| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible stable/zed: Bump down etcd version for zun https://review.opendev.org/c/openstack/openstack-ansible/+/883715 | 07:48 |
|---|---|---|
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_zun stable/zed: Install kata containers from source https://review.opendev.org/c/openstack/openstack-ansible-os_zun/+/883711 | 07:49 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Configure spice console on haproxy only when it is used https://review.opendev.org/c/openstack/openstack-ansible/+/883688 | 07:53 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Rename container_address to management_address https://review.opendev.org/c/openstack/openstack-ansible/+/883148 | 07:57 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Add management_ip option for metal hosts https://review.opendev.org/c/openstack/openstack-ansible/+/870113 | 09:54 |
| noonedeadpunk | I did bunch of rebases that have invalidated previous votes on them | 09:57 |
| noonedeadpunk | (as they were not "trivial" ones) | 09:57 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Bump upstream SHAs for Antelope https://review.opendev.org/c/openstack/openstack-ansible/+/883550 | 10:14 |
| opendevreview | Merged openstack/openstack-ansible master: Add 'tls' scenario https://review.opendev.org/c/openstack/openstack-ansible/+/881968 | 10:40 |
| noonedeadpunk | I'm trying to figure out 2 things now - why volume tempest test is failing with ceph scenario and why jammy upgrade fails | 11:08 |
| noonedeadpunk | manually vms are created without issues for ceph scenario | 12:08 |
| jrosser | are there two approaches | 12:10 |
| jrosser | like 2 steps of create volume, boot instance using volume | 12:10 |
| jrosser | vs. single step b-f-v of image=foo and the first step is done automatically | 12:10 |
| jrosser | i recall a bunch of work on cinder tempest tests recently to force waiting to be able to ssh in as well? | 12:19 |
| jrosser | perhaps that was volume detach specifically | 12:19 |
| noonedeadpunk | Yes, so what fails is volume detach at the moment. But when I've tried these patches to wait for ssh (newer tempest), there were waaay more failures... | 13:22 |
| noonedeadpunk | but it's not about tempest but smth related to the service config | 13:31 |
| noonedeadpunk | ` Detected user call to delete in-use attachment. Call must come from the nova service and nova must be configured to send the service token. Bug #2004555` but defining service_user doesn't help | 13:39 |
| noonedeadpunk | I wonder if it could be somehow related to a client or smth... | 13:45 |
| noonedeadpunk | aha, ok.... my bad.... | 13:48 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_nova master: Add auth credentials for service_user https://review.opendev.org/c/openstack/openstack-ansible-os_nova/+/883842 | 13:52 |
| jrosser | ahha yes i was just reading the bug | 13:55 |
| jrosser | and got as far as https://review.opendev.org/c/openstack/kolla-ansible/+/883110/3/ansible/roles/nova-cell/templates/nova.conf.j2 | 13:55 |
| jrosser | saw `valid_interfaces = internal` there as well | 13:56 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_cinder master: Define service_user for cinder services https://review.opendev.org/c/openstack/openstack-ansible-os_cinder/+/883844 | 13:57 |
| noonedeadpunk | I don't see valid_interfaces in config ref: https://docs.openstack.org/nova/latest/configuration/config.html#service-user | 13:58 |
| jrosser | https://opendev.org/openstack/keystoneauth/src/branch/master/keystoneauth1/loading/adapter.py#L46 | 14:00 |
| jrosser | TIL | 14:00 |
| noonedeadpunk | `interface and valid_interfaces are mutually exclusive` | 14:01 |
| noonedeadpunk | And we use interface usually | 14:01 |
| jrosser | which seems deprecated | 14:02 |
| noonedeadpunk | well. yes. we can replace with sed and mass-push the update | 14:03 |
| noonedeadpunk | haven't really seen any deprecation notice in logs though... | 14:03 |
| noonedeadpunk | but let's update it for B :) | 14:04 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Bump upstream SHAs for Antelope https://review.opendev.org/c/openstack/openstack-ansible/+/883550 | 14:07 |
