass token between region ? | 16:22 |
| noonedeadpunk | but yes, you should be able to do that | 16:22 |
| noonedeadpunk | um, your token is in cookies | 16:22 |
| noonedeadpunk | so it's always in browser, and horizon just set cookies for you upon login | 16:23 |
| spatel | you are saying if LDAP is backend and sharing password between region then horizon will do magic with cookies and just redirect to other region without login screen | 16:23 |
| noonedeadpunk | in theory - yes | 16:23 |
| noonedeadpunk | in practise - never used ldap | 16:24 |
| noonedeadpunk | (for keystone) | 16:24 |
| spatel | hmm! I am using LDAP but never thought i can do this setup :) | 16:24 |
| spatel | Now thinking to do that because its easy to handover single horizon instead 5 horizon URL | 16:24 |
| spatel | What if i sync user/password in mysql using ansible script between region ? | 16:25 |
| spatel | in that case it should work correct? | 16:25 |
| noonedeadpunk | Well, you can configure any horizon to serve any openstack deployment, even for some third-party provider | 16:25 |
| noonedeadpunk | and any amount of providers. But yes, question if how good switching between regions will be | 16:26 |
| spatel | noonedeadpunk do you have handy horizon variable to set muti-region using OSA? | 16:55 |
| spatel | I believe just need to set AVAILABLE_REGIONS = [ foo, bar] correct? | 16:56 |
| opendevreview | Merged openstack/openstack-ansible-plugins master: Limit maximum number of threads for parallel git clone https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869564 | 17:20 |
| admin1 | not sure if you guys know, but telling anyway .. when a bucket is created in horizon, what project is actually connecting to ceph and sending the bucket-create command ? | 17:31 |
| admin1 | but asking* | 17:31 |
| jrosser | noonedeadpunk: for your isolated deployment, did you override all of the openstack service repo URLs individually? | 17:34 |
| opendevreview | Merged openstack/openstack-ansible-os_nova master: Enable rbd download when nova_glance_rbd is in use https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/869073 | 18:01 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-plugins stable/zed: Limit maximum number of threads for parallel git clone https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869463 | 18:53 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-plugins stable/yoga: Limit maximum number of threads for parallel git clone https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869464 | 18:53 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-os_ironic stable/zed: Update IPA image for the Zed release https://review.opendev.org/c/openstack/openstack-ansible-os_ironic/+/869465 | 18:55 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-os_nova stable/zed: Enable rbd download when nova_glance_rbd is in use https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/869466 | 18:55 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-os_nova stable/yoga: Enable rbd download when nova_glance_rbd is in use https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/869467 | 18:56 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: Bump ansible version to 2.14.1 https://review.opendev.org/c/openstack/openstack-ansible/+/869599 | 19:13 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Add variable to control no_log in mq_setup role https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869602 | 19:19 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible-plugins master: Add variable to control no_log in service_setup role https://review.opendev.org/c/openstack/openstack-ansible-plugins/+/869604 | 19:26 |
| prometheanfire | damiandabrowski: that worked, thanks | 19:45 |
| opendevreview | Merged openstack/ansible-role-pki stable/wallaby: Ensure CA privatekey permissions https://review.opendev.org/c/openstack/ansible-role-pki/+/867634 | 20:01 |
| mgariepy | jamesdenton, what is the purpose of the br-tunnel here: https://review.opendev.org/c/openstack/openstack-ansible/+/867577/5/doc/source/reference/figures/networking-openvswitch-cn.drawio.png | 20:21 |
| mgariepy | ooops : https://review.opendev.org/c/openstack/openstack-ansible/+/867577/5/doc/source/reference/figures/networking-ovn-cn.drawio.png | 20:22 |
| mgariepy | this one .. ;) | 20:22 |
| mgariepy | when using ovs with iptables firewall driver you need a bridge for it but when using openvswitch fw driver you can just bind it to the interface ip directly. | 20:23 |
| mgariepy | with ovn you don't really have an option for iptables | 20:23 |
| jamesdenton | let me see... | 20:32 |
| jamesdenton | ahh, so it's really just meant to replace the bridge we had called 'br-vxlan' | 20:33 |
| jamesdenton | which, depending on your setup, may not really be a bridge | 20:34 |
| jamesdenton | it just needs to be a layer 3 interface | 20:34 |
| jamesdenton | the interface where the TEP (vtep) is configured | 20:34 |
| mgariepy | what i do it usualy only create a vlan on top of the bond | 20:34 |
| jamesdenton | yep, same thing, really. | 20:34 |
| jamesdenton | i didn't rock the boat here, and went based on the original diagram: https://review.opendev.org/c/openstack/openstack-ansible/+/867577/5/doc/source/reference/figures/networking-neutronagents.png | 20:35 |
| jamesdenton | br-vxlan near the bottom | 20:35 |
| jamesdenton | but we can forgo br-vxlan/br-tunnel altogether and just recommend a tagged interface | 20:35 |
| mgariepy | yep good. | 20:37 |
| opendevreview | Jonathan Rosser proposed openstack/openstack-ansible master: [doc] Update repositiories for mirroring https://review.opendev.org/c/openstack/openstack-ansible/+/868506 | 21:38 |
| prometheanfire | jamesdenton: oh, for ovn, how do we tell it which interface to use for the overlay network? | 22:27 |
| jamesdenton | ovn-encap-ip is a parameter for ovs, that specifies the local tunnel endpoint addr | 22:28 |
| jamesdenton | the playbooks set that automagically | 22:29 |
| jamesdenton | same mechanism that was used for lxb and ovs in the past | 22:29 |
| jamesdenton | https://github.com/openstack/openstack-ansible-os_neutron/blob/c00039c7b0c6006ad5feff1254d99c4245b9cd3e/tasks/providers/setup_ovs_ovn.yml#L38 | 22:29 |
| prometheanfire | ok, I'll look for the magic bits | 22:29 |
| prometheanfire | oh, thanks :D | 22:30 |
| jrosser | the way that neutron_local_ip is determined there is exactly the same as it always was for linuxbridge etc | 22:31 |
| prometheanfire | https://github.com/openstack/openstack-ansible/blob/master/playbooks/common-tasks/dynamic-address-fact.yml | 22:33 |
| opendevreview | Merged openstack/openstack-ansible stable/zed: Unset OSA-defined variables for bootstrap https://review.opendev.org/c/openstack/openstack-ansible/+/868270 | 22:36 |
| jamesdenton | IIRC you will want 'tunnel' defined under cidr_networks in o_u_c, then if you have an interface configured with an IP in that CIDR, its addr is used for neutron_local_ip | 22:41 |
| jamesdenton | otherwise, the default is the mgmt ip of the host. likely undesirable, but works | 22:42 |
| prometheanfire | ah, think we have that ya | 22:42 |
| prometheanfire | reused the same network from vxlan | 22:42 |
| jamesdenton | you can see what the host is using with the following: ovs-vsctl --columns=external-ids list open_vswitch | 22:44 |
| prometheanfire | ok, no issue then, nice to know how it works though, (ref of a ref, kinda) | 22:46 |
| opendevreview | Merged openstack/openstack-ansible stable/yoga: Install ZFS packages for bootstrap-host if needed https://review.opendev.org/c/openstack/openstack-ansible/+/868274 | 22:46 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