| noonedeadpunk | hopefully this pass now ^ | 14:07 |
| jrosser | that is odd it's not in our deprecation logs as it seems like that for many releases now | 14:11 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Switch ubuntu upgrade jobs to Jammy https://review.opendev.org/c/openstack/openstack-ansible/+/879890 | 14:20 |
| damiandabrowski | jrosser: I've looked into your comments about ironic & tls-backend. I think we have 2 topics to discuss. | 14:42 |
| damiandabrowski | 1. [iPXE] fetching image over https | 14:42 |
| damiandabrowski | It's not that simple as iPXE does not allow to disable cert verification. | 14:42 |
| damiandabrowski | The only option is to recompile iPXE with custom CA, so I suggest to mark it as out of scope for now. | 14:43 |
| damiandabrowski | https://lists.ipxe.org/pipermail/ipxe-devel/2014-November/003898.html | 14:43 |
| damiandabrowski | 2. You mentioned that in some cases, IPA can communicate with ironic conductor *backend* directly. | 14:43 |
| damiandabrowski | When exactly can this happen? Could you please share more details so I can update docs accordingly? | 14:43 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Ensure OVN is restarted on package update https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/883740 | 14:44 |
| noonedeadpunk | damiandabrowski: fwiw, I also had comments about docker & tls for zun | 14:45 |
| noonedeadpunk | but it's merged already and not sure how much we should worry about that | 14:45 |
| jrosser | damiandabrowski: the "default" config for ironic in OSA potentially puts br-bmaas as a isolated network | 14:46 |
| jrosser | and in that case the VIP is not accessible from the bmaas network | 14:47 |
| jrosser | so the solution there (as documented in my lxc example in the os_ironic docs) is to configure the ironic agent to make it's callback to the conductor which is responsible for it, directly | 14:48 |
| jrosser | as that has an interface on the bmaas network | 14:48 |
| damiandabrowski | noonedeadpunk: thanks for reminding me, I'll check if we can do something about it before release | 14:53 |
| damiandabrowski | jrosser: ah, i get it now, but IIRC you mentioned conductor before and I see an override only for inspector | 15:01 |
| damiandabrowski | https://opendev.org/openstack/openstack-ansible-os_ironic/src/branch/master/doc/source/configure-lxc-example.rst?display=source#L456 | 15:01 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Drop OVN package installation from ovn_config https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/883851 | 15:01 |
| damiandabrowski | https://opendev.org/openstack/openstack-ansible-os_ironic/src/branch/master/templates/inspector.ipxe.j2#L8 | 15:01 |
| jrosser | damiandabrowski: https://opendev.org/openstack/openstack-ansible-os_ironic/src/branch/master/doc/source/configure-lxc-example.rst?display=source#L443-L446 | 15:01 |
| jrosser | inspector is probably 3) on your list tbh | 15:02 |
| damiandabrowski | hmm, so it indicates that IPA can access service catalog. Do you know how is it possible on isolated bmaas network? | 15:05 |
| jrosser | no this is config for inspector/conductor | 15:06 |
| jrosser | normally they use the catalog to construct the URL passed to IPA to use for the callback | 15:06 |
| jrosser | so in this case `endpoint_override` means, "dont use the catalog, use this URL instead" | 15:07 |
| noonedeadpunk | we somehow broke tempest installation for distro scenario as well | 15:13 |
| noonedeadpunk | it's trying to follow source path | 15:13 |
| noonedeadpunk | https://zuul.opendev.org/t/openstack/build/dbe35ce58d43486db28bd727227516af | 15:13 |
| damiandabrowski | ahhh okok, and how exactly IPA receives ironic API URL from inspector/conductor? | 15:13 |
| damiandabrowski | as a response, after sending a request to ipa-inspection-callback-url ? | 15:13 |
| * NeilHanlon has to mentally context switch to not read IPA as in FreeIPA, the identity management suite | 15:18 | |
| noonedeadpunk | ++ | 15:18 |
| jrosser | damiandabrowski: it needs something to start with, so generally kernel command line parameter | 15:18 |
| jrosser | it's a disk image which is pxebooted, and it looks like the options are kernel command line parameter or lookup with mdns | 15:19 |
| damiandabrowski | yeah, but I can't see any parameter for setting ironic API URL here: https://opendev.org/openstack/openstack-ansible-os_ironic/src/branch/master/templates/inspector.ipxe.j2 | 15:20 |
| jrosser | but thats inspector | 15:28 |
| jrosser | i'm confused | 15:28 |
| jrosser | they are two different things | 15:28 |
| opendevreview | Merged openstack/openstack-ansible master: Configure spice console on haproxy only when it is used https://review.opendev.org/c/openstack/openstack-ansible/+/883688 | 15:28 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_tempest master: Make tempest respect service_install_method https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/883854 | 15:29 |
| jrosser | damiandabrowski: i'm not sure what we are talking about now tbh, ironic, or ironic inspector | 15:29 |
| damiandabrowski | i'm trying to figure out how this endpoint is passed to IPA :D | 15:29 |
| damiandabrowski | https://opendev.org/openstack/openstack-ansible-os_ironic/src/branch/master/doc/source/configure-lxc-example.rst?display=source#L443-L446 | 15:29 |
| jrosser | my example shows overrides for both the ironic insepctor callback and the ironic-python-agent callback | 15:29 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_tempest master: Make tempest respect service_install_method https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/883854 | 15:30 |
| damiandabrowski | ah wait, i think i understand it now | 15:30 |
| jrosser | damiandabrowski: well ironic writes a lot of files out dynamically as well | 15:30 |
| jrosser | for the pxebooting | 15:31 |
| damiandabrowski | ironic inspector container also has only bmaas network, right? | 15:31 |
| jrosser | theres a diagram :) | 15:31 |
| jrosser | i mean this also all "depends" of course | 15:31 |
| damiandabrowski | so ironic inspector <> ironic API communication also needs to be done without haproxy | 15:31 |
| jrosser | you can do what you like in any particular deployment | 15:31 |
| damiandabrowski | yeah, I'm talking only about your example | 15:31 |
| jrosser | make it be on metal, routed networks, add a firewall etc | 15:32 |
| jrosser | what ironic inspector <> ironic API? | 15:32 |
| damiandabrowski | or no, i think i still do not understand it.. | 15:32 |
| damiandabrowski | "# Direct IPA to callback directly to deploying ironic container (via BMAAS network)" | 15:32 |
| damiandabrowski | this comment in particular | 15:32 |
| spatel | jrosser I am getting this error during my upgrade of wallaby to Xena - https://paste.opendev.org/show/bKeyJvdFoldpPtI6iErQ/ | 15:33 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_tempest master: Clean up old rhel variable files https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/883856 | 15:33 |
| jrosser | damiandabrowski: what in particular? | 15:33 |
| spatel | should i do -e galera_ignore_cluster_state=true ? | 15:33 |
| noonedeadpunk | spatel: does all your galera backends are up? | 15:33 |
| noonedeadpunk | as this might indicate that one of them are unhealthy | 15:33 |
| damiandabrowski | how exactly service_catalog.endpoint_override affects IPA | 15:33 |
| spatel | Hmm I think its dead.. damn it let me check | 15:34 |
| damiandabrowski | i.e. what and how provides the value of service_catalog.endpoint_override to the IPA | 15:35 |
| spatel | noonedeadpunk only infra-1 mysql is down | 15:36 |
| jrosser | well ironic writes out a whole load of files into /httpboot, when a node is booted | 15:36 |
| jrosser | specific to that node | 15:37 |
| spatel | on infra-2 and 3 working fine.. | 15:37 |
| spatel | let me try to restart and see | 15:37 |
| damiandabrowski | jrosser: yes, i even deployed ironic AIO trying to understand that, besides initramfs and kernel there is only inspector.ipxe | 15:39 |
| damiandabrowski | but the value of ironic API endpoint is not provided in inspector.ipxe: https://opendev.org/openstack/openstack-ansible-os_ironic/src/branch/master/templates/inspector.ipxe.j2 | 15:39 |
| damiandabrowski | ahh sorry | 15:39 |
| jrosser | no becasue ironic != inspector | 15:39 |
| damiandabrowski | "when the node is booted" | 15:39 |
| jrosser | sorry i keep saying this | 15:39 |
| jrosser | inspector config != ironic config | 15:40 |
| damiandabrowski | yes, but IPA is a part of inspector, right? | 15:40 |
| jrosser | IPA is used by both yes i believe | 15:41 |
| damiandabrowski | ahhh ok | 15:42 |
| jrosser | ironic-python-agent is used as part of the inspection process (i am slightly guessing now) that the relevant inspector parts are dropped into the regular i-p-a thing with dib | 15:43 |
| jrosser | but i-p-a booted in ramdisk and used to write the actual OS to the physical disk during deployment with ironic, and also to wipe the disks during cleaning etc | 15:43 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_tempest master: Build wheels only for source isntalls https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/883857 | 15:44 |
| jrosser | afaik there is work going on currently in the ironic project to merge inspector and ironic into one singular thing | 15:44 |
| jrosser | damiandabrowski: i think that the reason you see a static config for the callback URL in the inspector case in the config file OSA writes is that nothing at all is known about the node at that point, the idea of inspector is to do discovery from zero, so you have to start somewhere | 15:46 |
| damiandabrowski | ahh thanks, it's complicated :| i'll go trough ironic docs once again trying to understand how exactly endpoint_override affects IPA | 15:46 |
| jrosser | but for an already enrolled node that is being deployed later using ironic then there are * options for how it happens, so the actual config to pxeboot (and somewhere in there goes the callback url) are written out dynamically | 15:47 |
| jrosser | unfortunatly it also cleans these files up afterwards so i cant jsut grab an old one and take a look | 15:47 |
| jrosser | damiandabrowski: we need to improve the CI here and add "virtualbmc" to our ironic tests to allow a virtual baremetal node to be booted | 15:49 |
| spatel | noonedeadpunk This is very odd, I have upgraded 3 production cluster without error but today on 4th one hitting lots of error and strange issue | 15:49 |
| spatel | Playbook stuck here - TASK [galera_server : Install galera role remote packages (apt)] | 15:49 |
| opendevreview | Merged openstack/openstack-ansible stable/zed: Bump down etcd version for zun https://review.opendev.org/c/openstack/openstack-ansible/+/883715 | 15:50 |
| opendevreview | Merged openstack/openstack-ansible master: Rename container_address to management_address https://review.opendev.org/c/openstack/openstack-ansible/+/883148 | 15:50 |
| jrosser | spatel: playbooks can get stuck like that if you have CTRL-C them at some point before and leave a python process behind holding the lock on the apt database | 15:50 |
| damiandabrowski | jrosser: ack, thanks for your patience! | 15:51 |
| spatel | jrosser I didn't do any CTRL+C yet.. just trying to look into logs and see if i can find something | 15:52 |
| spatel | Just test "apt update" in galera container and it works fine.. | 15:52 |
| jrosser | ok | 15:53 |
| spatel | where are ansible logs? | 15:53 |
| spatel | oh its on deployment node | 15:54 |
| spatel | no activity happening in /var/log/apt/term.log on galera container | 15:55 |
| spatel | container saying mysql is missing.. haha | 15:56 |
| spatel | let me reboot container and see | 15:56 |
| spatel | very odd that it has all mariadb packages install but still not able to find mysql command | 15:59 |
| spatel | jrosser look like galera-2 node in wonky stat, it has package half done and now i am not able to remove them even i can see them | 16:05 |
| spatel | is it safe to delete container ? | 16:05 |
| spatel | and re-run - openstack-ansible setup-infrastructure.yml -e 'galera_upgrade=true' -e 'rabbitmq_upgrade=true' -e package_state=latest | 16:05 |
| spatel | https://paste.opendev.org/show/bkzqzhzMDNKsaWE4DyyI/ | 16:06 |
| damiandabrowski | jrosser: found it! that's how ironic fetches API URL for IPA: | 16:07 |
| damiandabrowski | https://github.com/openstack/ironic/blob/69c8977e5094403ff8a06c1fdbac381ebfebc56a/ironic/drivers/modules/deploy_utils.py#L668 | 16:07 |
| damiandabrowski | then its passed as a ipa-api-url kernel parameter(so you were right about temporary files in /httpboot) | 16:08 |
| damiandabrowski | PS. it's only used when deploying node | 16:08 |
| jrosser | and here is the other end of that in the agent https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/config.py#L30-L37 | 16:08 |
| damiandabrowski | victory \o/ | 16:09 |
| NeilHanlon | #teamwork | 16:14 |
| spatel | Quick question how do I re-add fresh node in galara cluster? I have removed all package on node-2 but playbook doesn't like that | 16:18 |
| spatel | should i delete container and re-run play ? | 16:18 |
| jrosser | there are instructions for this in the guides | 16:25 |
| jrosser | https://docs.openstack.org/openstack-ansible/latest/admin/maintenance-tasks.html | 16:26 |
| spatel | jrosser I am reading that but in my case its whole container is dead and if i am re-running playbook then getting error that bring up mysql but i don't have mysql on that container | 16:30 |
| spatel | Tying to understand should I be going - openstack-ansible galera-install.yml --tags galera-bootstrap | 16:31 |
| spatel | or just openstack-ansible galera-install.yml will re-build my node-2 and add into existing node. | 16:32 |
| spatel | This is prod so taking baby step :( | 16:32 |
| damiandabrowski | spatel: if it's not primary galera node(`galera_cluster_members[0]`), I'd remove /var/lib/mysql on a broken container and just run `openstack-ansible galera-install.yml -l <broken_galera_container` | 16:51 |
| spatel | damiandabrowski how do i check that? This is node-2 (infra-2) | 16:52 |
| damiandabrowski | cd /opt/openstack-ansible; ansible -m ping galera_all[0] | 16:54 |
| spatel | ostack-sin-infra-1-1_galera_container-dbb207e8 | 16:56 |
| spatel | In my case node-1/3 is working one. | 16:56 |
| spatel | node-2 is dead | 16:56 |
| damiandabrowski | ok, feel free to run openstack-ansible galera-install.yml -l node-2 then | 16:56 |
| damiandabrowski | but to ensure that galera won't have any troubles with state transfer, I'd remove /var/lib/mysql content beforehand | 16:57 |
| spatel | Let me understand that means it will use node-1 to do SST to node-2 correct? | 16:57 |
| damiandabrowski | (only on node-2 ofc) | 16:57 |
| spatel | ok | 16:57 |
| damiandabrowski | hmm, IIRC donor is chosen randomly | 16:59 |
| damiandabrowski | but I'm not 100% sure | 16:59 |
| spatel | damiandabrowski I have one more question after upgrade why we keep older packages hanging there? - https://paste.opendev.org/show/bOw5MabaWbAEcJl0hnQo/ | 17:00 |
| spatel | 10.5 is wallaby and 10.6 is Xena | 17:01 |
| noonedeadpunk | Looks like 883857 is fixing distro jobs | 17:02 |
| damiandabrowski | hmm i'm not sure tbh, probably because it doesn't hurt to still have them | 17:05 |
| damiandabrowski | but since galera does not support downgrades, they are not really useful | 17:05 |
| noonedeadpunk | spatel: we actually should not... | 17:05 |
| noonedeadpunk | I'd say that's likely a bug as packages we remove should be here: https://opendev.org/openstack/openstack-ansible-galera_server/src/branch/master/vars/debian.yml#L60-L63 | 17:06 |
| noonedeadpunk | but wait.... | 17:06 |
| noonedeadpunk | what older packages? | 17:06 |
| noonedeadpunk | rc means these are removed from the system | 17:07 |
| damiandabrowski | ah, i was looking for 'apt' instead of 'package' in our code :D so noonedeadpunk is right | 17:07 |
| noonedeadpunk | sweet, 883550 almost passing as well | 17:09 |
| spatel | damiandabrowski I am getting this error when i run playbook -l node-2 | 17:10 |
| spatel | https://paste.opendev.org/show/byr2B536kxRVuZxzc8LD/ | 17:10 |
| spatel | This is rabbit hole :( | 17:13 |
| spatel | let me destroy container that is all left at this point | 17:14 |
| spatel | I believe i should do openstack-ansible galera-install.yml (instead -l node-2) | 17:15 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Ensure OVN is restarted on package update https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/883740 | 17:17 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_neutron master: Drop OVN package installation from ovn_config https://review.opendev.org/c/openstack/openstack-ansible-os_neutron/+/883851 | 17:18 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible master: Switch ubuntu upgrade jobs to Jammy https://review.opendev.org/c/openstack/openstack-ansible/+/879890 | 17:18 |
| noonedeadpunk | no, you totally should not use limit here | 17:20 |
| spatel | I have destroy whole container and now re-running setup-host + galera-install.yml | 17:22 |
| NeilHanlon | btw i'm around if anything needs a +2 (especially if I already reviewed it ;) ). will check on review dashboard periodically | 17:22 |
| NeilHanlon | happy monday :) | 17:22 |
| damiandabrowski | noonedeadpunk: weird, i was pretty sure that galera role supports limits O.o distro upgrade guide also suggests to run it with limits: | 17:25 |
| damiandabrowski | https://docs.openstack.org/openstack-ansible/latest/admin/upgrades/distribution-upgrades.html#:~:text=limit%20localhost%2Creinstalled_host*-,openstack%2Dansible%20setup%2Dinfrastructure.yml%20%2D%2Dlimit%20localhost%2Crepo_all%2Crabbitmq_all%2Creinstalled_host*,-openstack%2Dansible%20setup | 17:25 |
| noonedeadpunk | oh, huh, I might be wrong then... | 17:26 |
| opendevreview | Dmitriy Rabotyagov proposed openstack/openstack-ansible-os_octavia master: Implement support for octavia-ovn-provider driver https://review.opendev.org/c/openstack/openstack-ansible-os_octavia/+/868462 | 17:32 |
| spatel | damiandabrowski noonedeadpunk that works... all looking good now | 18:09 |
| spatel | we should add this kind of scenario in doc but again.. we can't add all use case...:( | 18:10 |
| spatel | Thank you!!! | 18:10 |
| opendevreview | Merged openstack/openstack-ansible-ceph_client master: Add immutable object cache documentation https://review.opendev.org/c/openstack/openstack-ansible-ceph_client/+/882830 | 18:24 |
| opendevreview | Merged openstack/openstack-ansible-ceph_client master: Add config and documentation for ceph perisistent write log cache https://review.opendev.org/c/openstack/openstack-ansible-ceph_client/+/882849 | 19:36 |
| opendevreview | Merged openstack/openstack-ansible-ceph_client master: Add releasenote for ceph immutable object cache and persistent write log cache https://review.opendev.org/c/openstack/openstack-ansible-ceph_client/+/882850 | 19:36 |
| noonedeadpunk | so, looks like all obvious issues are covered new | 20:57 |
| noonedeadpunk | couple of rechecks needed due to intermittent issues | 21:02 |
| noonedeadpunk | I've updated https://etherpad.opendev.org/p/osa-antelope-leftover-patches with patches just in case | 21:02 |
| jrosser | feels like we continually (like every few years) mess with the distro install for tempest | 21:02 |
| noonedeadpunk | yeah, when I was looking at the role I was wondering how it worked to be frank | 21:04 |
| noonedeadpunk | as it was more like lucky co-incidence | 21:04 |
| jrosser | idk if it is fixed now but there never used to be tempest plugins packaged for ubuntu | 21:04 |
| noonedeadpunk | no, it's not | 21:04 |
| noonedeadpunk | for ubuntu it's still source install | 21:04 |
| noonedeadpunk | though, we shouldn't try to build wheels in this case | 21:04 |
| noonedeadpunk | I'm not sure though if we should try to build wheels for tempest at all though | 21:05 |
| noonedeadpunk | as playbook run it against utility_all[0] always | 21:05 |
| noonedeadpunk | so wheels in this case kinda waste of time | 21:05 |
| jrosser | so here https://review.opendev.org/c/openstack/openstack-ansible-os_tempest/+/883857/1 | 21:05 |
| jrosser | isnt `venv_build_host != inventory_hostname` always false when there is no repo server | 21:06 |
| noonedeadpunk | I was thinking to hardcode "no" there... | 21:06 |
| jrosser | right - so this is to make the code more obvious then, rather than change behaviour? | 21:06 |
| noonedeadpunk | venv_build_host was failing to calculate | 21:06 |
| noonedeadpunk | due to our changes | 21:06 |
| jrosser | ahha i see | 21:06 |
| noonedeadpunk | with 'dict object' has no attribute 'ubuntu_22.04_x86_64' | 21:07 |
| jrosser | hrrm we probably miss a default somewhere | 21:07 |
| jrosser | but anyway | 21:07 |
| noonedeadpunk | But what default that would be... | 21:07 |
| noonedeadpunk | as venv_build_targets should be empty dict I guess | 21:09 |
| noonedeadpunk | but yeah, anyway | 21:10 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!